9.1: Analytical Findings (Subgroup A) - 7-teens/7-teens-DSA3101-2410-Project GitHub Wiki

A. Analysis of Key Factors Influencing Customer Purchasing Behavior

This part analyzes customer purchasing habits on Shopee to pinpoint the main factors that influence buying decisions. By examining trends in customer behavior, we aim to develop targeted strategies for marketing and engagement that align more closely with customer needs, ultimately boosting loyalty and driving revenue growth.

The technical implementation details, including code for each step of the analysis, can be found in the Key Factor Influencing Purchasing Behavior Notebook in the GitHub repository.

Part 1: Exploratory Data Analysis

Overview On 2019 Sales

Temporal Analysis of Shopee Sales 2019

1. Observed Component

The "Observed" graph represents Shopee’s daily sales throughout the year 2019.

Double-Date Sales Spikes: Major spikes appear to align with double-date sales events (e.g., 2.2, 11.11, 12.12), which might indicate their effectiveness in increasing sales.

2. Trend Component

The "Trend" component shows the general direction of daily sales over the year, without seasonal and irregular variations.

Mid-2019 Growth Period: There’s an upward trend in sales around the middle of 2019, which might reflect effective promotional strategies, a growing user base, or increased platform engagement.
End-of-Year Slowdown: Sales show a slight decline towards the end of the year, following peaks in double-date events (e.g., 11.11 and 12.12).

3. Seasonal Component

The "Seasonal" graph reveals recurring sales patterns, suggesting possible weekly and monthly cycles in customer behavior.

Weekend Peaks: A recurring cycle shows weekends might consistently generate more sales, suggesting customers could be more active on the platform during these days.
Potential for Monthly Cycles: Although double-date events appear to dominate monthly patterns, there might be potential to create anticipation for smaller, recurring monthly sales beyond the double dates (e.g., "End of Month Specials").

4. Residual Component

The "Residual" graph captures irregularities or unexpected fluctuations in sales beyond the trend and seasonal patterns.

Unexplained Highs and Lows: Certain dates show sales figures that deviate from both the trend and seasonality, potentially due to specific product launches, viral promotions, or external events (e.g., weather, holidays) that might have impacted customer behavior unexpectedly.

Analysis of Order Volume Based on Hours of Day

Peak Order Times (7-9 AM):
- There is a significant peak in orders during the early morning hours, especially from 7 AM to 9 AM. The highest spike is at 7 AM, where the number of orders reaches above 10,000.
- This indicates that early morning is a popular time for placing orders, which could be due to people starting their day, ordering breakfast items, or setting up for business hours.
Midday Consistency (11 AM - 3 PM):
- From around 11 AM to 3 PM, there is a steady volume of orders with a peak at 3 PM, though less intense than the morning peak. This likely corresponds to lunch orders or afternoon transactions.
- The order count in this period is relatively even, suggesting a stable demand.
Lower Demand Period (4 PM - 5 PM):
- There is a noticeable dip in order volumes between 4 PM and 6 PM, with the lowest point around 5 PM. This may indicate a period when fewer people are ordering, possibly due to it being the end of the workday or the time between lunch and dinner.
Evening Rise (6 PM - 11 PM):
- In the evening, orders rise again and remain relatively consistent from 7 PM through to around 11 PM. This pattern could be associated with dinner orders or people engaging in evening activities, leading to a renewed demand for goods or services.
Early Morning and Late Night Lows:
- There is minimal order activity from midnight until around 6 AM, which is expected since this is typically a low-demand period as most people are asleep.

Average Order per Day for Regular and Sales Date

Priority and Additional Mega Sales: Confirming the peaks from previous analysis, Shopee mega sales dates (e.g., 11.11, 12.12) show a higher (almost triple) average number of orders per day, highlighting their potential effectiveness in driving customer engagement. Customers might be more motivated to make purchases due to exclusive discounts, extensive marketing, and the anticipation built around these major events.

Shopee Gross Merchandise Value (GMV) Analysis 2019

Initial Analysis Based on Daily and Weekly GMV Trend

Special Dates Consistently Above Average GMV:
- The graph shows that GMV on special dates (highlighted sales events like 11.11, 12.12, etc.) consistently stays above the average GMV line. This might suggests that these special events are consistent in boosting GMV, making them high-performing days in terms of sales volume.
Regular Days with Mixed GMV Patterns:
- While there are peaks on some regular days that reach or even exceed the average GMV line, many other regular days fall below average. This creates a mixed pattern where certain regular days perform well, but a substantial number of low-GMV days contribute to a lower yearly average.

Average of Daily GMV for Regular vs. Special Dates

Higher Average GMV on Mega Sales Days:
- Both Priority and Additional Mega Sales show higher average GMV per day than regular dates, aligning with the observation that special dates tend to drive higher engagement.
Regular Dates:
- Although some regular days have high GMV, the substantial number of regular days with below-average GMV pulls down the overall average. This reinforces that while regular days can have their own peaks, many still underperform compared to special sale events.

Possible Key Factors Influencing Customer Purchasing Behavior Based on Temporal Analysis

Special Sales Events: Priority and Additional Mega Sales (e.g., 11.11, 12.12) consistently generate higher-than-average orders and GMV, suggesting that limited-time deals and extensive marketing during these events might encourage more purchases.
Weekly Cycles: Weekly trends indicate higher engagement on certain days, like weekends, implying that natural shopping habits may influence purchase timing, independent of large sales events.
Peak Hours: Based on the hourly analysis, there is a notable spike in order volume between 7 AM and 9 AM, with another increase in engagement during the evening hours from 6 PM to 11 PM. This pattern suggests that customers might be more active in the early morning and later in the evening, potentially due to morning routines and evening relaxation time.

Key Insights on Order Volume and GMV Based on Temporal Analysis

Mega Sales Events as Key Revenue Drivers Sales events like 11.11 and 12.12 contribute significantly to Shopee’s GMV, underscoring their importance as high-performing revenue days.
- Actionable Insight for Marketing Team: Focus marketing resources on these events to maximize reach. Tailor campaigns to highlight exclusive discounts and deals that drive urgency.
- Actionable Insight for Business Analyst Team: Monitor which customer segments respond most to mega sales events to refine audience targeting and future campaign designs.
Recurring Double-Date Events for Consistent Engagement Monthly double-date events like 2.2, 9.9, and 10.10 create predictable spikes in order volume. These events are effective in maintaining engagement between larger sales.
- Actionable Insight for Marketing Team: Position double-date sales as part of a recurring campaign series to establish them as anticipated shopping days. Increase pre-event awareness to keep engagement high.
- Actionable Insight for Business Analyst Team: Identify categories or products that perform exceptionally well during these events and use that data to drive stocking and promotional decisions.
Weekend Peaks Suggest Strategic Opportunity for Regular Promotions Weekend sales see consistent order volume increases, suggesting higher customer activity.
- Actionable Insight for Marketing Team: Consider launching regular “Weekend Deals” or “Flash Sales” to build on this natural increase in activity.
- Actionable Insight for Business Analyst Team: Segment customers who are highly active on weekends and target them with personalized offers, potentially increasing order frequency.
Possible Deals for Peak Hours: The identified peak hours (7 AM - 9 AM and 6 PM - 11 PM) represent optimal times for flash sales and limited-timeoffers, as customer engagement is naturally higher during these periods.
- Actionable Insight for Marketing Team: Launch flash deals and time-limited offers during these hours to leverage high traffic and drive urgency. Experiment with exclusive discounts or limited-quantity promotions to encourage more purchases during these peak periods.
- Actionable Insight for Business Analyst Team: Track order volume and GMV metrics during these peak hours to gauge the effectiveness of these flash deals and further fine-tune promotions based on response rates.
Inconsistent GMV on Regular Days Regular weekdays show variable GMV performance, with many days falling below the average.
- Actionable Insight for Marketing Team: Implement smaller, targeted promotions on regular days to boost sales consistency. Highlight product categories with steady demand to attract repeat purchases.
- Actionable Insight for Business Analyst Team: Track daily GMV trends and experiment with regular-day promotions to understand which tactics work best to elevate baseline sales.
Post-Mega Sale Engagement Strategies
- Insight: There is a slight decline in sales after large events, likely due to customer spending fatigue.
- Actionable Insight for Marketing Team: Shift post-event marketing focus towards value-based offers and loyalty rewards to encourage light engagement without overwhelming customers.
- Actionable Insight for Business Analyst Team: Assess long-term customer retention metrics post-event to fine-tune event spacing and frequency for sustained engagement without fatigue.

Overview on Shopee Products

Analysis on Order Volume Based on Product Category

1. High-Demand Categories

Men's Clothes, Health & Beauty, and Women's Clothes: These categories have the highest order counts, each exceeding 10,000 orders. This indicates that fashion and personal care are highly popular among customers. These categories likely cater to a broad demographic, covering a wide range of personal style and self-care needs.
Mobile & Accessories, Baby & Toys, and Home & Living: These categories also show strong demand, with order counts ranging between 6,000 and 8,000. The high order volume in Mobile & Accessories reflects the popularity of smartphones and related accessories, which are frequently purchased and replenished. The Baby & Toys and Home & Living categories suggest that customers view the platform as a source for family and lifestyle-supporting products.

2. Mid-Demand Categories

Groceries & Pets, Home Appliances, Sports & Outdoor, Automotive: These categories have moderate order volumes, with counts ranging between 4,000 and 5,000. This suggests steady customer interest in essential and recreational products, possibly indicating that these categories are used to fulfill routine or specific needs.
Women's Bags, Watches, Men's Bags & Wallets: These accessory categories show moderate demand, with order counts close to 3,000. These products might be viewed as occasional purchases, contributing to a steady but not overwhelming order volume.

3. Lower-Demand Categories

Cameras & Drones, Women's Shoes, Fashion Accessories: These categories have relatively lower order counts, below 2,500. This could indicate more selective or less frequent purchasing in these areas, possibly due to higher price points or niche customer interests.
Computer & Accessories, Games, Books & Hobbies, Men's Shoes: These categories also see lower order volumes, suggesting that they might cater to specific segments of the customer base with occasional purchasing needs.

4. Niche and Specialized Categories

Travel & Luggage, Gaming & Consoles, Tickets & Vouchers, Others: These categories have the lowest order counts, falling below 1,000. This suggests that they cater to more specialized or sporadic needs, and are not a primary focus for most customers. Travel & Luggage and Tickets & Vouchers may see occasional spikes due to seasonal events or promotions but otherwise remain low in demand.

Analysis on GMV Based on Product Category

1. High-GMV Categories

Mobile & Accessories, Cameras & Drones, and Watches: These categories have the highest GMV, each nearing or exceeding 600,000. The high GMV suggests that these categories contain high-value items, possibly with larger price tags, which contribute significantly to overall revenue. Mobile & Accessories leads, reflecting the high demand and value associated with smartphones and related gadgets. Cameras & Drones and Watches also indicate significant revenue generation, likely due to the high cost of items in these categories.
Home Appliances: With a GMV slightly below the top three categories, Home Appliances also generates substantial revenue, around 500,000. This category likely includes higher-priced items like kitchen appliances and household electronics, contributing to its high GMV despite potentially lower order volume compared to categories like fashion.

2. Mid-GMV Categories

Men's Clothes, Computer & Accessories, Sports & Outdoor: These categories have moderate GMV, with values ranging from 200,000 to 300,000. Men's Clothes may have a high order volume but lower price points, balancing out to a mid-range GMV. Computer & Accessories and Sports & Outdoor show solid performance in terms of revenue, indicating steady demand and moderately priced items.
Baby & Toys, Health & Beauty: Both categories fall within the mid-GMV range, suggesting that while these items might be ordered frequently, their lower individual prices contribute to a moderate overall GMV.

3. Lower-GMV Categories

Men's Bags & Wallets, Women's Clothes, Travel & Luggage: These categories show lower GMV, falling between 100,000 and 150,000. This could be due to a combination of moderate order volumes and relatively lower prices per item compared to high-GMV categories.
Women's Bags, Groceries & Pets, Home & Living: These categories also show relatively low GMV, indicating that while they might see frequent orders, the price points are generally lower, resulting in a smaller contribution to overall revenue.

4. Niche and Specialized Categories

Tickets & Vouchers, Automotive, Women's Shoes: These categories have some of the lowest GMV values, with less than 100,000 each. Tickets & Vouchers might be event-driven and could have limited consistent demand, while Automotive and Women's Shoes may reflect specialized purchases with occasional demand.
Games, Books & Hobbies, Fashion Accessories, Gaming & Consoles, Others: These categories generate minimal GMV, with values well below 100,000. This suggests they either cater to highly specialized interests or have low purchase frequency and lower price points.

Analysis on Order Volume by Price Range

1. Majority of Orders in Low Price Ranges:

The largest volume of orders falls within the (0, 1] price range, with over 25,000 orders. This indicates that most orders are for lower-priced items, which may suggest that customers frequently purchase inexpensive products, possibly as impulse buys or small, routine purchases.
The next price range, (1, 2], also has a significant number of orders, contributing to the high volume of low-cost transactions.

2. Decreasing Order Volume with Higher Price Ranges:

As the price range increases, the total number of orders decreases substantially. Price ranges from (2, 5] to (20, 50] still have a fair amount of activity, indicating some customer willingness to spend on moderately priced items.
Higher price ranges, especially those above (50, 100], see a sharp decline in the number of orders. Orders in ranges (100, 500], (500, 1000], and (1000, 10000] are minimal, showing limited demand for high-priced items.

3. High Demand for Affordable Products:

The cumulative percentage reaches approximately 80% by the (10, 20] price range, indicating that most of the orders are concentrated within the lower price ranges.
The data suggests that customers primarily shop for affordable products, with a majority of the orders concentrated in the lower price brackets. This could reflect a price-sensitive customer base that prefers value-driven purchases over high-cost items.

Looking at the total orders over all time, a similar trend still applies, with the majority of orders concentrated in the lower price ranges.

Analysis on Order Volume by Price Range All Time (2015-2019)

1. Majority of Orders in Low Price Ranges:

The (0, 1] price range has the highest volume of orders, surpassing 8 million orders, indicating that low-priced items continue to dominate total orders over time. This pattern suggests that customers are consistently purchasing low-cost items, possibly due to the platform's appeal for small, affordable purchases.
The (1, 2] price range follows with a substantial number of orders, reinforcing the idea that price-sensitive customers are a significant part of the customer base.

2. Decreasing Order Volume with Increasing Price Ranges:

As seen previously, there is a steady decline in order volume as the price range increases. Price ranges (2, 5] and (5, 10] maintain moderate activity, indicating that customers are willing to spend on moderately priced items, but demand drops considerably for higher price ranges.
Orders in the (20, 50] and above price brackets show a noticeable decrease, with orders in the (1000, 10000] range being almost negligible. This indicates that high-priced items contribute minimally to total order counts.

3. Sustained Demand for Affordable Products:

The cumulative percentage line shows that approximately 80% of total orders are covered by price ranges up to around (10, 20]. This confirms that a large proportion of orders come from low-cost items, with higher price ranges contributing less to the overall order count.
The data illustrates that customer preference for affordable products remains consistent over time. Lower price ranges consistently capture the highest order volumes, suggesting a predominantly price-sensitive customer base that continues to prioritize low-cost purchases.

Key Takeaways

Focus on Low-Cost Products: The platform may benefit from focusing on expanding low-cost offerings and providing promotions or discounts within the (0, 1] to (5, 10] price ranges to capitalize on demand.
Targeted Strategies for Mid-Range Price Brackets: Price ranges (5, 10] to (20, 50] still show notable activity, indicating an opportunity to target customers in these segments with upselling or cross-selling strategies to boost sales in the mid-price range.
Price Sensitivity: The steep decline in order volume as prices increase suggests that the customer base is generally price-sensitive. Price-sensitive strategies, such as loyalty rewards for frequent purchases in lower price ranges or installment options for higher-priced items, could encourage purchases across a broader range.
Consistent 80/20 Rule: With around 80% of orders coming from lower price ranges, optimizing inventory and promotions within these segments will likely yield high returns. The remaining higher price ranges could be targeted with specific marketing campaigns for niche audiences.

This analysis underscores the importance of affordable products in driving consistent order volume over time and highlights strategies to maintain engagement with a predominantly price-sensitive customer base.

Analysis of Discount Usage in Orders

1. High Usage of Discounts (60%):

A significant majority, 60% of orders, were placed using a discount. This highlights the importance of discounts in customer purchasing behavior, suggesting that many customers are inclined to make a purchase when there is a financial incentive or price reduction involved.
This may reflect a price-sensitive customer base that actively seeks discounts or promotions when shopping.

2. Orders Without Discount (40%):

The remaining 40% of orders were placed without any discount. While this is lower than the orders with discounts, it still represents a sizable portion of customers willing to purchase at full price.
This could indicate a segment of customers who prioritize convenience, product availability, or quality over cost savings.

Analysis of Impact of Rating on Orders from Sample Orders in 2019

Preference for Highly Rated Products:
- The significant number of orders in the 4 and 5-star bins implies that customers actively seek out products with positive feedback, possibly viewing high ratings as a measure of product reliability, quality, or customer satisfaction.
- The low to negligible presence of orders in the 1-3 star bins reinforces this idea, as customers appear to avoid lower-rated products.
Trust in High Ratings:
- Since the majority of orders are associated with products rated 4 stars or higher, it’s clear that customers trust ratings as an indicator of product quality. This behavior suggests that high ratings play an essential role in the purchase decision-making process, likely encouraging customers to feel more confident about their choice.
Impact on Purchase Decisions:
- The cumulative effect shows that 4- and 5-star rated products account for nearly 100% of purchases, supporting the notion that ratings are a critical factor influencing customer choice.
- Customers may view high ratings as a shortcut for assessing quality, reducing the risk of dissatisfaction.
Looking at the total orders by coarse rating bin over all time, a similar trend confirms that ratings play an important role in customer purchasing decisions.

Analysis of Impact of Rating on Orders All Time (2015-2019)

1. Strong Preference for Highly Rated Products:

Just as with the sample orders, the majority of all-time orders are associated with products in the 4- and 5-star rating bins. The 4-star rating bin alone accounts for a substantial majority of orders, with more than 2.5 million total orders, while 5-star products follow with over half a million orders.
This high concentration of orders within these top two rating bins underscores a customer preference for well-rated products, indicating that customers likely view ratings as a measure of quality and reliability.

2. Minimal Orders for Low-Rated Products:

There is virtually no order activity in the 1-3 star bins, showing that poorly rated products see minimal demand. This reinforces the notion that customers avoid lower-rated products when making purchasing decisions, focusing instead on items with strong positive feedback.

3. Ratings as a Consistent Driver of Sales:

The similarity in trends over both sample and all-time data highlights that ratings have a long-term impact on customer behavior. Customers appear to rely heavily on ratings as a filter when choosing products, reflecting the importance of positive ratings in maintaining product visibility and attractiveness.

Analysis of Orders by Fine Rating (Sample Orders and All Time)

1. High Preference for Top Ratings (5.0 and 4.9):

The majority of orders come from products with 5.0 and 4.9 ratings, with 5.0-rated products totaling around 25,000 orders and 4.9-rated products reaching nearly 40,000 orders.
Together, these two ratings account for a significant portion of total orders, as shown by the cumulative percentage line, which exceeds the 80% threshold within these top two ratings. This suggests that customers overwhelmingly prefer products with near-perfect ratings, indicating that high ratings play a crucial role in purchase decisions.

2. Gradual Decline in Orders for Slightly Lower Ratings (4.8 to 4.5):

Orders start to decline noticeably with ratings of 4.8 and lower. Products with ratings between 4.7 and 4.5 see progressively fewer orders, suggesting that even slight decreases in rating can impact customer interest.
This trend highlights that customers may perceive even small rating differences as significant when choosing products, favoring items that consistently receive higher reviews.

3. Minimal Orders for Ratings Below 4.5:

Products with ratings of 4.4 or lower have very few orders. This further emphasizes that customers are likely to avoid products with lower ratings, perhaps viewing them as less trustworthy or of lower quality.
The near-zero order count for ratings closer to 4.0 suggests that customers are highly selective and rely heavily on ratings as an indicator of product reliability.

4. Strong Positive Bias in Rating Preferences:

The cumulative line illustrates a classic Pareto distribution, where 80% of orders come from products rated 4.9 and above. This indicates that customers heavily favor top-rated products, and even slight drops in rating below 4.9 can significantly reduce a product’s attractiveness.

Conclusion on Customer Purchasing Behavior Based on Products

Based on an in-depth analysis of customer purchasing behavior, several key factors shape demand on the platform. These include Product Demand by Category, Price Sensitivity, Discount Influence, and Rating Impact. A clear understanding of both high-demand and lower-demand segments provides valuable insights for developing targeted strategies to maximize engagement and drive sales across a diverse customer base.

Analysis of Customer Purchasing Behavior

1. Product Demand by Category

High-Demand Categories: Categories such as fashion (especially men’s and women’s clothes), daily necessities (groceries, health & beauty products), and mobile accessories drive the majority of sales. These products reflect customer needs for both style and everyday essentials, making them core categories for customer engagement and retention.
Lower-Demand Categories: Niche categories like gaming & consoles, automotive, travel & luggage, and tickets & vouchers see fewer orders, but there is still a consistent, smaller audience interested in these products. These segments may cater to specific interests or one-off purchases and represent opportunities for targeted marketing.

2. Price Sensitivity

Strong Preference for Low-Cost Items: The majority of orders are concentrated in lower price ranges, with 80% of sales come from the lower end of the pricing spectrum. Customers show a strong preference for affordable products, suggesting a price-sensitive audience that frequently seeks value-for-money.
Demand for Higher-Priced Items: While limited, there is a segment of customers willing to spend on premium products, particularly in categories like electronics, fashion, and niche segments. This presents an opportunity to capture value-oriented customers who seek quality and exclusivity.

3. Discount Influence

High Impact of Discounts: Discounts drive 60% of orders, demonstrating that promotions and price reductions are key motivators for purchase decisions. Customers are more likely to buy when they perceive added value through cost savings, underscoring the importance of discount-driven strategies.
Discount Effect on Lower-Demand Items: Discounting can be an effective way to boost sales for niche or higher-priced items, encouraging trial purchases and increasing exposure for products that might otherwise receive limited attention.

4. Rating Impact

Ratings as a Primary Driver of Sales: Products with ratings of 4.5 and above dominate sales, with customers heavily favoring well-rated items. Even small differences near the top end (4.7-5.0) can have a significant impact on demand, indicating that high ratings are seen as a strong indicator of quality and reliability.
Limited Interest in Lower-Rated Products: Products rated below 4.0 experience minimal demand, suggesting that customers avoid items with low ratings. However, moderately rated items (4.0-4.4) still see some demand, particularly if they fall within price-sensitive or niche categories.

Strategic Recommendations

1. Marketing Team: Selection of Products for Flash Sales

For New Customers, Focus on High-Demand and Price-Sensitive Items: Prioritize flash sales for high-demand categories such as clothing, health & beauty, groceries, and mobile accessories. These categories align with core customer needs and are more likely to attract high engagement during promotional periods.
For Returning Customers, Segment Flash Sales by Customer Profile: Use customer segmentation to tailor flash sales. For example, frequent buyers of daily necessities could see flash sales on groceries and household items, while fashion-focused customers receive personalized deals on clothing and accessories. This personalization could increase engagement and conversion rates.
Create ‘Bundles with Purpose’: Offer bundled products in high-demand categories (e.g., “Self-Care Essentials” combining health, beauty, and fashion items, or “Stay-Connected Kits” with mobile accessories and tech items). Bundling related items can add value for customers and encourage larger purchases.
Gamified Flash Sales and Discounts: Implement interactive, gamified flash sales where customers can unlock additional discounts or rewards by engaging with the platform (e.g., sharing products, leaving reviews). This approach can make flash sales more engaging and drive additional traffic to the platform.
Tiered Discounts for Higher-Priced Products: For higher-ticket items, use a tiered discount model to encourage purchases (e.g., 10% off if purchased during the first hour of the flash sale, 5% off afterward). This approach taps into urgency while making premium products more accessible.

2. Quality Assurance Team: Action on Low Ratings

Address Issues in High-Volume, Moderate-Rated Products: Focus on products with high sales but moderately low ratings (4.0-4.4). Addressing common complaints for these items (e.g., quality or durability issues) can improve ratings, making them more attractive to potential buyers.
Implement a "Rapid Response" Feedback Program: For products that have experienced a sudden increase in low ratings, initiate a rapid response program where a dedicated team reviews the issue, contacts affected customers, and works with suppliers to make adjustments. This proactive approach can prevent further negative reviews.
Quality Assurance Badging for Improved Products: For products that have received significant quality upgrades, introduce a “Quality Improved” badge to signal to customers that prior issues have been addressed. This transparency can attract hesitant buyers and rebuild trust in previously low-rated products.
Customer Education on Product Usage: If low ratings are due to misunderstandings about product use (e.g., tech gadgets, DIY home items), consider creating product guides, videos, or in-app tips to help customers get the most out of their purchases, reducing complaints and returns.
Enhance Quality for Niche Products: For lower-demand, niche products, addressing quality issues can build loyalty among specialized customer segments. Improving satisfaction in these areas can encourage repeat purchases within smaller, dedicated audiences.

3. Algorithm Engineers: Product Display on Home Page and Search Algorithm

Prioritize Highly Rated, High-Demand Products: Modify search and homepage algorithms to favor products with ratings of 4.5 and above in popular categories. This aligns with customer preferences and boosts visibility for trusted, well-reviewed items, likely enhancing conversion rates.
Integrate Price Sensitivity into Recommendations: Display affordable, well-rated items prominently for customers who historically prefer low-cost purchases, while occasionally highlighting premium items with positive ratings for value-oriented customers.
Dynamic Sorting for Discounted Items: Adjust the algorithm to dynamically prioritize discounted products when active promotions are running. This ensures that price-sensitive customers see relevant discounts first, maximizing engagement during promotional events.
Balanced Display of Niche Products: To increase the visibility of lower-demand products, integrate a diverse selection of niche items in personalized recommendations, particularly for customers who have shown interest in specialized categories (e.g., gaming, travel).
Context-Aware Algorithm Adjustments: Adjust product displays based on time, season, and customer behavior. For example, show more seasonal products (e.g., summer fashion or winter gear) and daily essentials during specific periods. Context-aware displays create a timely and relevant experience for customers.
Personalized Rating-Based Recommendations: Integrate an algorithmic filter that considers individual customer behavior. If a customer frequently buys 4.0-4.4 rated products in specific categories, display similar rating ranges prominently, while highlighting top-rated items for customers who typically purchase 4.5+ products.
Feature Newly Improved Products with Higher Visibility: When products receive an upgrade or have addressed quality issues, give them a temporary boost in visibility on the homepage or search results with an “Improved Quality” label. This approach can reignite interest in products that may have seen demand drops due to prior issues.
Explore an “Explore New Products” Carousel: Implement a rotating carousel on the homepage showcasing new, unrated, or niche products that might otherwise lack exposure. This can encourage discovery among customers who are open to trying something new and different.

4. Brand and Deals Team: Exclusive Partnerships and Promotions

Identify High-Demand Brands for Exclusive Deals: Use sales and engagement data to identify top-performing brands within high-demand categories (such as clothing, beauty, and mobile accessories). Forge exclusive partnerships with these brands to offer special deals, early access to new products, or limited-edition items to boost exclusivity and appeal.
Create Exclusive Product Lines: Collaborate with popular brands to develop exclusive product lines or bundles that are only available on the platform. This can enhance the platform’s uniqueness and encourage customers to choose it over competitors for exclusive items.
Offer Co-Branded Promotions with Popular Brands: For flash sales, work with popular brands to offer co-branded promotions, such as bundled deals, free gifts with purchase, or loyalty points for branded purchases. These types of deals can attract brand-loyal customers and deepen their engagement with the platform.
Launch Brand-Led Campaigns Based on Rating and Sales Data: Feature top-rated, high-demand brands in curated campaigns (e.g., “Top-Rated Brands Week” or “Best-Sellers by [Brand]”). Use positive reviews and sales success stories to market these brands as quality-driven, increasing customer trust.
Negotiated Discounts for Consistently High-Performing Brands: Approach brands that consistently perform well on the platform and negotiate exclusive discounts. Highlight these discounts in special campaigns, positioning them as unique offers available only through this platform.

Summary of Recommendations by Team

Marketing: Tailor flash sales based on customer segments, offer bundles, and use gamification. Experiment with tiered discounts and niche-targeted campaigns.
Quality Assurance: Address quality issues proactively, launch a “Quality Improved” badge, develop educational guides, leverage high-volume buyers for feedback, and automate rating alerts for early intervention.
Algorithm Engineers: Prioritize high-rated items in context, optimize discount displays, introduce personalized rating-based recommendations, feature newly improved products, and showcase new or unrated items for discovery.
Brand and Deals: Partner with high-demand brands to create exclusive offers, limited-edition items, and co-branded campaigns. Negotiate discounts and collaborate on exclusive product lines to build strong brand-driven campaigns that increase customer loyalty.

By implementing these multifaceted recommendations, the platform can enhance customer experience, drive engagement, and maximize sales. This approach leverages data, personalization, quality, and innovation to create a dynamic shopping experience that aligns with customer needs and preferences.

Customer Demographics Analysis

Order Volume Based on Age

Highest Engagement Among 25-35 Age Group:
- The 25-30 and 30-35 age groups lead with the highest checkout volumes, each surpassing 18,000 checkouts. This indicates that customers in their late 20s to early 30s are the most active demographic on Shopee, showing strong engagement and contributing significantly to overall sales.
- 20-25 and 35-40 age groups also exhibit high engagement, with over 10,000 checkouts each, suggesting consistent activity from young adults and early middle-aged shoppers.
Moderate Activity in Middle Age Groups (40-55):
- Engagement decreases for the 40-45 and 50-55 age groups, with checkouts ranging from 5,000 to 10,000. This may suggest that middle-aged customers are active but not as engaged as younger demographics.
Lower Engagement from Older Age Groups (55+):
- Shoppers aged 55 and above (55-60, 60-65, and 65-70) have the lowest checkout numbers, with fewer than 5,000 checkouts each. This suggests lower engagement from older customers, possibly due to less familiarity with online shopping or lower comfort with e-commerce platforms.

Order Volume Based on Gender

Balanced Gender Distribution:
- Shopee’s customer base is nearly evenly split, with 51% of orders from females and 49% from males. Both genders actively participate on the platform, indicating a broad appeal across gender lines.

Analysis of Revenue by Age and Gender

1. Highest Revenue-Generating Segments:

The 25-30 age group generates the highest revenue overall, with females generating $453,015 and males $371,508. This age group is likely the most active in purchasing, possibly due to higher disposable income and a stronger inclination toward online shopping.
The 30-35 age group also shows high revenue, with females contributing $419,192 and males $393,163. These two age groups are the key drivers of revenue in this dataset.

2. Mid-Level Revenue Segments:

The 20-25 and 35-40 age groups contribute moderately high revenue. For the 20-25 group, females generate $230,007, while males contribute $209,235. In the 35-40 group, females generate $357,011 and males $299,582.
These age groups may represent younger adults establishing purchasing habits and mid-career professionals with more disposable income.

3. Lower Revenue Contribution in Older Age Groups:

Revenue contributions decline significantly in older age brackets, particularly from 45 and above. The 50-55 age group shows a decline, with females generating $82,142 and males $133,977. This downward trend continues into the 55-60 and 60-65 age groups, with minimal contributions.
These age segments may engage less in online shopping or have lower discretionary spending, although they still represent a notable part of the customer base.

4. Gender-Based Revenue Trends:

Across the most active age groups, females tend to generate higher revenue than males, especially in the 25-30 and 30-35 age groups. This trend might indicate higher engagement or a higher purchase frequency among female customers, likely influenced by categories like fashion, beauty, and lifestyle products.
In the older age groups, males contribute slightly more than females, suggesting a shift in purchasing behavior and product preferences.

Key Takeaways

Core Demographic: The 25-35 age range, especially females, represents the highest revenue-generating demographic. This group likely has both purchasing power and frequent engagement with online shopping.
Opportunity in Younger and Mid-Career Segments: The 20-25 and 35-40 segments contribute significantly, suggesting potential for targeted promotions to further increase engagement.
Older Demographics as Emerging Markets: Although revenue decreases significantly in the 45+ age groups, these demographics still play an essential role. Targeted campaigns might unlock additional revenue potential from these customers.

Recommendations

1. Marketing Strategies

Focus on 25-35 Age Group: Design marketing campaigns focused on the 25-35 age range, especially for females, promoting popular categories like fashion, beauty, and electronics.
Increase Engagement for 20-25 and 35-40 Age Groups: Personalized promotions on trending categories like lifestyle and tech products might help engage these segments.
Develop Campaigns for Older Age Groups: For customers 45+, create campaigns around products aligned with their interests, such as wellness, home, and leisure. Offering accessible support and educational resources might enhance their experience.

2. Product Selection and Promotion

Emphasize Gender-Specific Products: Highlight categories such as fashion and lifestyle for females, especially in the younger segments, while promoting products like home improvement and tech gadgets for males in older age brackets.
Seasonal and Occasion-Based Promotions: Run seasonal promotions for the core demographic (25-35) and consider holiday or event-driven promotions for older demographics.

3. Personalization and Customer Experience

Personalized Product Recommendations: Use age and gender data to tailor recommendations, with trending items for younger customers and more practical, lifestyle-oriented suggestions for older users.
Improve Experience for Older Demographics: Simplify navigation and provide easy-to-understand product information for older customers. Consider customer support with a focus on assistance for this demographic.

4. Data-Driven Insights and Feedback Loops

Collect Feedback from High-Revenue Segments: Capture feedback from the 25-35 age group to fine-tune marketing strategies and understand product preferences.
Track Trends in Lower Revenue Segments: Monitor trends within the 45+ age groups for any rising interest in specific categories, adjusting marketing efforts accordingly.

Key Takeaways

Core Demographic: The 25-40 age range, particularly females, represents the highest revenue-generating demographic. This group likely has high disposable income, frequent online purchasing habits, and an interest in a broad range of products.
Opportunity in Younger and Mid-Career Segments: While the 20-25 and 40-45 segments contribute moderately, there is potential to further engage these groups through tailored promotions or product recommendations.
Older Demographics as Emerging Markets: Although revenue drops significantly in older age groups, they still represent an important part of the customer base. Tailored marketing strategies could help capture additional revenue from this demographic.

Recommendations

1. Marketing Strategies

Focus on 25-35 Age Group: Design marketing campaigns targeting the 25-35 age range, particularly females, with products in high-demand categories (fashion, beauty, electronics, and home essentials). Exclusive deals, loyalty rewards, and flash sales could maintain engagement within this core demographic.
Increase Engagement for 20-25 and 40-45 Age Groups: For younger adults and mid-career professionals, consider personalized promotions that highlight popular categories such as lifestyle, fitness, and tech products. These age groups may respond well to trend-driven and convenience-focused campaigns.
Develop Campaigns for Older Age Groups: Create campaigns for the 45+ segment that focus on products aligned with their needs, such as health, home, and leisure. Offering educational content, guides, and simple navigation could improve their engagement with the platform.

2. Product Selection and Promotion

Highlight Gender-Specific Products: Given the higher revenue contribution from females, promote categories like fashion, beauty, and home essentials prominently. However, for male-dominated segments in older age groups, emphasize products such as electronics, automotive accessories, and home improvement items.
Seasonal and Occasion-Based Promotions: For the 25-35 core demographic, run targeted seasonal promotions (e.g., holiday sales, back-to-work deals) that align with lifestyle needs and seasonal demand. For the 50+ age range, consider occasion-based promotions (e.g., holiday gifting, family-oriented products) to encourage additional purchases.

3. Personalization and Customer Experience

Personalized Recommendations: Use age and gender data to personalize product recommendations. For instance, suggest trending fashion or tech products to younger customers, while recommending practical, high-quality home and wellness items to older customers.
Optimize User Experience for Older Demographics: Simplify the interface and provide clear product information for older age groups to enhance their shopping experience. Consider offering customer support with dedicated assistance for this demographic to address any purchase concerns.

4. Data-Driven Insights and Feedback Loops

Collect Feedback from High-Revenue Demographics: Gather feedback from the 25-35 age group to understand product preferences and potential areas for improvement. This data can inform targeted marketing efforts and product selection.
Monitor Trends in Lower Revenue Segments: Keep track of emerging trends within the 45+ age groups to identify opportunities for growth. For example, observe any increasing interest in specific categories or products and adjust marketing accordingly.

Overview on Individual Customers

Making Customer Metrics

1. Total Spending

Description: The total amount each customer has spent over all their purchases.
Purpose: Helps identify high-value customers and their overall contribution to revenue.

2. Average Order Value

Description: The average amount spent per order, calculated as total spending divided by checkout count over the last year.
Purpose: Indicates the typical spending level per transaction, useful for segmenting customers based on their purchasing power.

3. Campaign Count

Description: The total number of unique campaign days a customer has participated in.
Purpose: Measures engagement with promotional campaigns, allowing Shopee to identify customers responsive to promotions.

4. Average Order Per Month

Description: The average number of orders placed per month, based on the checkout count over the last year.
Purpose: Indicates purchase frequency on a monthly basis, helping to identify regular buyers versus occasional shoppers.

5. Average Discount Percentage

Description: The average discount percentage a customer receives across all orders.
Purpose: Shows sensitivity to discounts, allowing Shopee to tailor marketing efforts based on discount preferences.

6. Percentage of Items Discounted

Description: The percentage of a customer’s orders that include a discount.
Purpose: Measures how often a customer purchases discounted items, useful for identifying price-sensitive customers.

7. Average Purchase Frequency

Description: The average number of days between purchases for each customer.
Purpose: Provides insights into purchasing intervals, enabling predictions of when customers might buy again and identifying high-frequency shoppers.

8. Product Variety

Description: The number of unique product categories a customer has purchased from.
Purpose: Indicates diversity in purchasing behavior, with higher variety suggesting broader interests.

9. Checkout Count Last 1 Year

Description: The total number of checkouts a customer has made in the past year.
Purpose: Shows the overall engagement level with the platform, helping in identifying loyal customers.

10. Last Checkout Day

Description: The number of days since the customer’s last checkout.
Purpose: Helps identify dormant customers who may need re-engagement strategies.

11. Last Login Day

Description: The number of days since the customer last logged into the platform.
Purpose: Indicates customer engagement and platform activity, with recent logins suggesting active users.

12. Login Count (Last 10 Days & Last 60 Days)

Description: The number of times a customer logged in during the last 10 and 60 days.
Purpose: Shows recent engagement trends, helping to identify users who are becoming more or less active over time.

Monitoring Customer Metrics and Establishing Good Values

Effective monitoring of these customer metrics involves regular tracking and defining benchmarks or thresholds for each metric to evaluate customer engagement, spending behavior, and responsiveness to promotions. Here is a breakdown of how to monitor these metrics and suggested monitoring technique that can guide strategic decisions.

1. Total Spending

Good Value: Higher total spending is ideal. Identifying and nurturing high-spending customers can increase customer lifetime value (CLV).
Desired Trend: Increase.
Monitoring: Track total spending growth quarterly and annually per customer. Monitor for consistent increases, and flag any significant drops in spending among regular customers as a potential risk of churn.

2. Average Order Value (AOV)

Good Value: Higher AOV indicates higher purchasing power and is desirable as it implies customers are willing to spend more per transaction.
Desired Trend: Increase or at least remain stable.
Monitoring: Track AOV on a monthly and quarterly basis. A downward trend may indicate a shift towards lower-priced items, suggesting an opportunity to encourage bundling or cross-selling.

3. Campaign Count

Good Value: Higher campaign count means stronger engagement with promotions, indicating that customers are interested in deals and offers.
Desired Trend: Increase or maintain high engagement.
Monitoring: Track campaign participation across the last 6 months to identify regular participants. Sudden drops in engagement with campaigns from previously active customers could signal reduced interest, prompting re-engagement with targeted campaigns.

4. Average Orders Per Month

Good Value: Higher monthly order frequency is desirable as it indicates recurring engagement with the platform.
Desired Trend: Increase or maintain high consistency.
Monitoring: Track changes in monthly frequency over the last 12 months. Declines in regular monthly orders could suggest an opportunity for re-engagement or incentives to drive monthly purchases.

5. Average Discount Percentage

Good Value: Stability is generally preferred. High discount usage can indicate discount sensitivity, but extreme increases may reduce overall profit margins.
Desired Trend: Stable or moderate increase for discount-sensitive customers.
Monitoring: Track the average discount percentage each quarter. A high increase could indicate a shift toward more discount-seeking behavior, suggesting the need for balancing promotions with profitability.

6. Percentage of Items Discounted

Good Value: Stability or moderate increase, particularly for discount-sensitive customers. Higher values mean a customer is very discount-driven, but for profitability, this should ideally not be excessive.
Desired Trend: Stable or moderate increase for high discount-seekers.
Monitoring: Track quarterly to observe discount usage trends. Rising trends among high-value customers could warrant offering exclusive deals without compromising on margins.

7. Average Purchase Frequency

Good Value: Higher frequency (shorter intervals between purchases) is ideal, as it indicates strong repeat engagement.
Desired Trend: Increase (decrease in days between purchases).
Monitoring: Track purchase frequency every month. Any lengthening of the purchase interval for regular customers may indicate reduced engagement, suggesting they may need a reminder or offer to re-engage.

8. Product Variety

Good Value: Higher product variety indicates a customer with broad interests who is likely to stay engaged. This could also imply a higher likelihood of cross-selling opportunities.
Desired Trend: Increase in variety.
Monitoring: Monitor product variety annually. If variety decreases, it may suggest a narrowing of interests or a reduction in engagement, which could benefit from exposure to new product categories.

9. Checkout Count Last 1 Year

Good Value: A high checkout count over the year indicates loyalty and regular platform engagement.
Desired Trend: Increase.
Monitoring: Track on an annual basis and compare with previous years to assess customer loyalty trends. A drop in checkout counts among loyal customers may signal potential churn risk.

10. Last Checkout Day

Good Value: The fewer days since the last checkout, the better, as it implies recent engagement.
Desired Trend: Decrease (fewer days since last checkout).
Monitoring: Monitor weekly for high-value or loyal customers. An increasing trend in days since last checkout may indicate declining engagement, which may require a targeted promotion to encourage a return purchase.

11. Last Login Day

Good Value: Similar to the checkout metric, the fewer days since the last login, the better, as it reflects recent engagement with the platform.
Desired Trend: Decrease.
Monitoring: Monitor weekly to identify inactive customers. Rising values among regular or high-value customers can prompt targeted re-engagement strategies.

12. Login Count (Last 10 Days & Last 60 Days)

Good Value: A high login count in recent periods (last 10 days and last 60 days) indicates strong engagement with the platform.
Desired Trend: Increase.
Monitoring: Track these values monthly to detect any early signs of disengagement. Decreasing login counts may suggest waning interest, particularly for customers who were previously active, and may warrant re-engagement efforts.

Additional Monitoring Strategies

Set Alerts for Key Changes: Set up automated alerts for significant drops or increases in core metrics, especially for high-value customers (e.g., a sudden increase in days since last checkout, significant drops in purchase frequency, or last login).
Track Growth and Decline Patterns: Use dashboards to regularly track metrics that should ideally increase (e.g., total spending, AOV, frequency) and those that should be stable or moderately increase (e.g., discount-related metrics). This will help in quickly identifying customer behavior shifts.
Implement Cohort Analysis: Analyze customer cohorts based on similar metrics (e.g., high-frequency buyers, high AOV customers) to identify changes in engagement and spending habits within each cohort. This can help in creating tailored engagement strategies.
Benchmark and Compare with Industry Standards: Whenever possible, benchmark these metrics against industry averages to understand if the platform’s performance is competitive. High AOV, purchase frequency, and engagement with promotions are typically good indicators of platform health relative to competitors.
Annual Review of Desired Value Ranges: At least once a year, review the desired value ranges and thresholds for each metric to ensure they align with business goals, customer behavior changes, and industry trends.

By monitoring these metrics with these goals and trends in mind, Shopee can better identify and nurture high-value customers, engage price-sensitive users effectively, and prevent customer churn, ultimately driving sustainable growth and customer satisfaction.

Descriptive Customer Metrics for Shopee in 2019

1. Total Spending

Mean: $1,703.97
Median (50%): $249.92
Max: $171,898.58
Interpretation: There’s a high variability in total spending, with a significant difference between the mean and median, indicating a small number of high-spending customers skewing the average. The majority of customers have a relatively low total spend, with 75% spending less than $1,117. This suggests that most customers are occasional buyers with modest spending, while a few high-spenders are driving the higher average.

2. Average Order Value (AOV)

Mean: $46.74
Median: $22.99
Max: $3,580.93
Interpretation: The mean AOV is $46.74, but the median is $22.99, indicating that while some customers make high-value purchases, the typical order is much lower. This again shows a skewed distribution, with the majority of customers making smaller purchases.
Insights from Visuals: The log-transformed AOV shows a positively skewed distribution, where most customers’ AOV is relatively low, but a small subset has higher spending per order.

3. Average Discount Percentage

Mean: 28.28%
Median: 28.39%
Interpretation: Most customers receive an average discount of around 28%. The consistency between mean and median suggests a relatively normal distribution for discount percentages, with most customers receiving moderate discounts.
Insights from Visuals: Most customers fall within the 20-40% range for average discounts, with a steady cumulative increase.

4. Campaign Count

Mean: 2.63
Median: 1
Max: 20
Interpretation: Campaign participation is generally low, with the median indicating that most customers participate in only one campaign per year. A small group of highly engaged customers (top 25%) participates in three or more campaigns, suggesting potential for greater campaign engagement among the broader customer base.
Insights from Visuals: The campaign count distribution is heavily skewed to the left, with most customers participating in fewer than 5 campaigns. The cumulative line also rises sharply, indicating that only a few customers are highly engaged with campaigns.

5. Percentage of Items Discounted

Mean: 59.52%
Median: 60%
Interpretation: The median and mean values are close, suggesting a balanced distribution with most customers purchasing discounted items around 60% of the time. This indicates a strong preference for discounted items among the majority of customers, reflecting high price sensitivity.
Insights from Visuals: The distribution here is fairly even, with a slight skew toward higher discount percentages. The cumulative curve is gradual, showing a mix of low to high discount sensitivity among customers

6. Average Orders Per Month

Mean: 3.31
Median: 0.83
Max: 348.33
Interpretation: The large gap between the mean and median indicates a few high-frequency buyers skewing the average. Most customers place less than one order per month, with 75% placing fewer than two orders monthly, highlighting a need to increase purchase frequency among the majority.

7. Average Purchase Frequency (Days between Purchases)

Mean: 27.41 days
Median: 20 days
Interpretation: The median of 20 days suggests that a significant number of customers make a purchase approximately every three weeks, though the mean is slightly higher due to customers with longer gaps between purchases.
Insights from Visuals: The distribution shows a steep drop, indicating that the majority of customers purchase infrequently, with cumulative data showing a rapid plateau.

8. Product Variety

Mean: 8.53 categories
Median: 7 categories
Interpretation: Most customers purchase from around 7-9 categories. This suggests some level of product variety in customer purchases, but there may be opportunities to encourage more cross-category purchases to increase product variety.
Insights from Visuals: The distribution is relatively broad, with many customers purchasing across 4-12 categories, though the cumulative curve flattens at higher variety levels.

9. Checkout Count Last 1 Year

Mean: 39.71 checkouts
Median: 10 checkouts
Max: 4,180 checkouts
Interpretation: The mean is much higher than the median, indicating a small number of very frequent buyers. The majority of customers have lower checkout counts, with 75% having fewer than 22 checkouts per year.
Insights from Visuals: The log-transformed distribution reveals a heavy skew toward lower checkout counts, with most customers having fewer than 8 checkouts. The cumulative line shows that about 80% of customers have relatively few checkouts, confirming low engagement among a majority.

10. Last Checkout Day

Mean: 99.97 days since last checkout
Median: 68 days
Interpretation: Many customers have not made a recent purchase, with a median of over two months since the last checkout. This suggests potential churn among a significant portion of the customer base, warranting re-engagement efforts.
Insights from Visuals: This metric is skewed, with a significant portion of customers having high last checkout day counts, indicating they haven't made recent purchases.

11. Login Count (Last 10 Days & Last 60 Days)

Last 10 Days:
- Mean: 4.42, Median: 1
Last 60 Days:
- Mean: 25.26, Median: 7
Interpretation: The median login counts in both periods are relatively low, suggesting that a majority of customers are not logging in frequently. However, the higher mean values indicate that a smaller subset of customers logs in very often.
Insights from Visual (Login Counr Last 60 Days)s: The log-transformed login count shows a similar skew to checkout and purchase frequency metrics, with most customers logging in infrequently.

Analysis on Campaign Presence

Campaign Awareness Among Customers:
- With 66.5% of customers having participated in at least one campaign, it’s clear that the presence of campaigns has reached a significant portion of the customer base. This suggests that more than half of the customers are aware of and willing to engage with Shopee’s promotional campaigns.
Potential for Increased Engagement:
- 33.5% of customers have not participated in any campaigns, indicating an opportunity to increase engagement with this segment. This group might not be fully aware of the benefits or offerings of campaigns or may require additional incentives or tailored campaigns to encourage their participation.
Opportunity to Enhance Campaign Appeal:
- Understanding why some customers haven’t participated can provide insights for Shopee to make campaigns more attractive and accessible. Tailoring campaigns to the interests or needs of different segments, especially for non-participants, could improve their overall appeal.

Analysis of Shopee’s Current Customer Metrics

Based on the analysis of Shopee’s customer metrics, including campaign participation, spending patterns, frequency of purchases, and engagement with discounts, we can assess strengths, areas for improvement, potential solutions, and necessary stakeholders.

1. Strengths in Current Metrics

A. High Discount Engagement and Price Sensitivity

What’s Good: Shopee’s customer base shows strong engagement with discounts, with around 60% of items being discounted and an average discount percentage of approximately 28%. This indicates that the discount strategies are reaching the intended audience effectively.
Why It’s Effective: The high engagement with discounted items demonstrates that price sensitivity is well understood, and the discount strategies align with customer preferences.
Recommendation: Maintain a balance of moderate discounts to sustain engagement without sacrificing profitability. Occasional flash sales with larger discounts can stimulate additional engagement and excitement.

B. Campaign Awareness and Participation

What’s Good: With 66.5% of customers having participated in at least one campaign, Shopee has successfully raised awareness of its promotional campaigns. This level of engagement suggests that campaigns are visible to a majority of the user base.
Why It’s Effective: High campaign awareness is crucial for customer engagement and sales, as campaigns drive short-term traffic and can convert browsing customers into buyers.
Recommendation: Continue promoting campaigns effectively and monitor participation trends over time to maintain or increase this percentage.

C. Strong Diversity in Product Variety

What’s Good: With a mean of 8.5 categories per customer, Shopee’s customers are exploring a wide variety of products. This reflects the platform’s success in offering a broad product range and encouraging cross-category purchases.
Why It’s Effective: Higher product variety per customer indicates greater engagement across categories, which can increase overall spending and customer retention.
Recommendation: Maintain this variety and consider cross-selling strategies, such as bundling complementary products or personalized product recommendations to encourage even broader exploration.

2. Areas for Improvement

A. Low Average Order Value (AOV)

What Needs Improvement: The average order value (AOV) is relatively low, with a mean of $46.74 and a median of $22.99. This suggests that while customers are purchasing, they are often making smaller, lower-value transactions.
Improvement Strategy:
- Increase Basket Size: Encourage larger orders through bundling offers, minimum purchase discounts (e.g., “Save $10 on orders over $50”), or free shipping thresholds.
- Cross-Selling and Upselling: Highlight complementary items during checkout to increase the total order value.
Stakeholders Involved: Marketing Team (for creating bundle offers), Product Team (to improve the recommendation engine for cross-sell/upsell), and UX/UI Team (to display upsell items prominently at checkout).

B. Infrequent Purchase Frequency

What Needs Improvement: The average purchase frequency indicates that many customers are infrequent buyers, with a median of 20 days between purchases. A significant portion of the customer base has long intervals between transactions, indicating room to encourage repeat buying.
Improvement Strategy:
- Introduce Loyalty Programs: Create loyalty programs or reward systems that incentivize frequent purchases, such as offering discounts or points that accumulate with each purchase and can be redeemed for future discounts.
- Regular Re-Engagement Campaigns: Send reminders or personalized product recommendations to customers after a certain period of inactivity (e.g., after 20 days without a purchase).
Stakeholders Involved: Marketing Team (for loyalty program design), CRM/Customer Engagement Team (for sending automated re-engagement messages), and Data Analytics Team (to monitor purchase frequency and test re-engagement effectiveness).

C. Low Campaign Participation Among One-Third of Customers

What Needs Improvement: While a majority of customers have participated in campaigns, 33.5% have not engaged with any promotions. This segment represents an untapped opportunity.
Improvement Strategy:
- Educational and Awareness Campaigns: Implement campaigns that educate non-participants on the benefits of engaging with promotions and offer first-time participation incentives.
- Targeted Campaigns Based on Purchase History: Use customer data to create personalized, highly relevant campaigns that appeal to the specific preferences of non-participating customers.
Stakeholders Involved: Marketing Team (to design tailored campaigns), Data Analytics Team (to segment customers and personalize campaigns), and CRM/Customer Engagement Team (for targeted messaging and outreach).

D. High Inactivity for Last Checkout and Login Days

What Needs Improvement: The last checkout and login days indicate that a significant number of customers are inactive, which poses a risk for churn. The median last checkout day count of 68 suggests that many customers have not made a purchase in over two months.
Improvement Strategy:
- Re-Engagement Strategies for Dormant Customers: Develop targeted re-engagement campaigns for customers who have not logged in or made a purchase recently. “We Miss You” discounts, new product notifications, or exclusive deals can encourage them to return.
- Implement Automated Notifications: Use push notifications and email reminders to keep inactive users engaged with the platform and encourage browsing or purchasing.
Stakeholders Involved: CRM/Customer Engagement Team (for re-engagement campaigns), Data Analytics Team (to identify inactive customers and measure reactivation rates), and Product/Engineering Team (for push notification and email automation).

E. Low Participation in Multiple Campaigns

What Needs Improvement: Although 66.5% of customers have participated in a campaign, most engage in only one or two. Increasing repeat campaign participation could enhance customer lifetime value and overall engagement.
Improvement Strategy:
- Design Tiered Campaigns: Create multi-level campaigns that encourage participation over several stages (e.g., “Complete three campaign purchases to unlock an exclusive reward”).
- Incentivize Repeat Campaign Engagement: Offer rewards for customers who participate in multiple campaigns within a set period, such as additional discounts or entry into a prize draw.
Stakeholders Involved: Marketing Team (to design campaign rewarding structures), CRM Team (for sending reminders and notifications about tiered rewards), and Data Analytics Team (to track campaign participation trends).

Summary of Recommendations and Stakeholder Involvement

Metric	Improvement Needed	Possible Solution	Stakeholders Involved
Average Order Value (AOV)	Increase AOV by encouraging larger basket sizes	Bundle offers, minimum purchase discounts, cross-sell/upsell items	Marketing, Product, UX/UI
Purchase Frequency	Shorten intervals between purchases	Loyalty programs, regular re-engagement messages	Marketing, CRM, Data Analytics
Campaign Participation	Increase engagement among non-participants	Educational campaigns, first-time incentives, personalized promotions	Marketing, Data Analytics, CRM
Customer Inactivity	Reduce last checkout/login days	Re-engagement campaigns, automated notifications	CRM, Data Analytics, Product/Engineering
Repeat Campaign Engagement	Encourage multiple campaign participations	Tiered campaigns, multi-level rewards	Marketing, CRM, Data Analytics

By focusing on these areas of improvement, Shopee can enhance customer engagement, improve lifetime value, and increase overall sales. Collaboration across multiple teams—especially Marketing, Data Analytics, CRM, and Product—will be essential for implementing these strategies effectively and ensuring they align with customer needs and preferences.

Part 2: Segmenting Customers Based on Purchasing Habits

Why Actionable Customer Segmentation is Crucial?

Customer segmentation is a vital strategy for any platform aiming to maximize customer engagement, retention, and lifetime value. By dividing the customer base into distinct groups based on purchasing habits, Shopee can tailor its marketing, promotions, and engagement efforts to better meet the needs and preferences of each group.

A well-designed, actionable segmentation—one that’s precise enough to capture meaningful behavioral patterns but broad enough for easy implementation—empowers Shopee to deploy effective, efficient, and scalable strategies. Here’s why this is crucial:

1. Maximizing Marketing and Campaign ROI

Targeted Campaigns: Different customers respond to different types of promotions, discounts, and campaign formats. By segmenting based on factors like discount sensitivity, purchase frequency, and average order value, Shopee can tailor campaigns that resonate with each group.
Efficient Use of Resources: Without segmentation, marketing spend is likely spread thin across uninterested customers. With targeted campaigns, Shopee can allocate resources more effectively, reducing spend on low-engagement customers while investing more in high-potential segments.
Example: High-value customers may appreciate early access or loyalty rewards, while discount-sensitive customers respond better to flash sales or high-discount offers.

2. Improving Customer Retention and Reducing Churn

Understanding At-Risk Customers: Segmentation helps Shopee identify customers with declining engagement or long purchase intervals, signaling a risk of churn. Actionable segmentation enables Shopee to quickly identify and re-engage these customers through targeted retention strategies.
Personalized Re-Engagement: By segmenting customers based on their activity levels (e.g., recent purchasers vs. dormant customers), Shopee can deploy re-engagement campaigns tailored to bring back customers who haven’t visited or purchased in a while.
Example: Dormant customers might respond well to “We Miss You” campaigns with a special discount, whereas frequent customers might be incentivized to purchase with loyalty rewards or “members-only” offers.

3. Enhanced Product Recommendations and Shopping Experience

Relevant Product Recommendations: Segmentation allows Shopee to deliver personalized recommendations based on a customer’s purchasing history, category preferences, and engagement levels, making the shopping experience more relevant and enjoyable.
Increased Engagement through Personalization: When customers feel that Shopee understands their preferences, they’re more likely to return and explore. Actionable segmentation enhances this experience, allowing Shopee to target specific products and categories to each segment.
Example: A segment interested in electronics may receive recommendations for the latest gadgets or accessories, while customers who frequently buy household items might receive updates on new home essentials.

4. Effective Cross-Selling and Upselling

Boosting Average Order Value (AOV): Segmentation helps identify customers who may be more receptive to upselling and cross-selling tactics. For example, high-AOV customers could be introduced to premium items, while frequent buyers might be offered bundles or complementary products.
Category-Based Recommendations: Different segments may favor certain product categories. Shopee can leverage this insight to design cross-category promotions that cater to these preferences, ultimately increasing the number of items in each purchase.
Example: Customers purchasing baby products could be cross-sold items in related categories, such as toys or home organization, while customers buying electronics might be encouraged to add accessories or warranties.

5. Improving Customer Loyalty and Advocacy

Building Long-Term Relationships: Loyal customers are more likely to become brand advocates. By segmenting based on loyalty and repeat purchasing behavior, Shopee can strengthen relationships with its most devoted customers through exclusive perks and personalized experiences.
Encouraging Advocacy: A segment of highly engaged customers can be nurtured to become advocates, sharing their positive experiences and referring friends and family.
Example: High-loyalty segments could receive special “insider” offers, early product launches, or exclusive access to events, encouraging them to spread positive word-of-mouth and attract more users to Shopee.

Recommendation on Strategic Segmentation for Shopee

1. Customer-Facing Segmentation: Loyalty Tiers Based on Percentile Rank

Concept: Segment customers based on percentile rank, creating a loyalty tier system that is transparent to the customers themselves. This would encourage a competitive mindset as customers see their ranking and strive to achieve higher tiers.
Tiers:
- Classic (up to 50%)
- Silver (top 50-75%)
- Gold (top 75-90%)
- Platinum (top 90-100%)
Benefits of This Approach:
- Psychological Engagement through Competition: By explicitly stating percentile ranks, Shopee can tap into customers’ competitive nature, encouraging them to move up to higher tiers for more benefits.
- Reward-Based Retention: Each tier can offer increasingly valuable benefits (e.g., free shipping, exclusive discounts, early access to sales), incentivizing customers to remain active and increase spending.
- Transparency and Goal Orientation: Explicitly defined ranks and tier benefits give customers a clear sense of achievement and goals, which can foster brand loyalty.
- Exclusive Access to New Arrivals and Events: Offer early access to sales or new products for Gold and Platinum members, which could make these tiers more desirable.
- Personalized Tier Rewards: Customize rewards based on each customer’s spending patterns or preferred product categories. For instance, a Platinum customer who frequently buys electronics could receive targeted discounts on tech products.

2. Company-Facing Segmentation: Detailed Segmentation with RFM Analysis and Behavioral Metrics

Concept: Develop more nuanced segments using RFM (Recency, Frequency, Monetary) analysis, combined with other behavioral metrics such as discount sensitivity, product variety, preferred day, and preferred time. This segmentation isn’t directly shared with customers but provides internal insights for personalized engagement and resource allocation.
Recommended Company-Facing Segmentation Dimensions:

A. RFM Analysis (Core of Loyalty Segmentation)
- Recency: How recently a customer made a purchase. This metric helps identify active, lapsed, and dormant customers, which is critical for designing re-engagement campaigns.
- Frequency: How often a customer makes a purchase. Frequent buyers can be targeted with loyalty programs, while infrequent buyers may need reminders or incentives.
- Monetary: How much a customer spends. High-value customers may be suited for exclusive rewards and upselling, while lower-value customers might respond well to small discounts or low-priced recommendations.
Actionable Insight: RFM analysis allows Shopee to create highly targeted campaigns. For instance, high-revenue, frequent, and recent customers can be classified as “VIPs,” while high-spending but infrequent customers could be targeted with re-engagement incentives.

B. Activity Status
- Description: Track customer activity status (e.g. Active and Inactive) based on recent purchase activity with Shopee
- Actionable Insight: Customers who have been inactive can be targeted with more marketing to induce a re-engagement.
C. Discount Sensitivity
- Description: Identify customers who predominantly buy discounted items, as well as those who buy regardless of discounts.
- Actionable Insight: Use this segmentation to design specific offers. Discount-sensitive customers could receive flash sale notifications, while less discount-sensitive customers might be offered value-added services like faster shipping or premium options.
D. Preferred Day and Time of Purchase
- Description: Identify the days and times each segment prefers to shop.
- Actionable Insight: Schedule personalized notifications and targeted campaigns to reach customers during their preferred shopping windows. For instance, if a customer frequently shops on weekends, send them promotions timed for weekend access.
E. Campaign and Promotion Responsiveness
- Description: Track customer engagement with past campaigns and promotions to determine their responsiveness.
- Actionable Insight: Customers who have participated in multiple campaigns may appreciate personalized campaign invites, while those with lower engagement might respond better to single, high-value offers.

Making Customer Clusters Table

1. Shopee Loyalty Tiers

The Shopee Loyalty Tier system is a customer-facing segmentation model designed to classify customers based on their total spending and checkout count over the past year. By using quantile thresholds, this system segments customers into four distinct tiers—Classic, Silver, Gold, and Platinum—to reflect varying levels of engagement, spending, and loyalty.

Loyalty Tier Descriptions

Classic (0-50%)
- Benefits: Access to basic discounts, standard promotions, and general offers available on Shopee.
- Goal: Encourage increased engagement and spending to motivate customers to reach higher tiers.
Silver (50-75%)
- Benefits: Enhanced discounts, occasional early access to sales, and minor perks that provide a better value proposition compared to the Classic tier.
- Goal: Increase their loyalty by offering incentives that encourage more frequent purchases or higher spending.
Gold (75-90%)
- Benefits: Exclusive discounts, priority customer support, early access to major sales events, and special seasonal rewards.
- Goal: Retain their loyalty and encourage them to move into the Platinum tier through targeted upselling and premium offers.
Platinum (90-100%)
- Benefits: Highest level of discounts, VIP customer support, first access to new products and sales, and exclusive Platinum-only deals and events.
- Goal: Strengthen long-term loyalty by providing them with an elite experience and ensuring they continue to see unique value in remaining at the top tier.

Mechanism for Loyalty Tier Updates

Update Frequency: Tiers are updated monthly. Each month, customers’ total spending and checkout count from the past 12 months are re-evaluated, and their tier is adjusted if they have moved up or down based on the quantile thresholds.
Coupon Validity: Since tiers are updated monthly, coupons and benefits for each tier are valid for one month. At the beginning of each month, customers’ tiers are recalculated, and new coupons or perks are issued based on their updated status.
Communication of Tier Updates:
- Customers are notified of their updated tier at the start of each month via email, push notifications, or in-app messages.
- Shopee provides a summary of the benefits available to them in their current tier and encourages them to maintain or improve their standing.
- By explicitly showing customers their percentile rank (e.g., “You’re in the top 20% of Shopee customers!”), Shopee fosters a sense of achievement and motivates customers to reach the next tier.

2. RFM Segmentation

The RFM (Recency, Frequency, Monetary) model is a segmentation technique used to analyze customer behavior and identify loyalty patterns. By assessing how recently, how often, and how much a customer spends, RFM provides insights into customer engagement levels and helps Shopee tailor marketing efforts accordingly. This model is enhanced by a unique feature: discounting past values to prioritize recent transactions, similar to the concept of present value in finance.

Components of RFM Analysis

Recency: Measures the number of days since the last purchase. Customers with recent transactions are typically more engaged and likely to respond to marketing outreach.
Frequency: Captures the total number of purchases (or checkout count) in the last year. A higher frequency suggests a loyal customer who frequently interacts with Shopee.
Monetary: Represents the total spending over the last year, with recent spending weighted more heavily. This gives a realistic view of each customer’s value, favoring current purchasing behavior over historical transactions.

Discounting for Frequency and Monetary Values

To make recent purchases more influential in the model, Shopee could considering using 10% annual discounting (compounded daily) for frequency and monetary values. This approach, akin to the financial concept of present value, reduces the weight of older transactions and increases the importance of recent spending. The compounded daily discounting effectively captures the ongoing value of recent behavior, allowing Shopee to make decisions that reflect current customer loyalty and spending habits.

Mechanism for RFM Analysis

1. Update Frequency

Recommended Update Frequency: Monthly
- Rationale: Monthly updates strike a balance between actionability and operational efficiency. This frequency allows Shopee to regularly adjust segmentation and launch targeted marketing campaigns, re-engagement efforts, and loyalty rewards based on current customer behavior.
- Benefits of Monthly Updates:
  - Responsive to Recent Behavior: Monthly updates capture recent engagement changes without being overly sensitive to short-term fluctuations.
  - Actionable Insights for Marketing: With monthly RFM scores, Shopee can initiate targeted campaigns, adjust offers, and refine loyalty rewards to engage customers based on their latest behavior.
  - Resource Efficiency: Updating RFM scores monthly reduces the data processing and analytical workload compared to weekly updates, while still providing relevant insights for decision-making.
Alternative Option: Weekly Updates (for high-intensity promotional periods)
- Rationale: During peak shopping seasons or promotional events, such as holiday sales or flash sales, weekly RFM updates might be beneficial. This allows Shopee to respond more quickly to rapid changes in customer behavior.
- Suggested Use: If weekly updates are implemented, they could be limited to critical campaign periods (e.g., 11.11 or 12.12) to optimize marketing effectiveness without overburdening resources.

2. Application of Discounting in RFM Calculation

Daily Compounded Discounting Factor: The 10% annual discounting factor is compounded daily, assigning higher weight to recent transactions and gradually decreasing the importance of older purchases.
Implementation for Recency and Monetary:
- Frequency: Older purchases contribute less to the recency score, ensuring that customers with more recent activity appear more engaged.
- Monetary: For the monetary component, recent spending is valued higher than historical spending, resulting in an adjusted view of customer value that reflects current financial engagement.

RFM Segmentation Clusters with Quantile

In this RFM segmentation approach, we use quantiles to assign a score to each customer for the Recency, Frequency, and Monetary components. Each RFM component is divided into four quantiles, with scores ranging from 1 (lowest engagement) to 4 (highest engagement). We then calculate a total RFM score by summing the individual scores for each component. This total score (ranging from 3 to 12) helps identify common clusters within the customer base, based on their engagement and spending patterns.

Cluster Scoring

Recency Score:
- Quantile Scoring:
  - 4: Top 25% of customers with the most recent purchases.
  - 3: Customers in the 50-75% range (moderately recent).
  - 2: Customers in the 25-50% range (less recent).
  - 1: Bottom 25% of customers with the longest time since their last purchase.
Frequency Score:
- Quantile Scoring:
  - 4: Top 25% of customers with the highest purchase frequency.
  - 3: Customers in the 50-75% range (moderate frequency).
  - 2: Customers in the 25-50% range (lower frequency).
  - 1: Bottom 25% of customers with the lowest purchase frequency.
Monetary Score:
- Quantile Scoring:
  - 4: Top 25% of customers with the highest spending.
  - 3: Customers in the 50-75% range (moderate spending).
  - 2: Customers in the 25-50% range (lower spending).
  - 1: Bottom 25% of customers with the lowest spending.

Cluster Interpretation

Champion (RFM Score of 12)
- RFM Score Combinations: 4-4-4 (High Recency, High Frequency, High Monetary)
- Description: Champions are the top-tier customers with high scores in all three RFM dimensions. They are recent, frequent, and high-spending customers who are highly engaged and contribute significantly to Shopee’s revenue.
Loyal Customer (RFM Score of 10-11)
- RFM Score Combinations: 4-3-3, 3-4-3, 3-3-4 (High in two dimensions, moderate in one)
- Description: Loyal Customers are consistent buyers who are active and have decent spending, but one dimension (either frequency, monetary, or recency) is slightly lower than the other two.
Promising (RFM Score of 8-9)
- RFM Score Combinations: 4-2-2, 2-4-2, 2-2-4, 3-3-2, 3-2-3, 2-3-3 (High in one dimension, moderate in two)
- Description: Promising customers have a strong score in one RFM dimension but moderate scores in the other two, indicating engagement potential. These customers need only a little effort to move toward becoming Loyal Customers.
Need Attention (RFM Score of 6-7)
- RFM Score Combinations: 4-1-1, 1-4-1, 1-1-4, 3-2-2, 2-3-2, 2-2-3 (High or moderate in one dimension, low in others)
- Description: Customers in the Need Attention segment show activity in only one RFM dimension, or are moderately active across two but lack in at least one area. This group requires more effort than the Promising segment to become fully engaged.
Hibernating (RFM Score of 4-5)
- RFM Score Combinations: 3-1-1, 1-3-1, 1-1-3, 2-2-1, 2-1-2, 1-2-2 (Moderate to low in all dimensions, or only one moderate dimension)
- Description: Hibernating customers have low engagement, low spending, and are not recent purchasers. These customers have interacted with Shopee but are on the verge of becoming inactive or dormant.
Almost Lost (RFM Score of 3-4)
- RFM Score Combinations: 1-1-1, 2-1-1, 1-2-1 (Low in all dimensions or nearly all)
- Description: Almost Lost customers have low scores in all RFM dimensions, indicating minimal engagement and a high risk of churn. They may not respond to regular promotions and likely need substantial incentives

Interpreting clusters based on marketing effort needed to level up customers to Loyal cluster.

Segment Name	Common RFM Patterns	Marketing Effort Needed to Level Up to Loyal Customer
Champion	4-4-4	Low
Loyal Customer	4-3-3, 3-4-3, 3-3-4	Low to Moderate
Promising	4-2-2, 2-4-2, 2-2-4, etc.	Moderate
Need Attention	4-1-1, 1-4-1, 3-2-2, etc.	Moderate to High
Hibernating	3-1-1, 1-3-1, 2-2-1, etc.	High
Almost Lost	1-1-1, 2-1-1, 1-2-1	Very High

RFM Segmentation Clusters with Kmeans

K-means clustering is a popular technique for RFM segmentation as it allows for the grouping of customers based on similar distances in behavioral patterns. By calculating the distance between customers in the RFM space, K-means clustering identifies groups with similar purchase recency, frequency, and monetary value.

Mechanism

Standardize RFM Values:
- Before applying K-means, it’s essential to standardize the RFM data so that each dimension (Recency, Frequency, and Monetary) has equal influence on the clustering. This ensures that the clustering results are not skewed by differences in the scale of each component.
Define the Number of Clusters (K):
- Choosing the optimal number of clusters (K) is critical. We will use Elbow Method to determine the best K by analyzing how well-separated and cohesive each cluster is.
Calculate Distances:
- K-means calculates the Euclidean distance between each customer’s RFM values and the centroid of each cluster. Customers are then grouped based on the minimum distance to each cluster center, ensuring that each cluster has a high degree of similarity in RFM behavior.
Assign Cluster Labels and Interpret Results:
- After clustering, each customer is assigned to a cluster. These clusters are analyzed to understand the dominant characteristics of each group, such as high recency with low frequency or high monetary value, which can then be used to guide marketing strategies.

Cluster Naming and Analysis

Each cluster is named based on the dominant characteristics of the customers within it, allowing for targeted marketing approaches:

Cluster 3: Champions
- Characteristics:
  - Very Low Recency: Champions make frequent, recent purchases, indicating the highest level of engagement.
  - High Frequency and Very High Monetary Value: They purchase frequently and contribute significantly to revenue.
- Interpretation: Champions are Shopee's most valuable and loyal customers – brand advocates who frequently make high-value purchases.
Cluster 2: Loyal Customers
- Characteristics:
  - Low Recency: These customers have purchased recently, showing strong engagement.
  - High Frequency and High Monetary Value: They engage consistently and contribute significantly to revenue.
- Interpretation: Loyal Customers are highly valuable, frequently purchasing and showing brand loyalty.
Clusters 5 and 6: Promising
- Characteristics:
  - Low Recency: Customers in these clusters have engaged recently.
  - Moderate Frequency and Moderate to High Monetary Value: They purchase at a moderate rate and show promising spending patterns.
- Interpretation: Promising customers have potential to become more loyal if nurtured with the right incentives.
Cluster 1: At-Risk Customers
- Characteristics:
  - Moderate to High Recency: These customers have not purchased recently, but they were previously active.
  - Moderate Frequency and Monetary Value: They used to purchase more actively but have recently decreased their engagement.
- Interpretation: At-Risk Customers are slipping away. They were once engaged but are now less active, making them a priority for re-engagement efforts.
Cluster 4: Need Attention
- Characteristics:
  - Moderate Recency: These customers have purchased relatively recently, but not frequently.
  - Low Frequency and Low to Moderate Monetary Value: They purchase infrequently and spend a moderate amount.
- Interpretation: This group shows some engagement but lacks consistency in purchasing frequency or spending.
Cluster 0: Hibernating
- Characteristics:
  - High Recency: These customers have not engaged in a long time.
  - Low Frequency and Low Monetary Value: They purchase infrequently and have low total spending.
- Interpretation: Hibernating customers are essentially inactive, showing little to no recent engagement.

Comparison of RFM Segmentation with Quantiles vs. K-Means

1. Quantile-Based RFM Segmentation

Strengths:
- Simplicity: Quantile-based segmentation is straightforward and easy to understand. It divides customers into groups based on predefined percentile cut-offs, making it intuitive for stakeholders.
- Consistency: Since quantiles use static thresholds, the segmentation is stable and doesn’t change with small variations in data. This can be useful for quick, periodic updates (e.g., monthly) without recalculating clusters.
- Fast Processing: Quantile segmentation is computationally faster and better suited for large datasets that need frequent updates. This can be advantageous in real-time environments or when data is refreshed regularly.
Weaknesses:
- Less Structural Precision: The quantile method may create segments that don’t fully capture natural groupings in the data. This could lead to clusters with overlapping characteristics or less-defined boundaries.
- Limited Adaptability: Quantile segmentation assumes an equal number of customers in each segment, which may not accurately reflect behavioral patterns. It may overlook specific, smaller customer groups that are highly valuable.

2. K-Means Clustering for RFM Segmentation

Strengths:
- Data-Driven Grouping: K-means clustering uses distance-based calculations, creating clusters that represent true groupings in the data. This results in more structured, visually distinct clusters (as seen in the 3D plot).
- Flexibility: K-means can adapt to the underlying structure of customer behavior, allowing for clusters of different sizes. This flexibility allows Shopee to identify niche segments that quantile-based segmentation may overlook.
- Scalability: K-means can capture more granular customer segments, especially as data grows. For complex customer behavior, K-means provides a refined clustering that enhances segmentation accuracy.
Weaknesses:
- Complexity: K-means requires additional steps, including standardization of data and selection of the optimal number of clusters (K). This adds complexity to the analysis and may be less intuitive for stakeholders.
- Computational Demand: K-means can be computationally intensive, particularly for large datasets with frequent updates. It may be less suitable for real-time updates and may require periodic re-clustering to keep segments relevant.
- Sensitivity to Outliers: K-means can be affected by outliers, as these can distort the cluster centroids and affect the overall clustering result. Additional preprocessing may be required to handle outliers.

Summary Table

Method	Strengths	Weaknesses
Quantile-Based	Simple, consistent, fast for large datasets; easy to interpret and update frequently.	Less structural precision, limited adaptability, may not capture natural groupings in customer data.
K-Means Clustering	Data-driven clusters, flexible structure, scalable with granularity in segmentation.	Complex, computationally intensive, sensitive to outliers; may require periodic re-clustering.

Ultimately, the choice between these methods depends on Shopee’s priorities and what stakeholders need from customer segmentation:

If stakeholders prioritize fast, routine updates and simplicity, quantile-based segmentation may be more appealing.
If stakeholders are looking for precise and data-driven insights that can guide long-term strategies, K-means clustering may be the preferred choice.

This code assigns an activity status to each customer based on their most recent purchase date in relation to a specified cutoff date. By categorizing customers as "Active" or "Inactive," Shopee can better understand customer engagement levels, which is critical for targeted re-engagement strategies.

Activity Criteria: Customers are considered "Active" if they have made a purchase within the last 60 days relative to the cutoff date. If their most recent purchase was more than 60 days before the cutoff date, or if they have no purchase history, they are categorized as "Inactive."
Rationale for 60-Day Cutoff:
- Simplicity and Prompt Response: Setting a 60-day cutoff is straightforward, making it easy to identify and act on customers who are at risk of disengagement. This period allows Shopee to respond promptly with targeted re-engagement campaigns before customers drift further toward inactivity.
- Alignment with E-commerce Engagement Patterns: In many e-commerce settings, engaged customers typically make purchases every few weeks to a couple of months. A 60-day window captures customers who are still within a typical engagement cycle, ensuring that those marked as "Inactive" are genuinely in need of re-engagement.
- Optimal Marketing Balance: A 60-day period provides enough time for natural re-engagement without requiring overly frequent interventions. This balance allows marketing efforts to focus on customers who may need an incentive to return, optimizing the use of resources for maximum impact.
- Broad Applicability Across Categories: This cutoff is effective across diverse product categories. Whether customers are purchasing frequently bought items (e.g., groceries) or less frequent, higher-value items (e.g., electronics), the 60-day threshold is broadly applicable, making the "Inactive" status relevant for different types of customers.
Cutoff Date: The cutoff date serves as a reference point for determining recency. In this example, it is set to December 31, 2019, but it can be adjusted based on business needs, such as the end of a promotional period or fiscal quarter.
Use Case: The "Activity" status allows Shopee to effectively segment customers based on engagement, guiding targeted re-engagement campaigns for "Inactive" customers and proactive engagement strategies for "Active" customers.
- Active Customers are ideal targets for regular engagement and upsell strategies, reinforcing their relationship with Shopee.
- Inactive Customers can benefit from tailored re-engagement efforts, such as win-back offers, reminders, or special promotions. By focusing on reactivating these customers, Shopee can reduce churn and maintain customer loyalty.

4. Discount Sensitivity

Discount sensitivity is an important behavioral metric that reveals how likely customers are to make purchases when discounts are offered. By examining Average Discount Percentage (the typical discount percentage a customer receives on their purchases) and Percentage of Items Purchased at a Discount (the proportion of total purchases made with a discount), we can understand which customers are most responsive to promotions.

Using K-means clustering, customers are grouped into distinct Discount Sensitivity Clusters based on their discount behavior patterns. Here’s an analysis of each cluster:

Cluster Analysis and Naming

Cluster 0: Bargain Hunters
- Characteristics:
  - High Percentage of Items Purchased at Discount (70-100%): These customers primarily buy discounted items, indicating a high level of price sensitivity.
  - High Average Discount Percentage: They typically purchase items with substantial discounts, preferring significant price reductions.
- Interpretation: Bargain Hunters are highly discount-sensitive and wait for sales before purchasing.
- Actionable Insights:
  - Targeted Discounts: Use deep discount campaigns to attract them, such as seasonal sales or flash deals.
  - Personalized Promotions: Offer exclusive discount codes or promotions to keep them engaged.
Cluster 3: Deal Seekers
- Characteristics:
  - Moderate to High Percentage of Items Purchased at Discount (50-70%): These customers buy many discounted items but are not as dependent on discounts as Bargain Hunters.
  - Moderate Average Discount Percentage: They typically purchase items with modest discounts.
- Interpretation: Deal Seekers enjoy discounts but may still buy at regular prices if the value proposition is strong.
- Actionable Insights:
  - Loyalty Program Discounts: Encourage loyalty by offering modest, consistent discounts rather than only deep discounts.
  - Bundle Discounts: Offer small discounts for bundle purchases, encouraging more frequent purchases.
Cluster 1: Occasional Discounters
- Characteristics:
  - Low to Moderate Percentage of Items Purchased at Discount (20-50%): These customers occasionally take advantage of discounts but do not rely on them.
  - Low to Moderate Average Discount Percentage: They receive discounts occasionally but do not prioritize them in their purchasing decisions.
- Interpretation: Occasional Discounters are less price-sensitive and may respond more to product quality or brand value rather than discounts.
- Actionable Insights:
  - Value-Based Messaging: Highlight product quality, exclusivity, or other non-price-related value propositions to attract these customers.
  - Limited Discounts on Premium Items: Occasionally offer discounts on premium items to entice them without conditioning them to expect regular discounts.
Cluster 2: Full-Price Shoppers
- Characteristics:
  - Very Low Percentage of Items Purchased at Discount (<20%): These customers rarely buy discounted items, showing minimal discount sensitivity.
  - Very Low Average Discount Percentage: They pay close to full price on most purchases, indicating a focus on non-price factors.
- Interpretation: Full-Price Shoppers value the product or brand more than discounts and may prioritize quality, convenience, or brand loyalty.
- Actionable Insights:
  - Exclusive and Premium Offerings: Focus on exclusive products or premium services that align with their value orientation.
  - VIP Programs: Offer membership-based benefits rather than discount-based incentives to reinforce loyalty without discount dependency.

5. Preference on Timing and Category

A. Preferred Day of the Week

Description: Identifies the day of the week on which a customer most frequently makes purchases.
Purpose: Enables Shopee to schedule targeted promotions and campaigns on days when customers are more likely to shop, increasing the likelihood of engagement and conversions.

B. Preferred Day of the Week

Description: Identifies the hours of the day on which a customer most frequently makes purchases.
Purpose: Enables Shopee to schedule targeted promotions and campaigns on hours when customers are more likely to shop, increasing the likelihood of engagement and conversions.

C. Preferred Category

Description: Highlights the product category that a customer purchases most frequently.
Purpose: Facilitates personalized marketing and product recommendations by focusing on the customer’s primary area of interest, enhancing relevance and increasing customer satisfaction.

6. Campaign Participation

Campaign Participation measures the extent to which each customer engages in Shopee’s promotional campaigns, distinguishing between High Participation and Low Participation. This metric is based on the customer’s frequency of purchases during campaign periods in relation to the total number of campaigns available within a given timeframe.

Purpose: Identifying campaign participation helps Shopee to:

Optimize Targeting: Understand which customers are responsive to campaign-driven promotions and can be targeted for future campaigns.
Allocate Marketing Resources: Allocate marketing budgets effectively by focusing more resources on customers with a high likelihood of participating in campaigns.

Participation Score:

High Participation: If the ratio of campaign participation (campaign orders relative to the total campaign count) is above 40%, the customer is categorized as having "High Participation."
Low Participation: If the ratio is below or equal to 40%, the customer is categorized as having "Low Participation."

How to Use the Clusters

This guide outlines several actionable methods for Shopee’s marketing team to leverage customer clustering for targeted and effective campaigns. By selecting and filtering clusters, stakeholders can create tailored strategies that align with specific objectives and engagement levels. The flexibility of this clustering system allows stakeholders to dynamically adjust criteria, ensuring that campaigns remain relevant for both short-term promotional events and long-term engagement goals. Below are some potential scenarios where this clustering system could be effectively applied:

1. Preparing for High-Impact Campaigns (e.g., Double Date Sales)

Goal: Maximize reach and engagement in the lead-up to major sales events.
Suggested Cluster Filter:
- Active Customers with Low Campaign Participation: Identify active users who have historically shown low engagement in campaigns. This group can be targeted with intensive marketing to build familiarity with the upcoming campaign.
- Channels: Use a mix of in-app notifications, targeted ads on other platforms, and email reminders to repeatedly engage segment in the week before the sale.
- Personalized Incentives: Offer exclusive “early bird” discounts or reminders of special deals to encourage early participation.

2. Focusing on VIP and Loyal Customers for Exclusive Deals

Goal: Reward high-value customers with early access to premium deals, fostering loyalty and brand advocacy.
Suggested Cluster Filter:
- VIP and Loyal Customers (e.g., Platinum or Gold tiers) who prefer discounts (as identified by high Discount Sensitivity).
- Campaign Design: Provide early access to limited-time deals, premium items, or exclusive bundles. These customers appreciate being first to know, so give them a VIP experience with special privileges.
- Channels: Direct communication, such as SMS or email, to emphasize exclusivity and personal appreciation. Push notifications with personalized offers can also increase engagement.

3. Re-engaging Inactive Customers

Goal: Win back customers who haven’t purchased recently.
Suggested Cluster Filter:
- Inactive Customers with Moderate to High Campaign Participation: These customers have engaged in campaigns previously but have recently become inactive. Use re-engagement offers to rekindle interest.
- Campaign Design: Offer “We Miss You” deals, personalized recommendations based on past purchases, and win-back incentives like one-time discount codes.
- Channels: Email or SMS campaigns are effective for lapsed users, with follow-up reminders if engagement is not immediate.

4. Engaging Discount-Sensitive Customers with Flash Sales

Goal: Drive immediate purchases by appealing to price-conscious customers.
Suggested Cluster Filter:
- Bargain Hunters or Deal Seekers within Active or Recently Active Customers segments. These customers respond well to discounts and may be more likely to participate in time-sensitive promotions.
- Campaign Design: Flash sales, limited-time discounts, or bundle deals that create urgency.
- Channels: Real-time push notifications, in-app banners, and website highlights to capture attention immediately.

5. Promoting New Product Categories to Specific Category Shopper

Goal: Encourage purchases and introduce new products to engaged customers.
Suggested Cluster Filter:
- Shoppers with related preferred category who are Active: These customers frequently order from this category and are likely to be open to new product types.
- Campaign Design: Cross-category promotions, “complete your collection” bundles, or “you might also like” recommendations.
- Channels: Personalized in-app suggestions, email recommendations, and dynamic homepage banners showcasing new or trending products.

6. Testing Loyalty Program Impact on Frequent Buyers with Low Monetary Value

Goal: Encourage frequent, small-spending customers to increase their average order value.
Suggested Cluster Filter:
- Frequent Buyers with Low Monetary Value and High Discount Sensitivity: This group may benefit from loyalty incentives that encourage higher-value purchases.
- Campaign Design: Offer loyalty points or tier-based rewards for spending milestones, encouraging them to increase spending.
- Channels: Regular updates on loyalty status through in-app messages, personalized rewards reminders, and email notifications highlighting the benefits of moving up the loyalty tiers.

7. Leveraging Campaign Participation to Refine Strategy

Goal: Identify engagement trends to improve future campaigns.
Suggested Cluster Filter:
- Analyze Campaign Participation Rates Across Segments: Use data on campaign participation to refine messaging and timing for future campaigns.
- Action Steps:
  - High participation clusters can be encouraged to share their experiences, potentially acting as advocates.
  - Low participation clusters can be re-targeted with more attractive offers or reminders, especially during major shopping events.
- Insights: Campaign participation data helps in understanding which segments respond best to certain types of campaigns, allowing for more effective resource allocation in the future.

This clustering system allows Shopee’s marketing team to filter and target customer segments in a flexible, data-driven way. By selecting the right combinations of activity status, loyalty tiers, discount sensitivity, preferred timing, and campaign participation, stakeholders can optimize outreach for each unique campaign, maximizing both customer engagement and overall sales impact.

Combining Customer-Facing and Company-Facing Segmentation for Impact

By implementing both customer-facing and company-facing segmentation, Shopee is hoped to be benefited from a holistic approach that aligns customer experience with operational efficiency:

Customer-Facing: The loyalty tiers (Classic, Silver, Gold, Platinum) provide a clear framework for customers to engage competitively, work towards higher ranks, and enjoy exclusive benefits based on their spending rank. This tiered approach strengthens brand loyalty and encourages increased spending.
Company-Facing: Behind the scenes, more granular segmentation using RFM analysis, discount sensitivity, product variety, and other behavioral factors allows Shopee to target each customer group effectively. This segmentation provides a foundation for personalized marketing, optimized promotions, and improved customer service.

Stakeholders Involved

To implement this segmentation strategy successfully, cross-functional collaboration is crucial:

Marketing Team:
- Develop and implement customer-facing loyalty tier benefits and messaging.
- Design targeted campaigns based on internal segmentation insights (discount sensitivity, product variety, preferred shopping windows).
Data Analytics Team:
- Conduct RFM and behavioral segmentation analysis, updating segment definitions based on changing customer behavior.
- Monitor the effectiveness of segmentation-driven campaigns and loyalty tiers, adjusting the criteria as necessary.
CRM and Customer Engagement Team:
- Manage loyalty communications for different tiers, sending personalized messages and re-engagement campaigns.
- Track and analyze customer responses to loyalty incentives and adjust communications accordingly.
Product and UX/UI Teams:
- Ensure that the customer-facing experience reflects the loyalty tiers and segmented experiences.
- Implement features for seamless access to loyalty benefits and personalized recommendations for each customer segment.
Finance Team:
- Review and approve budget allocations for customer incentives and campaign investments, ensuring they align with projected ROI from targeted segments.
- Analyze the financial impact of segmentation-based discounts and loyalty benefits on overall profitability.

This two-way segmentation approach is hoped to create a customer-centric experience through loyalty tiers while gaining actionable insights for internal decision-making through detailed behavioral segmentation. By clearly defining loyalty tiers for customers and using RFM analysis, discount sensitivity, product variety, and other dimensions internally, Shopee is expected to improve customer satisfaction, engagement, and retention.

This segmentation strategy aligns well with Shopee’s business goals, allowing the platform to:

Engage customers with transparent loyalty benefits that reward frequent engagement.
Personalize experiences and offers based on precise behavioral insights.
Optimize resources by focusing efforts on high-potential segments and tailoring approaches for lower-engagement groups.

This combined approach is hoped to make customer interactions are meaningful, campaigns are effective, and business outcomes are maximized.

B. Analysis of Customer Retention and Lifetime Value

This phase of the project seeks to empower an e-commerce business with the insights needed to enhance customer retention and optimize Customer Lifetime Value (CLV), driving long-term profitability and competitive advantage. Given the fast-paced and highly competitive nature of e-commerce, understanding customer behavior, identifying at-risk customers, and deploying strategies to engage these customers effectively over time are crucial to sustained growth and maximizing revenue.

To achieve these goals, the project will focus on analyzing customer churn rates and retention patterns across diverse customer segments. Specifically, we aim to identify periods and segments with heightened risk of customer disengagement and to evaluate the impact of existing retention strategies on customer loyalty and long-term value. By segmenting customer data by demographics, engagement levels, and campaign types, we will provide a holistic view of customer retention and CLV, allowing the business to tailor retention initiatives and build a data-driven approach for customer loyalty.

The technical implementation details, including code for each step of the analysis, can be found in the Analysis of Customer Retention and Lifetime Value Notebook in the GitHub repository.

Here are the steps we take in the analysis:

1. Calculating Monthly Customer Churn Rates

Analysis of Monthly Customer Churn Rates

How to Calculate Monthly Customer Churn Rates?

The code calculates the monthly churn rate by following a structured approach that defines customer activity and measures the loss of monthly retention. Here’s a breakdown of the logic used:

Identify Each Customer's First Purchase Month:
- The code begins by determining the first purchase month for each customer. This is essential as it defines the starting point (or cohort month) for tracking a customer's future activity.
- The first_purchase_month for each customer is identified by grouping the dataset by customer_id and finding the minimum order date.
Define Customer Activity Status by Month:
- To calculate the churn rate, it's important to determine if a customer is "Active" in each month. A customer is considered active in a given month if they made at least one purchase within a certain period defined by the cutoff days.
- The is_active_in_month function checks whether each customer has made a purchase within a backtracked cutoff period (60 days) leading up to the end of the month. This accounts for recent purchases that indicate engagement.
- If a customer made a purchase in this cutoff period, they are marked as "Active"; otherwise, they are labeled as "Not Yet Customer" if they have no prior purchases.
Calculate Monthly Churn Rate:
- The code calculates churn by iterating through each month, tracking customers who were active in the previous month but did not make any purchases in the current month.
- The churn rate for each month is calculated with the formula:
  
  Churn Rate = (Lost Customers / Active Customers in Previous Month) * 100
  
  where "Lost Customers" represents the customers who were active in the previous month but did not remain active in the current month.
Verification for Months with 0% Churn:
- The code includes specific checks for months with a 0% churn rate. In these cases, it prints a confirmation message with the list of active customers in the previous and current months.
- For February and March, the 0% churn rate is due to the fact that all previously active customers remained active in the following month, showing complete retention.
Visualization:
- Monthly churn rates are stored in a dictionary and visualized as a bar chart, where each bar represents a month’s churn rate.

General Observations

The churn rates across months reveal patterns of customer retention and disengagement, providing insights into periods of strong retention and higher customer loss.

Stable Retention in Early Months:
- From February to June 2019, churn rates are relatively low, fluctuating between 0% and 6%. This suggests stable retention and effective initial engagement strategies that keep customers active in the early months.
High Churn in August and September:
- August and September show a significant spike in churn rates, reaching 10.04% in August and 27.88% in September. This indicates a period of increased customer disengagement, which could be due to seasonal factors, a drop in promotions, or unmet customer expectations.
Improved Retention in October and November:
- Churn rates drop sharply in October (2.41%) and remain stable in November (5.12%), indicating successful re-engagement efforts or an increase in customer engagement activities that have managed to retain customers effectively.
Churn Increase in December:
- In December, churn rises to 8.66%, possibly due to the holiday season, where spending behaviors and engagement patterns shift. This end-of-year churn increase highlights a seasonal challenge that may require specific strategies to retain customers.

Key Insights

Effective Early Retention:
- Low churn rates in the first half of the year (February-June) demonstrate effective engagement, indicating that the company’s initial retention strategies are performing well in maintaining customer activity.
Churn Spikes in Late Summer/Early Fall:
- The sharp increase in churn during August and September signals a potential issue with mid-year engagement. This period likely requires targeted strategies to sustain customer interest after the initial acquisition period.
Re-Engagement Success in Fall:
- The reduction in churn in October and November suggests effective re-engagement tactics, which could include timely promotions or improved retention efforts that draw previously inactive customers back.
Seasonal Churn Challenges in December:
- December’s increased churn rate indicates a potential impact from seasonal factors, underscoring the need for holiday-specific strategies to counteract seasonal disengagement.

2. Comprehensive Analysis of Customer Retention Rates

2.1 Analyzing Monthly Customer Retention Rates

Analysis of Monthly Customer Retention Rates

The heatmap visualization displays the monthly retention rates for different customer cohorts, defined by the month of their first purchase. Each row represents a unique cohort based on the first purchase month (from January 2019 to December 2019), and each column represents subsequent months in which these customers made repeat purchases. The retention rates, expressed as percentages, indicate the proportion of customers from each cohort who returned in subsequent months.

Why Analyzing First Purchase Cohorts is Important

Analyzing first purchase cohorts provides valuable insights into customer retention trends over time. By grouping customers based on the month of their first purchase, we can track how effectively the retention strategy encourages repeat engagement in subsequent months. Cohort analysis helps identify patterns in customer behavior, uncover issues with long-term loyalty, and evaluate the effectiveness of retention efforts. Additionally, this approach highlights whether specific months or marketing campaigns are more successful at attracting and retaining customers, allowing for targeted improvements in retention strategies.

General Observations

Initial Cohort Engagement: The first month after the initial purchase typically shows a strong retention rate only in specific cohorts, such as January 2019 (72%), July 2019 (85%), and August 2019 (100%). However, even in these cohorts, the retention rate tends to decrease over time, indicating diminishing engagement as the months progress.
Cohort Retention Decay: As time progresses, there is a clear decay in retention rates across cohorts. For example, the January 2019 cohort starts with a 72% retention rate in February 2019, which gradually declines to 44% by December 2019.
Low Retention for Later Cohorts: Later cohorts (e.g., from September 2019 onward) display lower retention rates. For instance, the retention rate for the September 2019 cohort drops to 0% in October and remains at 0% in both November and December 2019.

Key Insights

Early Cohorts Show Strong Initial Retention: Cohorts such as January, February, and March 2019 maintain relatively higher retention rates over a few months. For instance, the March 2019 cohort maintains a reasonable retention rate of 51% even three months after the initial purchase. Similarly, the February 2019 cohort maintains a reasonable rate of 55% even four months after the initial purchase.
Retention Drops Rapidly Over Time: Across all cohorts, there is a noticeable decline in retention rates within a few months. This suggests that while customers initially engage with repeat purchases, maintaining engagement becomes challenging over longer periods.
Retention Stability for August 2019 Cohort: The August 2019 cohort demonstrates unique retention patterns, maintaining an exceptionally high retention rate of 100% in both the first and second months following their initial purchase. By year-end, the cohort still sustains a retention rate of 33-44%. These trends suggest that specific factors associated with August 2019 may have positively influenced customer retention during these months, possibly due to targeted marketing efforts, seasonal effects, or unique product offerings. Further analysis could uncover the key drivers of this enhanced retention performance.
Challenges in Retaining Later Cohorts: The retention rates for cohorts starting from September 2019 are particularly low, with most rates falling to 0% within a few months. This drop could imply seasonal trends, diminishing customer interest, or issues with the retention strategy.

Summary of Effectiveness of Current Retention Strategy

The current retention strategy appears partially effective but lacks long-term engagement capabilities. While initial retention rates for some early cohorts are relatively strong, there is a significant decline within a few months, with some cohorts showing retention rates as low as 0% just a few months after their initial purchase. This trend indicates that the retention strategy may successfully attract customers initially but struggles to maintain their interest over time.

Effective Elements: The strategy is effective in engaging customers initially, as seen with high retention rates within the first few months (e.g., 72% -73% for January 2019). This suggests that initial promotions, onboarding, or first-purchase incentives are working well.
Ineffective Elements: The rapid decline in retention across most cohorts implies that the strategy is not fostering long-term loyalty or repeat purchases beyond a few months. Additionally, the very low retention rates for later cohorts indicate that the strategy may not be adapting effectively to changing customer needs or seasonal demands.

2.2 Analyzing Customer Retention Rates by Age

Analysis of Retention Rates Based on Age

The heatmaps represent retention rates segmented by age group, illustrating the likelihood of repeat purchases for each cohort over time. The x-axis represents the months after the first purchase, while the y-axis shows the initial purchase month for each cohort. The color intensity highlights the retention rate, with darker colors representing higher retention. This visualization reveals age-specific trends, providing insights into each age group's engagement patterns.

General Observations

Under 18 Age Group:
- Retention Patterns: There is a strong initial retention rates for each cohort in the first month. This indicates that new customers in this age group are initially engaged and likely to make repeat purchases in the month following their first purchase. However, there is a notable decline in retention rates as months progress. For example, for the March 2019 cohort, retention drops from 100% in the first month after their initial purchase to 50% in the second month and continues to decline rapidly, reaching 0% by September 2019 and remaining at that level through the end of the year. Additionally, retention varies significantly across cohorts. Some cohorts, like January 2019 and February 2019, maintain relatively higher retention rates, while others, such as March 2019 and May 2019, drop quickly to 0%, indicating a lack of long-term engagement in certain groups.
- Long-Term Retention: There is a general trend of diminishing retention rates over time. For several cohorts, retention falls to 0% after a few months, especially for those starting later in the year, like March 2019 and May 2019. This suggests that customers in this age group struggle with long-term engagement and are less likely to make repeat purchases after a few months. Additionally, the cohorts from the beginning of the year (e.g., January 2019) exhibit slightly better long-term retention than later cohorts, though this still declines significantly over time. This could indicate seasonal purchasing behaviors or marketing campaigns that were more effective earlier in the year.
- Interpretation: Since the initial retention of this age group is high but declines sharply, the business might benefit from implementing strategies to maintain engagement beyond the first few months. Loyalty programs, personalized offers, or reminder notifications could encourage repeat purchases. Besides, considering the varying retention rates across cohorts, it would be useful to analyze the effectiveness of any campaigns or engagement strategies employed at different times of the year. Identifying successful practices from high-retention cohorts like January 2019 could provide insights to replicate in future campaigns.
18-24 Age Group:
- Retention Patterns: Retention for this age group is relatively strong in the initial months, particularly for cohorts that made their first purchase early in the year. The January 2019 cohort begins with 73% retention after in the first month after initial purchase, maintaining around 70% for the next few months, showing a steady but gradual decline. Similar patterns are observed in the February and March cohorts, though they decline slightly faster.
- Long-Term Retention: By June to October 2019, retention rates continue to decrease gradually, with most cohorts stabilizing at around 30-45%. For example, the March cohort reaches 32% by October 2019, indicating a consistent drop-off over time. Retention rates decline even further toward the end of the year, with most cohorts dropping to around 0-20% in November and December.
- Interpretation: This age group likely needs a mix of engaging content and value-driven incentives to sustain interest beyond the first few months, as engagement drops off without substantial long-term reinforcement.
25-34 Age Group:
- Retention Patterns: The January 2019 cohort shows high retention in the initial months, with retention rates starting at 77% and remaining around 75-79% through March, April, and May. This reflects strong initial engagement. Other cohorts show a gradual decrease in retention over time, but still maintain moderate rates for a few months after the first purchase. For instance, the May 2019 cohort retains around 39-44% in the following months of June and July.
- Long-Term Retention: Retention rates tend to decrease gradually over time. For instance, the January cohort stabilizes around 57% by July and August, while later cohorts have lower long-term retention, often between 30-40%. This indicates that customers in this age group may initially engage well but gradually reduce repeat purchases over time.
- Interpretation: Customers in this age group appear to find value in the offerings, suggesting that a tailored long-term retention strategy could leverage their initial interest to foster loyalty and repeat engagement.
35-44 Age Group:
- Retention Patterns: Across multiple cohorts, initial retention remains high in the initial month of each cohort's purchase. However, by the second and third months, retention rates start declining noticeably. For example, The February 2019 cohort drops to 55% in March and further declines to 49% in April, while the March 2019 cohort retains 47% in April before continuing to decrease. This trend shows a consistent pattern of high engagement initially, followed by a gradual but steady decline over time.
- Long-Term Retention: Retention rates remain relatively low in the long term, with some months showing 0% retention in later stages. The occasional increase (such as the October 2019 cohort) suggests possible seasonal or campaign-driven engagement.
- Interpretation: This age group demonstrates an initial willingness to re-engage but experiences a significant drop-off after the first few months. Loyalty-building efforts might be effective if applied soon after the initial purchase to maintain engagement, as long-term retention tends to be lower without ongoing initiatives.
45-54 Age Group:
- Retention Patterns: This age group displays a strong initial retention rate, with retention beginning to drop off around the three after the first initial purchase. However, the decline is less steep, with several cohorts retaining around 35-50% by later months.
- Long-Term Retention: Long-term retention rates taper off, with most cohorts, like June 2019, dropping to around 17% by October 2019. However, a few cohorts, such as July 2019 and August 2019, continue to maintain higher retention rates, ranging from 60% to 100% in the later months. The August 2019 cohorts show notably high retention (100%) through several months, suggesting specific factors—such as campaigns or seasonal trends—that may have influenced their engagement.
- Interpretation: The retention pattern indicates that the this age group has moderate loyalty, with consistent engagement in the early months. However, retention gradually declines over time. This demographic may respond well to targeted loyalty initiatives, especially as engagement levels start to stabilize. Offering exclusive content, loyalty rewards, or periodic re-engagement campaigns could help boost retention in the long term.
55-64 Age Group:
- Retention Patterns: Initial retention is high but experiences a steady decline over the following months. For example, the January 2019 cohort retains around 70% by the June 2019 and continues to decrease gradually. Other cohorts, like March 2019 and April 2019, also show a significant drop in retention within the first few months, indicating a steady loss of engagement after the initial period.
- Long-Term Retention: Long-term retention stabilizes at lower rates (typically around 10-20%) for early cohorts, indicating less sustained engagement over time compared to younger groups. Some cohorts, such as May 2019, show retention rates of 8% in later months, while others, like August 2019, briefly maintain high retention (100%) before dropping off completely to 0%.
- Interpretation: While this age group has lower long-term engagement, they demonstrate moderate loyalty in the mid-term. Retention strategies that emphasize personalized experiences or value-based rewards might encourage longer-lasting engagement, especially for cohorts showing initial responsiveness to engagement efforts.
65+ Age Group:
- Retention Patterns: This group shows high initial retention, particularly for the January cohort, but quickly decreases in subsequent months.
- Long-Term Retention: Retention declines significantly over time for most cohorts, with only a small portion of customers remaining engaged in the long term, often stabilizing at low levels or dropping to 0% for some cohorts.
- Interpretation: The rapid disengagement indicates that current strategies may not resonate well with this age group. This suggests that a more tailored retention strategy, focusing on personalized communication or value-driven engagement, could help retain their interest and encourage repeat purchases among this demographic.

Key Insights

Effective Initial Retention Across All Age Groups: All age groups display strong initial retention rates, indicating effective initial engagement strategies, likely through targeted advertising or onboarding processes that successfully attract new customers.
Age-Specific Retention Decline Patterns: Younger age groups (Under 18 and 18-24) experience rapid declines in retention, suggesting that the current retention strategy lacks elements to sustain their interest. The middle age groups (25-54) have better retention, suggesting potential for building longer-term loyalty within these age brackets.
Middle-Aged Cohorts Show Higher Long-Term Engagement: The 25-54 age groups exhibit slightly better retention, suggesting a higher likelihood of long-term engagement. These cohorts might benefit from further loyalty-driven strategies to enhance their retention potential.
Older Demographics Show Declining Engagement: The 55-64 and 65+ age groups display lower long-term retention, indicating that these demographics might require different approaches for effective engagement.

Summary of Effectiveness of Current Retention Strategy

Effective Elements:
- The strategy is effective in engaging customers initially, as evidenced by high first-month retention rates. For example, the January 2019 cohort shows retention rates around 60-70% across all age groups in the first month after their initial purchase, suggesting that initial promotions, onboarding processes, or first-purchase incentives are effective in capturing customer interest.
- The middle age groups (25-54) exhibit relatively stable retention over time compared to other groups, indicating that the current strategy has some appeal to this demographic, potentially due to loyalty programs or value-driven offerings.
Ineffective Elements:
- There is a rapid decline in retention across most cohorts after the first few months, especially among younger (Under 18 and 18-24) and older age groups (55+). This suggests that the strategy lacks elements that foster long-term loyalty and repeat purchases.
- The retention strategy does not seem to adapt effectively to changing customer needs or preferences across age groups, particularly for the youngest and oldest demographics, which experience the most significant drop-offs. This indicates an opportunity to diversify engagement tactics based on age-specific interests and preferences.

2.3 Analyzing Customer Retention Rates by Gender

Analysis of Retention Rates by Gender

The heatmaps illustrate retention rates of female and male customers across monthly cohorts, showing how likely customers from each initial purchase month (cohort) are to make repeat purchases in subsequent months. Darker colors represent higher retention rates, with lighter colors indicating lower retention or lack of engagement.

General Observations

Female Customers:
- The retention rate starts strong in the first few months for each cohort, but it drops consistently over time.
- For example, the January 2019 cohort shows a 71% retention rate in the first month after their initial purchase, then fluctuating between 50-73% until reaching 43% by the end of 2019.
- Later cohorts (e.g., May and June) see retention drop more quickly, with a rapid decline to around 0-15% retention by the end of the year (November and December 2019).
Male Customers:
- Similar to females, male customers also show high retention in the first few months but display slightly more stability in certain cohorts (e.g., the August 2019 cohort maintains a higher retention rate through subsequent months).
- However, like females, retention rates drop over time, with many cohorts reducing to about 0-15% by the end of the year (November and December 2019).

Key Insights

Initial Retention Success: Both genders show a high initial retention rate, suggesting effective engagement strategies at the beginning of the customer journey.
Rapid Decline Over Time: Retention rates drop significantly over time for both genders, indicating that the current strategy lacks long-term engagement.
Slight Variability Across Genders: Male customers appear to have marginally higher retention in some later months, though the overall trend of decline is present for both genders.

Summary of Effectiveness of Current Retention Strategy

The current retention strategy appears partially effective but requires enhancements for long-term engagement. While it successfully attracts customers initially, there is a noticeable drop-off in retention rates within a few months, especially among female customers, indicating that the strategy may be limited in maintaining customer interest over time.

Effective Elements:
- The strategy is effective in driving initial engagement, as seen in the high first-month retention rates across both genders. For example, several female and male cohorts demonstrate retention rates of around 50-100% in the initial months after their first purchase, indicating that early marketing efforts, promotions, or onboarding processes are effectively capturing customer interest.
- Male customers exhibit slightly more stable retention rates over time compared to female customers, indicating that some aspects of the strategy may resonate better with male customers, particularly during the first few months post-purchase.
Ineffective Elements:
- Retention rates decline significantly after the first few months for both genders, with female customers experiencing a steeper drop-off in retention rates. This suggests that the current retention strategy does not adequately sustain engagement beyond the initial purchase, leading to a rapid loss of interest.
- The strategy lacks a gender-specific approach, which may be contributing to the high drop-off rates. There is no evident customization to address the unique preferences and engagement drivers for male and female customers, indicating a gap in tailored long-term engagement efforts.
- The very low retention rates (as low as 10-20%) within a year imply that the strategy does not effectively foster loyalty or encourage repeat purchases over time. This indicates an opportunity to incorporate personalized and gender-specific engagement tactics to better meet customers’ evolving needs.

Overall, while the current strategy is successful at capturing initial interest, it does not provide sufficient follow-through to retain customers over the long term, particularly for female customers. Addressing these gaps with targeted engagement strategies for each gender could help strengthen retention and build customer loyalty.

2.4: Analyzing Customer Retention Rates by Engagement Level (High vs Low Engagement)

Analysis of Retention Rates Based on Engagement Level

The heatmaps display retention rates segmented by customer engagement level, differentiating between high engagement and low engagement customers. The x-axis represents the months following the initial purchase, while the y-axis shows the cohort's first purchase month. Color intensity indicates retention rates, with darker shades representing higher retention.

General Observations

High Engagement Customers:
- Retention Patterns: High engagement customers exhibit strong initial retention, with the January 2019 cohort showing a retention rate of 82% in the first month, followed by consistently high rates (e.g., 82-84%) over the next few months.
- Long-Term Retention: Retention gradually decreases over time but stabilizes at a relatively higher rate for high-engagement customers compared to low-engagement ones. For example, most cohorts retain around 40-50% of customers about four months after their initial purchase. However, by November 2019, retention levels for most cohorts drop further, generally stabilizing between 20-30%. Interestingly, retention rates rises again in December 2019, with some cohorts reaching levels of 50-70% and even up to 100%.
- Interpretation: High engagement customers maintain more stable retention, indicating that these customers are more likely to remain loyal and continue engaging with the brand over a longer period.
Low Engagement Customers:
- Retention Patterns: Retention among low engagement customers drops significantly within the first few months. The March 2019 cohort starts with a 46% retention rate in the first month, but it drops to around 34% within the next two months.
- Long-Term Retention: The retention rate for low engagement customers decreases sharply, stabilizing at low levels. Most cohorts retain around 30% about four months after their initial purchase, and by the end of the year, retention rates fall to 0-10%.
- Interpretation: Low engagement customers show minimal long-term retention, suggesting that these customers are not as invested in the brand and may require additional incentives or engagement to retain their interest.

Key Insights

High Initial Retention for Both Engagement Levels: Both high and low engagement customers show strong initial retention, suggesting that the initial engagement strategy effectively attracts customers, regardless of their engagement level.
Significantly Better Retention Among High Engagement Customers: High engagement customers retain at a much higher rate over time than low engagement customers, indicating that engagement level is a strong predictor of long-term retention. This highlights the importance of fostering high engagement early in the customer journey.
Rapid Decline in Retention for Low Engagement Customers: Low engagement customers show a quick drop in retention, with most cohorts retaining less than 20% of customers by the middle of the year. This indicates that low engagement customers are more likely to churn, highlighting a potential area for improvement in retention strategies for this segment.

Summary of Effectiveness of Current Retention Strategy

The current retention strategy appears partially effective in maintaining engagement for high engagement customers but struggles to retain low engagement customers over the long term. While the initial strategy effectively attracts both high and low engagement customers, there is a significant disparity in long-term retention rates between these two groups.

Effective Elements:
- The strategy is effective in securing initial purchases across both high and low engagement customers, as evidenced by high first-month retention rates for both segments. This suggests that initial promotions, product offerings, or onboarding strategies are successful in attracting customers initially.
- High engagement customers exhibit sustained retention over time, indicating that the strategy is effective in building loyalty within this segment. Retention rates for high engagement customers remain relatively high even after several months, suggesting that aspects of the strategy resonate well with this group.
Ineffective Elements:
- Retention rates for low engagement customers decline sharply within the first few months, stabilizing at very low levels by the end of the year. This indicates that the strategy does not adequately address the needs or interests of low engagement customers, leading to high churn rates in this segment.
- The lack of differentiation in retention strategies for high and low engagement customers limits the effectiveness of the strategy, particularly for the low engagement group. A one-size-fits-all approach may not be sufficient to maintain long-term interest and loyalty across varying engagement levels.

Overall, while the current retention strategy successfully engages high engagement customers over the long term, it lacks the flexibility and personalization needed to retain low engagement customers effectively.

2.5 Analyzing Customer Retention Rates by Promotional Campaign Type

Analysis of Retention Rates Based on Promotional Campaign Type

The heatmaps show retention rates across different promotional campaign types, with high priority campaigns and additional mega sales campaigns on the top and high priority seasonal sales dates and other seasonal sales dates on the bottom. The x-axis represents the months following the initial purchase, while the y-axis indicates the cohort's first purchase month. Color intensity represents retention rates, with darker colors signifying higher retention.

General Observations

High Priority Mega Sales Date Campaign:
- Retention Patterns: High priority mega sales campaigns show strong initial retention rates, with many cohorts maintaining 60-100% retention in the early months. For example, the January 2019 cohort sustains 80-90% for up to nine months.
- Long-Term Retention: Although retention declines over time, it remains relatively high compared to other campaigns, with some cohorts retaining around 50% by the end of the year.
- Interpretation: This campaign type successfully attracts customers and maintains a strong retention rate over time, suggesting that high-priority mega sales dates are effective in driving long-term engagement.
Additional Mega Sales Date Campaign:
- Retention Patterns: Retention rates for additional mega-sale date campaigns are lower than those for high-priority campaigns, showing a steeper decline after the first few months. For example, the July 2019 cohort begins with a strong retention rate of 78% but quickly drops to 11% three months after their initial purchase.
- Long-Term Retention: Retention stabilizes at a lower rate, with most cohorts retaining approximately 30% or fewer customers by the end of the year. Notably, by April, retention declines further, reaching as low as 8% by year-end.
- Interpretation: While additional mega sales campaigns attract initial interest, they are less effective in retaining customers over the long term compared to high priority campaigns.
High Priority Seasonal Sales Date Campaign:
- Retention Patterns: High priority seasonal campaigns exhibit excellent retention, with retention rates often exceeding 90% in the early months. The January 2019 cohort, for instance, retains more than 95% for the first few months.
- Long-Term Retention: Retention remains strong, with certain cohorts retaining over 60% of customers in the month following their initial purchase, and some cohorts even sustaining 100% retention rates until November 2019. In December 2019, retention rates decline from November but stabilize around 50-80%, reflecting a high level of customer engagement and loyalty.
- Interpretation: High priority seasonal sales dates are highly effective for customer retention, making this campaign type particularly valuable for long-term engagement.
Other Seasonal Sales Date Campaign:
- Retention Patterns: Retention rates for other seasonal sales campaigns begin at moderate levels and decline relatively quickly compared to high-priority seasonal campaigns. For example, the January 2019 cohort starts with a retention rate of approximately 85%, which decreases to 72% six months after the initial purchase and further declines to 60% ten months after the initial purchase.
- Long-Term Retention: Retention stabilizes at low levels by the end of the year, with most cohorts retaining less than 30% of customers.
- Interpretation: While these campaigns attract initial interest, they do not retain customers as effectively as high-priority seasonal sales, suggesting they may benefit from additional engagement tactics.

Key Insights

Higher Retention with High Priority Campaigns: Both high priority mega sales and high priority seasonal sales campaigns demonstrate significantly higher retention rates over time compared to additional mega sales and other seasonal sales dates. This indicates that prioritizing certain campaigns can lead to stronger customer loyalty.
Retention Challenges with Additional Campaigns: Retention rates for additional mega sales and other seasonal sales dates decline rapidly, indicating that these campaigns may lack elements that encourage long-term engagement.
Importance of Campaign Type in Retention Strategy: The data suggests that campaign type plays a crucial role in retention. High priority campaigns (whether mega or seasonal) are more effective in retaining customers than additional or lower-priority campaigns.

Summary of Effectiveness of Current Retention Strategy

The current retention strategy appears effective in the short term for all campaign types but only sustains long-term retention for high priority campaigns. While initial retention rates are high across all campaign types, there is a significant disparity in long-term retention between high priority and additional campaigns.

Effective Elements:
- High priority campaigns (both mega sales and seasonal) show strong retention over time, suggesting that the strategy for these campaigns effectively promotes sustained customer engagement.
- The initial retention rates across all campaigns are high, indicating that promotional efforts are successful in attracting customers initially, regardless of campaign type.
Ineffective Elements:
- Retention rates for additional mega sales and other seasonal campaigns decline significantly within a few months, stabilizing at lower levels by the end of the year. This suggests that these campaign types may not be providing enough value or incentives to maintain long-term customer interest.
- The lack of differentiated retention tactics for high priority versus additional campaigns limits the overall effectiveness of the strategy. A tailored approach may be needed to enhance long-term retention across different campaign types.

Overall, the current retention strategy works well for high priority campaigns but requires additional tactics for sustaining engagement in additional and lower-priority campaigns.

3. Comprehensive Analysis of Customer Lifetime Value (CLV)

3.1 Analyzing Monthly Customer Lifetime Value (CLV)

Analysis of Monthly Customer Lifetime Value

The Customer Lifetime Value (CLV) metric is a powerful indicator of the total revenue a company can expect from a customer throughout their relationship. Understanding CLV on a month-to-month basis allows businesses to measure the effectiveness of their engagement and retention strategies over time. By examining the trends in average purchase value, purchase frequency, and customer lifespan, we can pinpoint how well current strategies are working to maintain customer engagement and foster long-term loyalty.

CLV Calculation

To calculate Customer Lifetime Value (CLV), we use three key components:

Average Purchase Value (APV): This is the mean revenue generated per purchase each month. It is calculated as:
- APV = Total Revenue / Total Purchases
- A high APV indicates that customers are making valuable purchases, which can be influenced by factors like product pricing and promotional discounts.
Average Purchase Frequency (APF): This metric captures the average number of purchases a customer makes within a month. It is calculated by taking the average of each customer's monthly purchases:
- APF = Average Number of Purchases per Customer in a Month
- APF is crucial as it indicates how frequently customers are engaging with the brand. High APF values suggest strong brand engagement, with customers returning often.
Average Customer Lifespan: This represents the average duration (in years) a customer remains actively engaged with the brand. It is calculated by finding the time between a customer’s first and last purchase and averaging across all customers:
- Average Lifespan (Years) = Days between First and Last Purchase / 365
- A longer lifespan indicates that customers are staying with the brand for an extended period, contributing to sustained revenue.

Using these components, Customer Lifetime Value (CLV) is calculated as:

CLV = APV * APF * Average Customer Lifespan (Years)

This formula provides a holistic view of the total value that each customer can bring to the company, considering their spending habits, frequency of purchases, and length of engagement.

Monthly Breakdown of Key Factors

Average Purchase Value by Month:
- Overview: The APV fluctuates throughout the year, with significant peaks in January, March, and July where values exceed $40. The average purchase value is highest in January at 46.20, which then drops in February to 37.41. March experiences a slight recovery with an average of 40.58. The values fluctuate throughout the year, with another noticeable peak in July (40.82), and a significant low in December at 31.65. This variation in APV could be linked to seasonal promotions, high-priority campaigns, or holiday periods that encourage higher spending per transaction.
- Seasonal Patterns: The high value in January suggests that customers may engage in post-holiday spending. The increase in July might reflect seasonal factors or mid-year promotional events. December’s lower average purchase value is interesting, as it is typically a high-sales month, indicating that purchases might be smaller in value, possibly due to discounts or holiday deals.
- Interpretation: Higher APV months suggest that campaigns targeting these periods are effective in encouraging larger purchases. The drop in APV during other months may signal a need to maintain interest or enhance the appeal of products and services outside peak seasons.
Average Purchase Frequency by Month:
- Overview: The APF varies, with November and December showing particularly high engagement, reaching up to 9.30 and 8.63, respectively. These values indicate that customers are making more purchases per month during these periods, likely due to seasonal or promotional events.
- Engagement Patterns: Early months, such as January through April, display moderate purchase frequencies, averaging around 5 purchases per customer. This indicates consistent but not intensive engagement during these months.
- Interpretation: The peaks in purchase frequency during November and December align with seasonal sales, which likely play a key role in boosting engagement. However, to create more balanced engagement, strategies to maintain frequency in non-peak months could be explored.
Average Customer Lifespan:
- Overview: The average customer lifespan remains short and stable at around 0.08 years (approximately one month). This brief lifespan implies that many customers are making purchases only within a limited period before disengaging.
- Interpretation: The short lifespan highlights a potential gap in retention efforts, as customers are not staying with the brand for extended periods. Improving the lifespan could significantly enhance overall CLV by encouraging repeat purchases over a longer timeframe.
Customer Lifetime Value (CLV) by Month:
- Overview: CLV fluctuates significantly across months, with notable highs in January, July, November, and December ($21.13, $21.35, $27.28 and $22.45, respectively), while other months, such as February and June, show lower values around $15.
- Seasonal Impact: CLV peaks during high-purchase months, indicating that promotions and targeted campaigns effectively increase customer value during these times.
- Interpretation: The high CLV values in certain months show that campaigns can drive short-term engagement and revenue, but the fluctuations suggest that sustained engagement is lacking. A more balanced approach could help maintain a steady CLV across all months.

Key Insights

Seasonal Campaigns Significantly Boost CLV: The highest CLV values are observed during peak promotional months (e.g., January and November), indicating that these campaigns effectively increase customer spending and engagement.
Short Customer Lifespan Limits Long-Term CLV Growth: Despite successful initial engagement, the short average lifespan suggests that customers disengage quickly. Enhancing retention strategies to prolong the lifespan could lead to more consistent revenue streams.
Inconsistent CLV Trends Across the Year: The sharp peaks and valleys in CLV reveal a reliance on seasonal promotions for customer engagement, suggesting that a more consistent strategy might yield steadier long-term results.

Summary of Effectiveness of Current Retention Strategy

The current retention strategy is effective in generating short-term engagement but falls short in sustaining long-term customer relationships. While the seasonal campaigns successfully boost purchase frequency and value, the limited customer lifespan indicates that these efforts are not fostering prolonged engagement.

Effective Elements:
- High Engagement During Campaigns: Seasonal sales and targeted promotions are effective in driving repeat purchases, especially in months like January, July, and November. These efforts increase CLV during specific periods.
- Initial Attraction: Promotional tactics successfully attract customers and encourage multiple purchases within the campaign period, indicating effective short-term engagement strategies.
Ineffective Elements:
- Lack of Sustained Engagement: The retention strategy lacks mechanisms to maintain customer interest post-campaign, resulting in a short average lifespan. This suggests that the strategy fails to foster loyalty beyond the initial purchase period.
- Inconsistent CLV: Monthly CLV fluctuations highlight that engagement is not consistent year-round. Customers are not motivated to stay active outside promotional events, which limits long-term revenue potential.

Overall, while the strategy generates strong engagement during targeted campaigns, it lacks the continuity needed for sustained customer relationships, leading to missed opportunities for steady revenue growth.

3.1.1 Calculating and Evaluating the Customer Acquisition Cost (CAC) in Relation to Customer Lifetime Value (CLV)

Analysis of Customer Acquisition Cost (CAC) and CLV-to-CAC Ratio

The goal of this analysis is to assess the efficiency of acquisition costs during promotional campaigns by comparing the Customer Acquisition Cost (CAC) with the Customer Lifetime Value (CLV).

Customer Acquisition Cost (CAC) represents the average cost incurred to acquire a new customer through campaign efforts. It is calculated by dividing the total campaign expenses by the number of new customers gained during these promotional periods. A lower CAC means it costs less to acquire each customer, which can be desirable if the acquired customers bring high long-term value to the business.

To evaluate if these acquisition costs are worthwhile, the CLV-to-CAC Ratio is used. This ratio compares the expected revenue a customer will bring throughout their relationship with the brand (CLV) against the cost of acquiring them (CAC). It provides insight into whether the company’s marketing spend is being used effectively.

Insights from the Analysis

New Customer Acquisition through Campaigns: The analysis identified the number of unique new customers acquired during the campaigns by filtering for customers whose first purchase occurred within the campaign period. This metric helps to understand how effective the campaigns are in attracting new customers.
Total Campaign Costs: By summing all campaign-related expenses, we gain a clear view of the investment made in customer acquisition efforts. This total cost is an essential factor for calculating CAC.
Calculation of CAC: Dividing the total campaign cost by the number of new customers acquired gives us the average CAC, representing the cost per customer. A high CAC indicates that acquisition is expensive relative to the volume of new customers gained, while a low CAC implies a more cost-effective acquisition strategy.
CLV-to-CAC Ratio: The CLV-to-CAC Ratio is a critical metric in evaluating acquisition efficiency. A high CLV-to-CAC ratio (≥3) typically suggests that acquisition costs are justified and that the customers acquired bring substantial long-term value. A moderate ratio (between 1 and 3) may indicate room for improvement in either CLV or CAC. A low ratio (<1), as observed in this analysis, implies that the cost to acquire customers may be too high relative to the revenue they generate.
Interpretation of Findings: In this case, the low CLV-to-CAC ratio (<1) suggests that current acquisition costs may not be sustainable, as the lifetime value generated by these customers does not justify the high acquisition costs. This result highlights a potential inefficiency in campaign spending, indicating that the brand is spending more on acquiring customers than they are likely to bring in revenue.

Conclusion:

Since the CLV-to-CAC Ratio is low, it is essential to dig deeper into the Customer Lifetime Value (CLV) by breaking it down across various customer segments. Analyzing CLV by attributes such as age, gender, campaign type, and engagement level allows us to pinpoint specific segments that may be driving profitability or, conversely, adding to acquisition inefficiency. This segmented analysis will highlight which customer groups are bringing the most value to the brand and which may need more focused retention strategies to enhance their lifetime value.

By identifying high-CLV segments, the company can tailor its acquisition and retention strategies to attract and retain these profitable groups, ultimately improving the overall CLV-to-CAC ratio. For low-CLV segments, this analysis will help the company make informed decisions on whether to adjust acquisition spending or implement targeted initiatives to increase engagement and value.

Therefore, in the next step, we will conduct a detailed CLV analysis segmented by age, gender, campaign type, and engagement level. This will enable a more precise approach to customer acquisition and retention, ensuring that resources are allocated to strategies that maximize profitability and sustain long-term customer relationships.

3.2 Analyzing Customer Lifetime Value (CLV) by Age

Analysis of Customer Lifetime Value (CLV) Based on Age Group

Analyzing Customer Lifetime Value (CLV) based on age group provides insights into which age segments bring the highest value to the company and which may require different engagement strategies. By examining metrics like average purchase value, purchase frequency, customer lifespan, and CLV for each age group, we can identify high-value segments and opportunities for improvement in customer engagement and retention.

General Observations

Average Purchase Value (APV) by Age Group:
- The 65+ age group has the highest average purchase value at $47.85, which suggests they are inclined to make fewer but higher-value purchases.
- The Under 18 age group has the lowest APV at $18.33, possibly due to lower spending power or different purchasing preferences.
- Interpretation: High APV in older age groups indicates a potential to enhance engagement among these customers with premium or high-value offerings.
Average Purchase Frequency (APF) by Age Group:
- The 25-34 age group shows the highest purchase frequency at 15.79 purchases per year, indicating a highly engaged segment.
- In contrast, the Under 18 and 65+ age groups exhibit very low purchase frequencies (0.15 and 0.20, respectively).
- Interpretation: Younger customers may have limited spending frequency, while older customers might benefit from incentives to increase engagement frequency.
Average Customer Lifespan by Age Group:
- The 25-34 age group also has the longest customer lifespan at 0.65 years, while the 65+ group has the shortest at 0.54 years.
- Interpretation: The 25-34 age group not only engages frequently but also maintains a longer relationship with the brand, making them a crucial demographic for CLV.
Customer Lifetime Value (CLV) by Age Group:
- The 25-34 age group has the highest CLV at $375.94, followed by the 35-44 age group at $272.61.
- The Under 18 and 65+ age groups contribute the lowest CLV values at $1.58 and $5.26, respectively.
- Interpretation: The highest CLV in the 25-34 age group indicates this demographic as a prime target for retention strategies, while the lower CLV in older and younger age groups suggests potential areas to enhance value through tailored engagement.

Key Insights

High Value in the 25-34 Age Group: The 25-34 age group shows strong engagement, high purchase frequency, and the longest customer lifespan, leading to the highest CLV. This segment is highly valuable and should be prioritized for sustained engagement.
Opportunities to Improve CLV for Younger and Older Age Groups: The Under 18 and 65+ age groups exhibit low engagement levels, short lifespans, and minimal CLV contributions. Targeted strategies are needed to drive engagement and increase their overall value.
Potential to Increase APF in the 65+ Age Group: While the 65+ group has a high average purchase value, their low purchase frequency limits their CLV. Introducing incentives to increase their purchase frequency could enhance their overall lifetime value.

Summary of Effectiveness of Current Retention Strategy

The current retention strategy appears to be highly effective for the 25-34 age group but less effective for younger and older age segments. While the brand successfully engages the 25-34 demographic with frequent purchases and longer relationships, it struggles to maintain similar engagement levels among the Under 18 and 65+ age groups.

Effective Aspects:
- The strategy effectively engages the 25-34 demographic, yielding the highest CLV due to strong purchase frequency and customer lifespan.
- The 35-44 age group also shows relatively high CLV, benefiting from similar engagement and retention efforts.
Ineffective Aspects:
- The retention approach does not sufficiently engage the Under 18 and 65+ age groups, leading to low CLV in these segments.
- Limited focus on incentivizing higher purchase frequency in older customers or enhancing the spending power of younger customers contributes to lower value in these demographics.

Overall, while the strategy demonstrates success in key demographics, addressing gaps in engagement with younger and older customers could enhance overall brand value.

3.3 Analyzing Customer Lifetime Value (CLV) by Gender

Analysis of Customer Lifetime Value Based on Gender

The analysis of Customer Lifetime Value (CLV) based on gender provides insights into spending behavior, purchase frequency, engagement duration, and overall value contribution of male and female customers. CLV is derived from three key components: average purchase value, average purchase frequency, and average customer lifespan. These components reflect how much, how often, and for how long customers engage with the brand.

General Observations

Average Purchase Value by Gender:
- Observation: Female customers have a slightly higher average purchase value of 37.40$ compared to male customers, who have an average purchase value of 36.55$.
- Interpretation: Female customers tend to spend marginally more per transaction. This could be influenced by product preferences, with certain items potentially being more popular among female customers.
Average Purchase Frequency by Gender:
- Observation: Female customers exhibit a higher average purchase frequency (42.50) than male customers (37.33).
- Interpretation: This suggests that female customers are more engaged with the brand and make purchases more frequently. This pattern could indicate effective targeting of female customers through marketing or product offerings that resonate more with this demographic.
Average Customer Lifespan by Gender:
- Observation: The average customer lifespan is similar across genders, with female customers showing a slightly longer engagement period (0.62 years) than male customers (0.61 years).
- Interpretation: The minimal difference in customer lifespan suggests that both male and female customers have similar retention rates, with gender not being a strong differentiator in terms of engagement duration.
Customer Lifetime Value (CLV) by Gender:
- Observation: Female customers have a higher CLV, amounting to 988.10$, compared to male customers, whose CLV is 837.75$. This difference is primarily driven by higher purchase frequency and average purchase value among female customers.
- Interpretation: Female customers are generating more lifetime value due to higher engagement, suggesting that strategies targeting female customers could yield better returns on investment.

Key Insights

Female Customers Contribute Higher CLV: Female customers show both a higher purchase frequency and a slightly higher purchase value, resulting in a significantly greater CLV compared to male customers.
Similar Retention Across Genders: Both male and female customers have a similar lifespan, indicating consistent engagement levels across genders.
Greater Engagement Among Female Customers: Female customers' higher purchase frequency suggests a stronger relationship with the brand, possibly due to tailored marketing campaigns or product offerings that appeal more to female customers.

Summary of Effectiveness of Current Retention Strategy

The current retention strategy appears more effective with female customers, as reflected by their higher purchase frequency and CLV. While both genders have similar retention durations, female customers show greater engagement and spending behavior, indicating a successful alignment of the brand's offerings with female preferences. However, the strategy may be less effective in maximizing engagement among male customers, as they demonstrate lower purchase frequency and CLV.

Effective Aspects:
- The strategy successfully drives higher engagement and lifetime value among female customers.
- Retention rates are comparable across genders, indicating a balanced approach to customer retention.
Areas for Improvement:
- Lower CLV and purchase frequency among male customers suggest potential untapped opportunities.
- The current strategy might benefit from a more targeted approach to boost male customer engagement.

3.4 Customer Lifetime Value (CLV) by Engagement Level (High vs Low Engagement)

Analysis of Customer Lifetime Value (CLV) Based on Engagement Level (High vs Low Engagement)

This analysis evaluates Customer Lifetime Value (CLV) across different engagement levels—High Engagement and Low Engagement. Engagement level is a significant indicator of how frequently and consistently customers interact with the brand, impacting their purchase frequency, average purchase value, and overall lifespan with the brand.

General Observations

Average Purchase Value by Engagement Level:
- Overview: Both high and low engagement groups show similar average purchase values, with only a slight difference ($36.85 for high engagement vs. $37.94 for low engagement).
- Interpretation: This similarity suggests that both groups spend comparably per transaction. Therefore, the difference in overall CLV is likely influenced more by frequency and lifespan rather than individual purchase value.
Average Purchase Frequency by Engagement Level:
- Overview: Customers with high engagement levels have an average purchase frequency of 72.37, while those with low engagement average only 9.19 purchases.
- Interpretation: The stark contrast in purchase frequency indicates that high-engagement customers are significantly more active, which directly enhances their CLV. This suggests that fostering frequent interactions is key to improving lifetime value.
Average Customer Lifespan by Engagement Level:
- Overview: High-engagement customers have an average lifespan of 0.78 years, compared to 0.47 years for low-engagement customers.
- Interpretation: High engagement not only encourages frequent purchases but also extends the customer's relationship with the brand. This longer lifespan further contributes to higher CLV.
Customer Lifetime Value (CLV) by Engagement Level:
- Overview: The CLV for high-engagement customers is $2075.01, drastically higher than the $162.74 for low-engagement customers.
- Interpretation: High engagement yields a substantial return in CLV, underscoring the importance of engagement-focused retention strategies. The substantial difference in CLV indicates that customer engagement level is a critical driver of long-term profitability.

Key Insights

Engagement Level Directly Impacts CLV: Customers with high engagement demonstrate a significantly higher CLV due to increased purchase frequency and longer customer lifespan.
Retention through Frequent Engagement: The data shows that boosting engagement frequency can lead to substantial improvements in CLV, as high-engagement customers are not only more active but also maintain a longer relationship with the brand.
Sustained Engagement Yields Greater Profitability: The long lifespan and high purchase frequency of engaged customers underscore the financial benefits of investing in strategies that encourage continuous interaction.

Summary of Effectiveness of Current Retention Strategy

The current retention strategy appears effective in driving engagement among high-value customers but may fall short in fostering engagement across the broader customer base. The high CLV for engaged customers demonstrates the value of engagement-focused strategies; however, the low CLV among less engaged customers suggests that more can be done to uplift this group.

Effective Aspects:
- The strategy effectively maximizes value among already engaged customers, as seen by the high CLV, purchase frequency, and lifespan for this group.
- High-engagement campaigns and touchpoints seem to be successful for customers who are already interacting frequently with the brand.
Ineffective Aspects:
- The retention strategy lacks elements to convert low-engagement customers into high-engagement ones. This leads to a wide gap in CLV between the two groups.
- Limited efforts to boost engagement frequency and lifespan in the low-engagement group result in a significant disparity in lifetime value.

Overall, while the strategy excels with engaged customers, it requires expansion to encompass and uplift the low-engagement segment to achieve more balanced CLV growth across all customer groups.

3.5 Customer Lifetime Value (CLV) by Promotional Campaign Type

Analysis of Customer Lifetime Value (CLV) Based on Promotional Campaign Type

This analysis assesses Customer Lifetime Value (CLV) across four distinct promotional campaign types: High Priority Mega Sales Date, Additional Mega Sales Date, High Priority Seasonal Sales Date, and Other Seasonal Sales Date. Each campaign type’s effectiveness is evaluated based on average purchase value, purchase frequency, customer lifespan, and resulting CLV.

General Observations

Average Purchase Value by Campaign Type:
- High Priority Mega Sales Date: Average purchase value is $30.13, indicating customers tend to make moderately valued purchases during these campaigns.
- Additional Mega Sales Date: Similar to High Priority Mega Sales, with an average purchase value of $29.63.
- High Priority Seasonal Sales Date: Highest purchase value among all campaign types at $45.36, suggesting this campaign effectively encourages high-value purchases.
- Other Seasonal Sales Date: Lowest purchase value at $27.34, implying these campaigns attract customers who make smaller purchases on average.
- Interpretation: High Priority Seasonal campaigns appear successful at motivating customers to spend more per transaction, whereas Other Seasonal Sales attract lower-value purchases.
Average Purchase Frequency by Campaign Type:
- High Priority Mega Sales Date: Average purchase frequency is 4.10, indicating moderate customer engagement.
- Additional Mega Sales Date: Slightly lower frequency at 3.93, showing consistent customer engagement but less frequent than Other Seasonal Sales.
- High Priority Seasonal Sales Date: Lowest purchase frequency at 2.49, highlighting limited repeat purchases within this campaign.
- Other Seasonal Sales Date: Highest purchase frequency at 4.78, suggesting customers are more inclined to return during recurring seasonal events.
- Interpretation: Other Seasonal Sales are effective in generating repeated interactions, likely due to the familiarity or recurring nature of these campaigns.
Average Customer Lifespan by Campaign Type:
- High Priority Mega Sales Date: Customer lifespan is relatively short at 0.08 years, indicating limited post-campaign engagement.
- Additional Mega Sales Date: Marginally longer lifespan at 0.13 years, but still reflects short-term engagement.
- High Priority Seasonal Sales Date: Minimal lifespan close to 0 years, signifying that this campaign attracts one-time purchases.
- Other Seasonal Sales Date: Longest lifespan at 0.27 years, showing that these campaigns successfully foster longer-lasting customer relationships.
- Interpretation: Other Seasonal Sales appear more successful in establishing ongoing customer relationships, unlike high-priority, one-time events, which attract short-term buyers.
Customer Lifetime Value (CLV) by Campaign Type:
- High Priority Mega Sales Date: CLV is $10.26, indicating moderate value contribution with limited repeat purchases.
- Additional Mega Sales Date: CLV increases to $14.82, reflecting slightly better customer retention and value than High Priority Mega Sales.
- High Priority Seasonal Sales Date: Lowest CLV at $0.31, showing minimal long-term impact despite high individual transaction value.
- Other Seasonal Sales Date: Highest CLV at $35.85, demonstrating strong returns due to high purchase frequency and longer customer lifespan.
- Interpretation: Other Seasonal Sales generate the highest CLV, indicating that consistent, recurring events are more effective at sustaining profitable customer relationships than high-intensity, single-day sales.

Key Insights

Recurring Campaigns Drive Higher CLV: Other Seasonal Sales Date campaigns outperform in terms of CLV, purchase frequency, and customer lifespan, highlighting the effectiveness of recurring seasonal promotions.
One-Time High-Priority Sales Lack Longevity: High Priority campaigns (both Mega and Seasonal) tend to attract high initial purchases but fail to sustain long-term engagement, leading to lower CLV.
Frequency and Relationship Duration are Critical: The CLV data underscores that sustained customer engagement and relationship duration contribute more significantly to profitability than one-off, high-value transactions.

Explanation for Near-Zero CLV Value for the High Priority Seasonal Sales Date Campaign

The analysis of the "High Priority Seasonal Sales Date" campaign confirms that it was a single-day event, taking place only on November 29, 2019. The following code was used to verify this information.

# Check the date range for "High Priority Seasonal Sales Date"
campaign_date_range = sales_data[sales_data['campaign_name'] == 'High Priority Seasonal Sales Date']['order_time']
print("Campaign start date:", campaign_date_range.min())
print("Campaign end date:", campaign_date_range.max())

# Distribution of customer lifespans in this campaign
hp_seasonal_lifespans = customer_lifespans[customer_lifespans['campaign_name'] == 'High Priority Seasonal Sales Date']
print("Customer lifespan distribution:\n", hp_seasonal_lifespans['lifespan'].describe())

Below is a breakdown of the key findings:

Single-Day Campaign: The campaign ran from 00:15 to 23:56, indicating that all customer purchases occurred within a single 24-hour period.
Customer Lifespan: Due to the single-day duration, every customer has a lifespan of 1 day (with a minimum value set to 1 day for customers who made purchases on the same day). The distribution statistics for customer lifespan—mean, min, max, 25th percentile, and so on—all show a value of 1.0, reinforcing that all purchases were made exclusively within this day.

These findings highlight that the "High Priority Seasonal Sales Date" campaign was likely a flash sale or a special promotional event that attracted purchases within a constrained timeframe.

Rationale for Setting a Minimum Lifespan of 1 Day

In calculating Customer Lifetime Value (CLV) for this analysis, we set a minimum lifespan of 1 day for customers in single-day campaigns for the following reasons:

Avoiding Zero-Day Lifespans: Without this adjustment, customers with same-day purchases would have a calculated lifespan of zero days. This zero value could distort further calculations, particularly in metrics like customer lifetime value (CLV), where lifespan is a factor. A zero-day lifespan would reduce CLV calculations to zero, inaccurately reflecting the customer’s engagement with the campaign.
Accurately Representing Customer Engagement: Even though purchases occurred within a single day, a minimum lifespan of 1 day more accurately reflects the customer’s active participation in the campaign. This approach captures the fact that each customer engaged with the campaign, even if only briefly, and provides a consistent way to account for customer engagement in short-duration campaigns.
Ensuring Consistency in Lifespan Metrics Across Campaigns: Applying a minimum threshold of 1 day for single-day campaigns maintains consistency across different campaigns when calculating metrics like average customer lifespan. Without this adjustment, single-day campaigns would contribute zero values, skewing comparisons with longer campaigns and potentially obscuring insights in cross-campaign analyses.

By implementing a minimum lifespan threshold, the analysis remains robust, accurately capturing customer engagement in short campaigns and maintaining consistency in lifespan metrics across different campaign types.

Summary of Effectiveness of Current Retention Strategy

The current retention strategy appears effective for seasonal, recurring campaigns (especially Other Seasonal Sales Date), as these events foster frequent purchases and longer customer relationships, resulting in higher CLV. However, the strategy is less effective for high-priority, single-day campaigns, which drive high-value transactions but fail to retain customers long-term.

Effective Aspects:
- Recurring seasonal campaigns are successful in promoting repeat purchases and longer customer lifespans, which directly contribute to higher CLV.
- Other Seasonal Sales Date campaigns achieve the best balance of purchase frequency and lifespan, resulting in the highest CLV.
Ineffective Aspects:
- High-priority campaigns, such as High Priority Mega Sales Date and High Priority Seasonal Sales Date, generate high transaction values but lack retention mechanisms to convert one-time buyers into long-term customers.
- The retention strategy does not capitalize on post-campaign engagement opportunities, particularly for high-value customers acquired during high-priority events.

Overall, the strategy is highly effective for seasonal campaigns but requires improvements to increase customer retention from high-priority, one-time campaigns to enhance long-term value.

Overall Summary of Current Retention Strategy Effectiveness

The analysis of customer retention and lifetime value provides a comprehensive view of the strengths and limitations in the current retention strategy. While the strategy effectively drives initial customer engagement through promotional campaigns and generates short-term value, it falls short in sustaining long-term loyalty and maximizing customer lifetime value (CLV). Below is a detailed summary of its effectiveness:

Effective Aspects:

High Initial Engagement: Promotional campaigns, particularly high-priority sales events, have successfully drawn in new customers and reactivated existing ones. This is evidenced by strong first-month retention rates following campaign peaks, suggesting that the strategy is adept at generating immediate, short-term engagement. High initial customer engagement indicates that the company has effective promotional mechanics in place, attracting significant traffic and interest around key sales events.
Short-Term Retention Success: Across various demographic groups (age, gender), promotional campaign types, and engagement levels, the current retention strategy successfully maintains customer interest in the short term. This initial success demonstrates the effectiveness of targeted campaigns in capturing immediate engagement and driving return visits shortly after purchase.
High CLV During Campaign Peaks: CLV analysis across campaign types highlights substantial revenue during seasonal and mega sales events. The elevated CLV during these periods underscores the importance of well-timed campaigns in driving revenue and customer value. These spikes in CLV illustrate the capability of the current strategy to capitalize on consumer enthusiasm around key events, generating significant short-term profits.

Areas Lacking Effectiveness:

Decline in Long-Term Retention: Retention rates show a marked decline after the initial engagement period of a few months. This reveals a challenge in transitioning customers from one-time or infrequent purchasers to regular buyers. The gap in long-term retention points to a need for more consistent, post-campaign engagement strategies to sustain customer interest, ensuring that initial engagement momentum is maintained.
Over-Reliance on Campaign Peaks for CLV: The analysis reveals a heavy dependence on sales periods for driving CLV, with noticeable drops in value during non-promotional months. This dependence suggests a lack of consistent revenue streams from regular engagement, highlighting a need to diversify strategies to drive spending in off-peak periods and create more stable revenue.
High Drop-Off Among Low Engagement Customers: Low-engagement customers experience rapid declines in retention rates, suggesting that the current strategy does not adequately reach or motivate this segment. The quick disengagement among this group indicates that targeted engagement tactics for lower-frequency customers could significantly raise their CLV and increase their chances of becoming regular buyers.

In conclusion, the current retention strategy demonstrates clear strengths in driving initial engagement and generating substantial short-term value, particularly during high-impact promotional campaigns. However, its effectiveness wanes in sustaining customer loyalty and maximizing CLV over the long term. This analysis reveals specific gaps, such as the marked decline in long-term retention, over-reliance on campaign-driven peaks for revenue, and limited engagement with low-frequency customers. To build on the initial success and address these limitations, the strategy requires more targeted, consistent post-campaign efforts, as well as personalized engagement tactics for underperforming customer segments. Implementing these adjustments can enhance retention, foster customer loyalty, and ensure a more stable, long-term revenue stream.

C. Analysis of The Most Effective Marketing Channels and Campaigns

This part of the project presents an analysis of the most effective marketing channels and campaigns for Shopee, the top e-commerce platform in Singapore. The primary goal of this analysis is to identify which marketing strategies yield the highest return on investment (ROI), conversion rates, and which campaigns obtain the highest average order values (AOV). By examining various campaign types, including Mega Sales, Flash Sales, Seasonal Sales, Next Day Delivery, Livestream Exclusive, and Bundle Promotions, this analysis provides valuable insights on customer tastes and preferences in an online shopping landscape.

The technical implementation details, including code for each step of the analysis, can be found in the Analysis of Marketing Channel and Campaigns Notebook in the GitHub repository.

1. Key Metrics of Analysis:

Determine which marketing channels, such as KOL (Key Opinion Leaders) marketing, social media, email, SMS, in-app, and website marketing, yield the highest return on investment (ROI), conversion rates and click-through rates.

Why ROI?

ROI provides a quantitative measure of how effectively marketing campaigns convert investments into profits. By assessing ROI, Shopee can identify which marketing channels are driving revenue and which are not performing as expected.
Understanding the ROI of different marketing channels allows Shopee to allocate resources more effectively. By investing more in high-ROI channels and reducing spend on less effective ones, the company can maximize its marketing budget's impact. ROI analysis provides clear evidence of which marketing strategies are working. This helps stakeholders make informed choices about future marketing investments and strategies.
Evaluating ROI not only measures financial outcomes but also provides insights into customer behavior and preferences. Understanding what drives higher ROI enables stakeholders to improve future marketing strategies and utilize different marketing strategies on different target groups.

Why Consider Conversion Rate?

The conversion rate indicates how well a marketing channel turns customer exposure into actual sales. By analyzing conversion rates, Shopee can identify which channels are most effective in driving purchases, helping to refine marketing strategies and optimize efforts.
Analyzing the conversion rate enables Shopee to allocate resources more effectively. Focusing on high-performing channels allows for more efficient spending and maximizes return on investment (ROI). Conversion rates directly correlate with sales and revenue. By focusing on channels that lead to higher conversion rates, Shopee can increase total sales and order volumes, ultimately boosting overall revenue.

Why Click-Through Rate (CTR)?

CTR reflects how engaging and appealing an advertisement or marketing message is to the audience. A higher CTR indicates that more users are interested in the content, suggesting that the messaging, visuals, and overall campaign strategy are resonating with potential customers.
It provides insight into which campaigns are successfully driving traffic and attracting interest, helping to identify best practices and areas for improvement.
CTR can indicate which types of content, offers, or promotions are more attractive to customers. By examining CTR trends, Shopee can gain valuable insights into consumer preferences, enabling more targeted and relevant marketing efforts.

Why Average Order Value (AOV) of campaigns

AOV provides insights into how much customers are willing to spend per transaction. By analyzing AOV across various campaigns, Shopee can better understand customer spending habits and preferences, which may be useful in improving their pricing and promotional strategies.
AOV analysis can help Shopee adjust its marketing strategies to target customers more effectively. For example, if a particular campaign yields a higher AOV, similar tactics can be replicated or modified for future campaigns.
Campaigns with higher AOVs may warrant more investment in terms of marketing spend. By identifying which campaigns yield higher AOV, Shopee can allocate its budget more strategically to maximize returns.
AOV serves as a key performance indicator that allows Shopee to benchmark the success of different marketing initiatives. This comparative analysis helps in understanding what works best in driving higher-value purchases.

2. Analysis of marketing channels and campaign types

2.1 Key insights from analysis of marketing channel ROI

2.1.1 Cost efficiency

In terms of Cost per Sale (CPS) and Cost per Visitor (CPV)

Social Media is the most cost efficient channel
SMS is the least cost efficient channel

Cost per Visitor vs. Total Visitors (With bubble size indicating revenue generated from each channel) Cost per Visitor vs. Total Visitors (With bubble size indicating revenue generated from each channel)

The above visualization helps identify the efficiency of each marketing channel in driving traffic (visitors) at a minimal CPV. The lower to CPV, the more cost effective the channel is in driving traffic.

Social Media is highly efficient with the lowest CPV and highest visitor volume, making it highly cost-effective in generating traffic. The large bubble size also indicates the high revenue that Social Media generates.
In-App has a moderate CPV with high number of visitors. The large bubble size shows that In-App channel is still able to generate high revenue.
KOL, Website and Email shows a middle ground with moderate CPV and visitor counts. But KOL is the channel that drives the most revenue despite that.
SMS has the lowest efficiency with the highest CPV but lowest visitor volume, making it the least cost-effective channel for visitor generation. This makes SMS marketing less ideal in generating large scale traffic unless it is used in highly targeted campaigns where the high CPV can be rationalized.

Cost per Sale vs. Total Sales (With bubble size indicated revenue generated from channel)

The above visualization helps identify the efficiency of each marketing channel in driving sales at a minimal CPS.

Social Media has the highest efficiency with the lowest CPS and high sales volume, making it the most cost-effective in generating conversions. Coupled with the high revenue generated, Social Media would yield a high ROI.
In-App and KOL have a moderate CPS with good sales volume, with a balanced cost-to-sale ratio, making them relatively effective for conversions.
Conversions are more expensive through Email and Website, with total revenue lacking behind of Social Media, In-App and KOL.
SMS has the lowest efficiency with high CPS and low total sales, indicating that conversions via SMS are costly. SMS also contributes to the least total revenue.

In terms of revenue generated, the marketing budget spend on KOL, In-App and Social Media channels are more justified, given their larger contributions to revenue and overall lower CPV and CPS.

Social Media is the most efficient in both metrics, and should definitely be considered for scaling efforts in its marketing for optimized returns.

SMS is the least efficient in both metrics, hence scaling down on its marketing budget or using it for more niche audiences or specific high-impact campaigns can be considered.

Optimization Opportunities: Email and Website could become more effective by targeting optimizations in either CPV (for traffic generation) or CPS (for conversions), potentially through focused campaign types like retargeting or loyalty programs.

For further analysis, we analyse Cost per Sale as a percentage of AOV.

KOL, Social Media and In-App have relatively lower Cost-Per-Sale as a percentage of AOV, indicating that these channels are more effective in converting visitors to sales and generating more revenue without overspending, which in turn will lead to a higher ROI.

However, the percentage range for those channels are around 25% to 42%, which is still relatively high, indicating that cost-effectiveness might be limited. Nonetheless, we can see that those channels brings the highest total sales, indicating that the ROI of KOL, Social Media and In-App channels are among the highest.

SMS's percentage is way above 100%, which indicates that it is extremely cost-ineffective in driving orders of high value, in turn likely to lead to low ROI. Furthermore, SMS drives the lowest total sales, which in combination leads to poor ROI.

2.1.2 Conversion rate, Bounce rate and Click-through rate

As an additional analysis, we can look at the conversion rates, bounce rates and click-through rates of the various channels in determining their respective engagement.

Average Conversion Rate by Channel

KOL, In-App and Social Media has the highest Average Conversion Rates of 6.8% followed by 4%. These channels are most effective in turning visitors into sales. Cost per visitor of those channels are also relatively low.

Average Click-Through Rate (CTR) by Channel

KOL has the highest CTR (3%), followed by Email and Social Media (5%). This indicates that these channels are most effective in engaging consumers with the content shared through the channel.
Website (0.75%) and In-App (1%) have relatively lower CTR. Indicating that the content shared through these channels are less successful in attracting the attention of Shopee's potential customers.

Average Bounce Rate by Channel

Email (35%) has the lowest bounce rate, followed by KOL (40%) and SMS (45%), indicating that visitors who arrive via these channel are more likely to engage with the content and navigate further, contributing to a higher conversion rate. From this we can infer that Email is generally a more targeted form of marketing that shows customers ads that suit their interest.
High bounce rates for Website (60%) and In-App and Social Media (around 55%) suggest that visitors may not find the landing pages relevant or engaging enough, leading to immediate exits. We can infer that these channels are less targeted towards customers which often lead to only quick engagements with the content before exiting.

Overall insights for engagement metrics

KOL marketing is shown to be one of the better channels in engaging customers and converting them to sales for Shopee.

Marketing channels such as Email have been useful in engaging customers to view and engage with the marketing content, but less effective in converting these customers to sales.

2.2.3 Total revenue over time

For additional analysis, we look at the revenue generated from each channel over time.

A spike in revenue is observed for KOL and Social Media in November. This suggests a highly successful marketing campaigns likely involving KOL Referrals, Livestreams and Social Media ads.
In-App, KOL and Social Media: These channels consistently generate higher revenue, indicating their profitability in engaging customers.
Email, SMS and Website are lagging behind in terms of contribution to overall revenue.

Seasonal patterns

The End-of-Year season generated the highest revenue across multiple channels, especially so in November, likely due to the Shopee's mega 11.11 sales.

2.2.4 ROI Evaluation and overall analysis

Finally, evaluating the ROI of each marketing channel, we can see that:

KOL (Key Opinion Leaders) has the highest ROI, exceeding 280%. This suggests that influencer marketing is the most effective in driving sales.
- KOL is also one of the most cost effective channels with high Click-Through Rate and high Conversion Rates
- Shopee works with KOL (e.g. Livestreamers and brand ambasadors) that has proven to effectively promote brand awareness and sales. This suggests the success of Shopee's Affiliate Marketing Solutions (AMS) which helps sellers expand their brand presence by connecting them with KOLs and other affiliate networks that provide content that drives high engagement which successfully converts visitors to sales.
Social Media also has a high ROI, exceeding 220%. This suggests that investments in social media campaigns yield substantial returns.
- Social Media is also the most cost effective channel and has a relatively high CTR compared to other channels.
- Shopee uses a wide variety of Social Media Marketing techniques on Social Media platforms such as Instagram, FaceBook and TikTok, with catchy jingles that proves to effectively draw in customers and sales, and these have proven to be effective in generating high number of sales.
In-App marketing also holds a strong ROI but is lower than Social Media and KOL. This indicates some effectiveness but may need further optimization.
- In-App marketing sits on the average in terms of cost-effectiveness.
SMS, Email, and Website show considerably lower ROIs (below 50%). This suggests that these channels are not as effective at generating revenue compared to the others. However, Email marketing has shown its potential in engaging customers with its content, hence Shopee can look towards further optimizing Email marketing to generate more leads.

2.1.5 Optimizing Marketing Channels and ROI

1. Marketing channels with the top ROI

Investment in KOL and Social Media marketing channels are well-spent.

The higher than industry average for Cost per Sale as a percentage of Average Order Value is justified by the high number of sales generated from the marketing channel, in turn leading to higher ROI.

2. Marketing channels with average ROI

In-App marketing has the third highest ROI, but lacking in terms of engagement metrics.

Shopee should leverage In-App Marketing to drive Mid-Funnel Engagement by implementing personalized product recommendations and limited-time offers in-app to encourage repeat purchases.
Possible ways are to Integrate gamified experiences (e.g., points, badges) and flash deals within the app to keep users engaged and enhance the effectiveness of other campaigns.

3. Need for Optimization

The lower ROI channels (Email, SMS, Website) might require a re-evaluation of their strategies such as :

Further improve the targeting and segmentation of Email campaigns through anayzing customer search and purchase history, and extend similar ways of marketing to SMS marketing as well.
Increase the reach of Email marketing since it has proven its potential in engaging potential customers.

2.2 Key insights on impact of various promotional campaigns on sales

2.2.1 Total sales, AOV and unique customers by campaign type

To analyse the impact of the various Shopee campaigns on sales, we will mainly look at the total sales generated via each campaign, the AOV of each campaign and the unique customers brought by each campaign.

1. Total Sales by Campaign Types

Mega Sales significantly outperformed all other campaigns, with an extremely high total sales ($99,184.88) compared to other campaigns. Bundle promotions, however, has the lowest sales ($2,504.25).

2. Average Order Value by Campaign Types

The Livestream Exclusive campaign has the highest average order value (AOV) of $48.95. This indicates that customers purchasing through Livestream campaigns tend to spend more per transaction, which is a positive indicator for profitability.
The AOV across other campaigns (ranging from about $35.52 to $42.19) shows relatively healthy spending, but the difference in AOV suggests that different promotional strategies may lead to varying customer spending behaviors.

3. Unique Customers by Campaign

Mega Sales again stands out with 1,495 unique customers, indicating its broad reach and appeal. This suggests that the campaign was successful not only in attracting existing customers but also in acquiring new ones.
Other campaigns like Bundle Promotions and Flash Sale attracted significantly fewer unique customers (74 and 79, respectively), indicating potential areas for improvement in customer acquisition strategies.

Overall observations from visualisations

The Mega Sales campaign appears to be the most effective across all metrics, suggesting that strategies used in this campaign could be analyzed and potentially replicated in other campaigns.This also shows how the Mega Sales campaign have best met its promotional campaign aims through increasing sales greatly and attracting much higher unique customers.
The high average order value of Livestream Exclusive indicates that customers tend to purchase more or are willing to purchase items with higher prices when they watch livestreams with live demonstrations and honest reviews of the products.
Mega Sales has the second highest AOV of ($42.19) indicates that there may be effective cross-selling strategies in place or marketing tactics that promote bulk-buying that could be leveraged further. This also shows how the Mega Sales campaign have met its promotional campaign aim of boosting sales and increasing AOV.
Low unique customer counts in Bundle Promotions and Flash Sales, suggest that there may be opportunities to improve outreach and engagement in these campaigns, so as to boost engagement level and eventually boost sales too. This shows that these campaigns could be improved to better meet its promotional campaign's aims.

2.2.2 Evaluating total sales of each campaign by category

Now, we will analyse what are the top categories where each campaign's sales are concentrated.

1. Mega Sales

The Mega Sales campaign has the highest total sales across multiple categories, particularly in Mobile & Accessories, Men Clothes and Health & Beauty, indicating a strong performance in these product categories. This suggests that products under these categories resonate well with customers during significant promotional events like the Shopee Mega Sales.
Categories like Home Appliances and Women Clothes also show noteworthy sales, highlighting their appeal during the Mega Sales.

2. Bundle Promotions & Seasonal Sales

Both Bundle Promotions and Seasonal Sales perform well in categories like Mobile & Accessories, though its total sales are generally lower compared to the Mega Sales. Seasonal sales also have relatively high sales under Home Appliances.

3. Flash Sale

Flash Sale has a relatively high sales for Mobile & Accessories and Baby & Toys categories.

4. Livestream Exclusive

Livestream Exclusive performs particularly well in Mobile & Accessories. Other categories such as Home Appliances and Health & Beauty performed relatively well too. This suggests that having exclusive livestream deals on products under these categories could potentially drive sales by attracting a targeted audience interested in real-time demonstrations and exclusive offers.

5. Next Day Delivery

Products under Mobile & Accessories, Health & Beauty and Women Clothes categories are rather popular for the Next Day Delivery Campaign. This trend indicates that customers highly value convenience and quick fulfillment for essential and frequently used items.

Overall observations from visualisations:

Overall, Mobile & Accessories, Health & Beauty and Home Appliances are consistently high-performing categories across various campaigns. The popularity of these categories may reflect evolving customer behavior, where customers are increasingly inclined to invest in technology, personal care products, and household items.

Overall findings for AOV of each campaign type

Among the six campaigns analysed, Livestream Exclusive stands out with the highest AOV of $48.95, indicating that getting KOLs to promote products on campaign day is useful as people tend to purchase more when watching KOL livestreams.
Additionally, Mega Sales stand out with the second highest AOV of $42.19, indicating that there is strong customer interest in bulk buying during Mega Sales. This suggests that offering deep discounts during prominent shopping events linked to Mega Sales dates may effectively encourage larger transactions. The 'Mobile & Accessories' category, with total sales of over $50,000, underscores the importance of this segment in driving revenue during Mega Sales.
The Mega Sales event recorded a total of 2,366 orders and generated $99,184.88 in sales, resulting in an AOV of $42.19. It has the 2nd highest AOV out of the other campaigns and the substantial volume of orders reflects strong consumer engagement during this period. Categories like 'Mobile & Accessories,' 'Men's Clothes,' and 'Health & Beauty' were pivotal in achieving these sales figures. This also shows that customers are attracted to the deep discount provided on Mega Sales, so there are more orders though AOV is not the highest.
Bundle Promotions has the lowest AOV at $32.52, which signifies that Bundle Promotions may not have met its promotional campaign aim as compared to the other campaigns since the main aim of Bundle Promotions is to increase AOV.

2.2.3 Analyzing AOV by campaign and marketing channels

Insights from visualizations:

Mega Sales

Mega Sales have a moderate AOV ranging from $18.43 to $45.38. As the AOV for all marketing channels are in a similar range, it suggests that large promotional events like Mega Sales are effectively marketed across the different channels. The moderate AOV during Mega Sales may also be due to high volume of low-value transactions as Mega Sales is the biggest campaign in Shopee which yields the highest total sales, and many products are sold at deep discount prices.
Affordable products such as small electronics, fashion items or daily essentials with lower AOVs may be heavily featured on Mega Sales, causing many customers to focus purchasing budget-friendly items during Mega Sales.

Bundle Promotions

SMS marketing is doing much better than the other marketing channels for Bundle Promotions with an extremely high AOV of $293.56. This indicates that SMS marketing is particularly effective at driving high-value orders for Bundle Promotions. On the other hand, Website (AOV of $3.24) and Social Media (AOV of $15.46) did not do so well.

Flash Sale

Flash Sales have a low to moderate AOV ranging from $7.99 to $59.65. However, Email is an exception with an extremely low AOV of $1.48. Flash Sales are designed to create urgency and encourage quick purchases. This can lead to a focus on low-cost items that customers feel they can buy impulsively rather than on higher-value purchases. The extremely low AOV for Email may indicate that audience engaging with email promotions may not be the same as those engaging with Flash Sales through other marketing channels.

Livestream Exclusive

Livestream Exclusive also has a high AOV of $225.31 from SMS marketing. Email also have a rather high AOV of $80.65. This suggests that customers are willing to spend significantly more when being provided with marketing content through SMS and Email marketing.

Seasonal Sales

Seasonal sales show decent AOVs throughout all marketing channels. In general, its AOV across all marketing channels are rather low compared to other campaign types.

Next Day Delivery

AOV for social media is the highest at $46.16, while Email has the lowest AOV of $19.25, reveals important insights into customer behavior, engagement strategies, and the effectiveness of various marketing channels. This may be because Social media platforms are inherently interactive, allowing brands to engage customers in a dynamic manner. Users often interact with content, such as comments, shares, and likes, fostering a sense of trust around the products being promoted, especially when assured with Next Day Delivery. This engagement can translate into a willingness to spend more per transaction.

Overall observations from visualizations:

Overall, the effectiveness of a marketing channel may vary depending on the campaign type. For example, SMS marketing show exceptional performance in Bundle Promotions and Livestream Exclusive, while Email works well for Bundle Promotions and Livestream Exclusives and Social Media works well for Flash Sale and Next Day Delivery. This highlights the importance of aligning the right campaign type with the most effective marketing channels.
Some campaign types, like Mega Sales and Seasonal Sales, show relatively consistent AOVs across channels, indicating effective marketing strategies. In contrast, others exhibit more variability, suggesting that certain channels may not be as well-suited for particular campaigns.

Overall findings from visualizations

1. SMS Marketing for Bundle Promotions (Extremely High AOV of 93.56)

SMS campaigns tend to target customers who are already engaged and willing to receive personalized offers. Since SMS is a direct and personal form of communication, it can be highly effective in reaching high-value customers who are more likely to spend more on exclusive bundles.
SMS campaigns often create a sense of urgency by sending limited-time offers or exclusive discounts. Customers who receive SMS messages may feel more inclined to take advantage of high-ticket bundle offers, which may lead to higher AOVs.

2. SMS Marketing for Livestream Exclusive (High AOV of 25.31)

Livestreaming allows for real-time interaction between the presenter and the audience. SMS is highly effective for promoting Livestream Exclusive deals because it is a personal channel that pushes immediate notifications. Customers who receive SMS updates about livestreams may be more likely to engage with the content, leading to higher AOVs.
During livestreams, products can be shown in real time, which may encourage customers to purchase higher-priced items they might have been unsure about before. SMS notifications about "limited-time" offers during the Livestream can create a sense of urgency, prompting customers to buy more expensive products.

3. Relatively high AOV for Email and KOL Marketing for Livestream Exclusive

KOL Marketing: KOL marketing plays a significant role in Livestream Exclusive campaigns. Influencers and opinion leaders have established trust and a following, which increases the likelihood that their audience will spend more during a livestream, especially if the KOL promotes premium products. Their endorsement provides social proof, encouraging customers to make higher-value purchases.
Email Marketing: Email marketing for Livestream Exclusives can be highly targeted. Shopee Singapore can send personalized offers to customers based on their previous behavior or interests (e.g., if a customer has purchased electronics before, they might receive an email about exclusive tech bundles). The personalization of offers can increase the AOV significantly because customers are more likely to buy products that align with their preferences.

4. Social Media Marketing has the highest AOV of 9.65 for Flash Sales

Social media is inherently designed to drive engagement and spur quick action, making it an ideal channel for Flash Sales, which often create urgency. Social media platforms allow customers to interact with content, share deals with their friends, and quickly make a purchase. The interactive nature of social media (likes, shares, comments) helps customers feel more connected to the deals, encouraging them to spend more.
Targeted Flash Sale Campaigns: On platforms like Facebook, Instagram, or TikTok, Shopee can use targeted ads to promote Flash Sales to specific customer segments who are more likely to make higher-value purchases. For example, showing ads for premium products on Flash Sale during peak times can encourage higher spending.

2.3 Conclusion

The above analyses of Shopee Singapore's marketing channels and campaigns highlights that KOL and Social Media marketing are the most effective strategies with the highest ROI, and Mega Sales with the highest total sales. Given these strengths, Shopee should consider increasing investment in these areas and refining advertisement strategies to capitalise on their success.

Strategic Implications for ROI and AOV

Shopee should prioritise KOL and Social Media marketing channels, leveraging their high ROI and CTR. Increasing budget allocation to these marketing channels can potentially improve brand engagement and conversion rate, thus impacting sales performance.
Given the findings on AOV, Shopee should optimise promotional strategies for Livestream Exclusive and Mega Sales. Tailoring offerings to maximise customer interest in these campaigns can significantly enhance revenue and customer satisfaction.
Implementing strategies that improve customer retention for campaigns such as Next Day Delivery and Flash Sales**. These can potentially enhance overall customer lifetime value, thereby contributing positively to ROI.
Optimise low ROI channels strategically. Channels such as SMS, Website and Email are less efficient in terms of cost and contribution to ROI, however, it surprisingly contributes to exceptionally high AOV especially for Bundle Promotions and Livestream Exclusives. This suggests that these channels may not be the most effective for broad, cost-efficient customer acquisition but are highly impactful for engaging customers in high-value transactions.

Use these channels to drive their high-AOV purchases strategically, such as for Bundle Promotions and Livestream Exclusives, instead of routine marketing to leverage their ability to deliver targeted offers that encourage customers to make larger purchases, without overextending the budget on these channels.

Summary

By aligning marketing efforts with consumer preferences and campaign objectives, Shopee Singapore can significantly enhance its overall effectiveness in driving sales and increasing Average Order Value (AOV). The analysis reveals that KOL and social media marketing channels yield the highest ROI, while strategies like Mega Sales and Livestream Exclusive demonstrate strong potential for increasing AOV.

Continuous analysis and optimisation of these strategies will be necessary for maintaining competitiveness in the dynamic e-commerce landscape. Shopee should focus on refining its campaigns to better resonate with customers, leveraging data insights to adapt to changing consumer behaviors and preferences.Investing in effective campaigns that resonate with customers will ultimately lead to improved sales performance and customer retention.

In summary, by strategically investing in effective marketing channels and continuously optimizing campaign performance, Shopee Singapore is well-positioned to thrive in the evolving e-commerce market.

3. Note on Limitations

This analysis was conducted using a synthetically generated dataset created to simulate marketing performance metrics. While efforts were made to closely approximate real-world E-commerce data, we have to recognize several limitations that arise from the nature of synthetic data and the assumptions on which it is based:

Assumptions-based data: The dataset was generated based on predefined assumptions, including average order values, conversion rates, click-through-rates and bounce rates. These assumptions may not accurately reflect real customer behaviors or marketing and campaign dynamics, leading to potential discrepancies between the synthetic data and actual performance metrics.
Lack of real-world variability: The synthetic data may lack the variability and fluctuations present in real-world data. We have attempted to incorporate seasonal trends by defining usual Shopee sale dates and weighted the marketing and campaign data acccordingly. However, factors such as unforeseen market shifts, unique customer interactions, or change in consumer preferences are not fully captured, which can limit the generalizability of the insights.
Potential bias in channel and campaign performance: Assumptions made about each marketing channel's and campaign's effectiveness may introduce biases. For example, since conversion rates and costs were set based on hypothetical performance expectations based on average E-commerce metrics, the relative effectiveness of each channel is skewed, affecting ROI and other comparative metrics.
Simplified customer behavior: Synthetic data often may not capture the complexity of real interactions. For instance, brand loyalty, repeat purchases, or cross-channel interactions might not be fully represented, potentially impacting analyses of long-term value or cross-channel synergies.
Limitations in predictive validity: Insights derived from synthetic data should be interpreted with caution if applied to forecasting or strategic planning. Since the data does not reflect actual historical patterns of Shopee, any projections or predictions may not hold true under real-world conditions.

Conclusion: We have prepared the synthetic datasets due to lack of public datasets and statistics regarding Shopee's marketing channel and campaign's cost, revenue and performance and engagement metrics. We have used the datasets as a framework for our exploratory analysis to draw analytical insights to address Shopee's business objectives. Future analyses incorporating real performance metrics and historical data are recommended to achieve more accurate and actionable insights.