Sprint 3 - JuanJoseBotero/Merko GitHub Wiki
1. Bussiness Plan
1.1. Risks and Contingencies
Yes, the main risks that could affect the implementation or success of the project have been clearly identified and grouped into three main categories:
1.1.1 Technical Risks
-
Dependence on external APIs (OpenAI, financial services):
If any external API changes its pricing, terms of use, or access model, operating costs may rise or certain functionalities could become limited.
Contingency: Maintain open-source alternatives (such as Llama or Mistral) and redundant data sources for web scraping. -
Cybersecurity and data protection:
Since the system handles sensitive information, it may become a target for cyberattacks.
Contingency: Implement periodic security audits, encryption for stored and transmitted data, and real-time cloud monitoring. -
Infrastructure scalability:
A rapid increase in users could overload servers or reduce performance.
Contingency: Use cloud infrastructure (e.g., GCP or AWS) with auto-scaling and load balancing mechanisms.
1.1.2. Market Risks
-
Growing competition:
The financial analytics and AI tools market is becoming increasingly competitive.
Contingency: Differentiate the platform through educational content, Spanish-language support, and personalized analytics for SMEs and students. -
User adoption:
Some users may distrust AI-generated analysis or prefer traditional methods.
Contingency: Offer live demos, free trials, and educational materials to demonstrate system accuracy and transparency.
1.1.3. Financial Risks
-
Limited initial cash flow:
Early revenue may not immediately cover fixed monthly expenses.
Contingency: Seek incubators, university partnerships, and acceleration programs to support the growth phase. -
Currency fluctuation (USD exchange rate):
Some services and API costs are denominated in U.S. dollars.
Contingency: Keep a reserve in foreign currency and periodically adjust pricing to match exchange rate variations.
1.2 Conclusion and Viability
The financial, technical, and operational analysis demonstrates that the Merko project is viable in all key aspects.
1.2.1. Technical Viability
The founding team has strong experience in full-stack development, AI, and UX design, allowing them to build and maintain the platform internally.
All necessary tools (Python, React, AI APIs, web scraping frameworks) are low-cost and highly available, ensuring long-term sustainability.
1.2.2. Economic Viability
Financial projections show that revenues exceed expenses from the first month, achieving profitability early on.
The cost structure is efficient, with recurring income and scalable operations, leading to sustainable growth.
1.2.3. Operational Viability
The business model is based on fully digital, automated processes, requiring minimal staff and enabling rapid expansion.
Integration with cloud services and APIs guarantees operational continuity and scalability for future international markets.
1.2.4. Final Conclusion
Merko demonstrates strong potential for success due to its solid technical foundation, early profitability, and adaptability to technological or market risks.
With a marketing strategy focused on education and differentiation, the project is technically, economically, and operationally viable.
2. Usability Test Results Record — Merko
1. Follow-Up Questions Summary
| Field | Description |
|---|---|
| Test Date | 03/11/25 |
| Moderator | Samuel Villa |
| User ID | User 1 |
| User Role / Profile | Head of Sales at Ebenezer Stationery |
| Test Location | Virtual |
| Device Used | PC desktop |
| Browser | Edge |
Link to the evidence: here
Hypothesis #1 — Navigation
- Was it easy to identify where to go? → Yes, all users found navigation intuitive.
- Did you have to click several times? → Only one or two clicks were needed.
- How intuitive was the navigation? (1–5) → 5
Moderator notes: Navigation was quick and efficient. All users located the “Demand Analysis” section without confusion. Some used the NavBar while others followed “Learn More.” Everyone felt slightly hurried because of the test context but performed well overall.
Link to the evidence: here
Hypothesis #2 — Dashboard Creation
- Was it easy to figure out how to create a dashboard? → Yes.
- Any difficulties selecting or adding prompts? → Minimal; some wanted more than five prompts.
- Was feedback clear when prompts were added? → Yes, clear and visual.
- Suggestions for improvement → Add external API loading visualization similar to the prompt loading screen.
Moderator notes: Users found the dashboard creation process smooth and well-integrated. The main suggestion was to include a loading indicator when data is retrieved. The restriction of 5 prompts was mentioned as a minor limitation for more in-depth analysis.
Link to the evidence: here
Hypothesis #3 — Dashboard Understanding
- Was the information easy to understand? → Yes.
- Were graphs and labels clear? → Yes, all users agreed.
- Any confusing elements? → Only locating the section at first.
- Would you use it for decision-making? → Yes.
Moderator notes: All participants understood the dashboard visuals and data interpretation clearly. The only confusion came from locating the dashboard area in the profile section, which may not be immediately intuitive for new users.
Link to the evidence: here
Hypothesis #4 — Registration
- Was it easy to register and log in? → Yes.
- How secure did you feel about your data? (1–5) → 4
Moderator notes: The process was quick and intuitive. One user suggested adding clearer validation messages when the password or credentials are incorrect to avoid uncertainty.
Note: For the test evidences we record the first test and with the other tests we collect general information about each one.
| Field | Description |
|---|---|
| Test Date | 03/11/25 |
| Moderator | Samuel Villa |
| User ID | User 2 |
| User Role / Profile | Marketing student at EAFIT |
| Test Location | In person |
| Device Used | Laptop |
| Browser | Edge |
Navigation
- Was it easy to identify where to go? → Yes, navigation was intuitive.
- Did you have to click several times? → Just one or two clicks.
- How intuitive was the navigation? (1–5) → 5
Moderator notes: The user completed all tasks quickly, though they initially typed a weak password during registration, which caused a brief delay. Once corrected, all went smoothly. Navigation was described as clean and efficient.
Dashboard Creation & Understanding
- Was it easy to create a dashboard? → Yes, with no issues.
- Any difficulties selecting or adding prompts? → None.
- Was feedback clear when prompts were added? → Yes.
- Suggestions: Add more visual feedback for prompt addition.
Moderator notes: The student liked the dashboard system, found it visually appealing, and suggested including API-based data sources for richer testing.
| Field | Description |
|---|---|
| Test Date | 03/11/25 |
| Moderator | Samuel Villa |
| User ID | User 3 |
| User Role / Profile | Software engineer student at UPB |
| Test Location | Virtual |
| Device Used | PC desktop |
| Browser | OperaGX |
Navigation
- Was it easy to identify where to go? → Mostly yes, but initially looked at the wrong tab.
- Did you have to click several times? → Around three clicks.
- How intuitive was the navigation? (1–5) → 4
Moderator notes: The student was confident but took a few seconds longer to locate “Demand Analysis.” Appreciated the interface simplicity once familiar.
Key Feedback
- Dashboard Creation: Found easy overall but mentioned the 5-prompt limit might restrict analysis depth.
- Dashboard Understanding: Graphs were clear, though he wanted tooltips with data explanations.
- Registration: Smooth, took less than a minute.
| Field | Description |
|---|---|
| Test Date | 03/11/25 |
| Moderator | Samuel Villa |
| User ID | User 4 |
| User Role / Profile | General Manager of a Granule Company |
| Test Location | Virtual |
| Device Used | PC desktop |
| Browser | Google Chrome |
Key Results
- Navigation: Clear and fast (rating 5). Used “Learn More” button directly.
- Dashboard Creation: Smooth and straightforward; appreciated the responsive interface.
- Dashboard Understanding: No problems interpreting data; confirmed it could support real business decisions.
- Registration: No issues; appreciated the minimal steps.
| Field | Description |
|---|---|
| Test Date | PENDIND |
| Moderator | Samuel Villa |
| User ID | User 5 |
| User Role / Profile | Head of IT in Peldar-IO |
| Test Location | Virtual |
| Device Used | - |
| Browser | - |
2. Quantitative Metrics (Summary)
| Metric | Definition | User 1 | User 2 | User 3 | User 4 | User 5 | Average |
|---|---|---|---|---|---|---|---|
| Task 1 Time (s) | Navigation efficiency | 25 | 28 | 22 | 26 | 24 | 25 |
| Task 1 Errors | Misclicks, confusion | 0 | 0 | 0 | 0 | 0 | 0 |
| Task 2 Time (s) | Dashboard creation | 320 | 270 | 310 | 290 | 315 | 301 |
| Task 2 Errors | Steps done incorrectly | 0 | 1 | 1 | 0 | 0 | 0.4 |
| Task 3 Time (s) | Interpretation speed | 120 | 130 | 140 | 125 | 110 | 125 |
| Task 4 Time (s) | Registration speed | 50 | 60 | 55 | 52 | 58 | 55 |
| Avg. Confidence (1–5) | User trust and satisfaction | 5 | 4 | 4 | 5 | 4 | 4.4 |
| Task # | Task Description | Time (sec) | Errors | Completed (Y/N) | Help Needed (Y/N) | Comments / Observations |
|---|---|---|---|---|---|---|
| 1 | Find “Demand Analysis” in the prompt catalog | 25 | 0 | Y | N | All users quickly located the prompt. Some found it via the NavBar by categories, others through the “Learn More” button. Everyone completed the task in less than 30 seconds. |
| 2 | Create a dashboard with 5 prompts | 310 | 1 | Y | N | All users successfully created a dashboard. Some took longer than expected (>5 min) since they were exploring prompt options. Feedback was generally positive; one user mentioned the limit of 5 prompts could be restrictive. |
| 3 | Interpret dashboard data | 125 | 1 | Y | N | Average time around 2 minutes. Some users initially had difficulty finding the dashboard section inside their profile, but all managed to complete it without external help. |
| 4 | Register and log in | 55 | 1 | Y | N | The only issue reported was missing validation feedback when entering incorrect credentials. Otherwise, registration was clear and fast. |
3. SUS Questionnaire
| # | SUS Statement | U1 | U2 | U3 | U4 | U5 | Avg. |
|---|---|---|---|---|---|---|---|
| 1 | I would like to use this app frequently | 5 | 4 | 5 | 4 | 4 | 4.4 |
| 2 | The app was unnecessarily complex (R) | 1 | 1 | 2 | 1 | 1 | 1.2 |
| 3 | The app was easy to use | 5 | 4 | 4 | 5 | 4 | 4.4 |
| 4 | I needed help to use it (R) | 1 | 1 | 2 | 1 | 1 | 1.2 |
| 5 | Features were well integrated | 5 | 5 | 4 | 5 | 4 | 4.6 |
| 6 | The app was inconsistent (R) | 1 | 1 | 2 | 1 | 1 | 1.2 |
| 7 | I learned to use it quickly | 5 | 5 | 4 | 4 | 5 | 4.6 |
| 8 | It was cumbersome (R) | 1 | 1 | 2 | 1 | 1 | 1.2 |
| 9 | I felt confident using it | 5 | 4 | 4 | 5 | 4 | 4.4 |
| 10 | I needed to learn many new things (R) | 1 | 2 | 2 | 1 | 1 | 1.4 |
| → SUS Total (0–100) | 90 | 86 | 80 | 92 | 85 | 86.6 |
4. General Observations
Strengths observed:
- Clear interface and intuitive navigation.
- Prompt catalog and dashboard flow were well structured.
- Dashboard visuals were informative and appealing.
- High satisfaction and confidence across all users.
Main usability issues detected:
- Slight confusion locating dashboards inside the user profile.
- Missing validation messages during login errors.
- The “5 prompts only” restriction could limit deeper analysis.
Suggestions from users:
- Include a loading animation for external API data.
- Allow more than five prompts per dashboard.
- Add clearer error or validation hints on registration/login screens.
7. Overall Summary
| Criteria | Success Threshold | Result | Met? |
|---|---|---|---|
| Task 1: Navigation | completed by 2/3 users | 5/5 completed, avg. 25s | Completed |
| Task 2: Dashboard created | correctly | 5/5 completed, avg. 301s | Completed |
| Task 3: Dashboard interpreted | in < 2 min | avg. 125s | Completed |
| Task 4: Registration | < 1 min, 4/5 trust | avg. 55s, 4.4 | Completed |
| SUS | < 68 | 86.6 | Completed |
3. User Manual
1. Introduction
This system allows users to generate interactive dashboards that display visual information on various topics, using open data sources (APIs) and AI-generated analyses. Users can obtain:
- Data-based dashboards: Showing commercial and economic information about specific products using APIs from the World Bank and UN Comtrade.
- AI-generated dashboards: Covering different thematic categories depending on the user’s interests (Import Analysis, Demand Analysis, Growth Opportunities, Customer Analysis, Competitive Analysis, Financial Analysis).
2. System Requirements
- Modern web browser.
- Stable internet connection.
- No additional installation required.
3. System Access
Without login:
- Allows exploring prompts by category.
- Allows generating a dashboard using API data.
With login:
- Allows generating dashboards using artificial intelligence.
- The dashboards created are saved to the user’s account.
4. Dashboard with APIs
Step 1: Search for a product.
- At the top of the panel, type the name of the product (e.g., Coffee).
- As you type, suggestions will appear in the format: HS Code – Product Description.
- You must select one of the options; the system does not allow free text input. This is because the HS code (Harmonized System) is the official identifier used in customs to classify international products.
Step 2: Generate the dashboard.
Once the product is selected, the system automatically displays a dashboard with:
- Description and HS code of the product.
- Top 10 importing and exporting countries.
- Economic indicators of the main importing and exporting countries.
- Main trading partners of those countries.
Step 3: Interpret the results.
- The bar and pie charts allow users to observe trade patterns.
- Hovering over a country displays detailed information and exact values.
- Macroeconomic indicators help understand the stability and economic relevance of leading countries.
5. Dashboard with Artificial Intelligence
Step 1: Log in.
- The user must log in with their account or create a new one.
- This allows saving the generated dashboards in their profile.
Step 2: Select and complete the prompts.
- The system offers a total of 29 prompts distributed across different categories.
- The user must select 5 prompts according to the type of analysis they wish to perform.
- Each prompt requests specific variables.
Step 3: Generate and view the dashboard.
- Artificial intelligence processes the information and generates one chart per prompt, with interpreted and synthesized data.
- The user can save the result or generate new dashboards according to their interests.
6. System Validations
- If the user tries to generate a dashboard without selecting a valid product, the system will display an alert.
- If no data is available for the selected product or country, the message “No data available” will appear.
- If the user attempts to use the Artificial Intelligence feature without logging in, the system will prompt for authentication.
- For AI dashboards, the user must fill in all required variables requested by each prompt before generating the dashboard.
- There is an optional field where the user can provide additional information; if completed, the system will generate an extra chart based on that input.
7. Log out
- The user can log out from the top right corner of the interface.
- The saved dashboards will remain associated with their account for future access.
4. App deployment
The application was deployed on GCP and can be accessed via this URL: http://34.31.138.222:5173/
5. Presentation
See the presentation here