Home - grishasen/proof_of_value GitHub Wiki

πŸ“Š CDH Value Dashboard

The CDH Value Dashboard is a flexible, open-source application developed to help Pega clients measure the value of Customer Decision Hub (CDH) initiatives. It bridges the gap between technical implementation and measurable business impact through intuitive, story-driven reporting, engagement/conversion tracking, and Customer Lifetime Value (CLV) monitoring.

🧩 Key Capabilities

Capability Description
Engagement Lift Tracks impact of actions on user engagement (e.g. CTR, lift vs control).
Conversion Lift Measures conversion rate changes from CDH actions using test vs control.
CLV Improvement Analysis Evaluates changes in RFM parameters (recency, frequency, monetary value).
Business Experiment Analytics Supports A/B testing insights via z-score, chi-square, confidence intervals.
Operational Efficiency Monitoring Reports on model performance (AUC, Precision), data properties (variance, percentiles, volumes).
Chat with data Conversational data analysis.

πŸ“Œ Reporting Principles

  1. Tell a Story
    Start with the business narrative. Select metrics that support and reinforce the narrative for clarity and impact.

  2. Audience-Specific Views

    • Executives: High-level impact and ROI.
    • Product Teams: Detailed performance and usage data.

  3. Include Comparisons

    • Time-based trends
    • Benchmarks
    • Control groups (Hold-out, No Action, Crowd Wisdom)

  4. Control Group Design

    • Essential for all experimental evaluations
    • Configurable within NBA strategy extension points
    • Outcomes written back to Interaction History (IH)

πŸ“˜ Application Terminology

Term Description
Metric Quantitative performance indicator (CTR, conversion rate, CLV).
Marketing Dashboard Visual display of performance across selected metrics.
Report A combination of plots, filters, and datasets tailored to a business story.
RFM Customer segmentation by Recency, Frequency, Monetary value.
CLV Estimated total value of a customer across their lifecycle.

πŸ—οΈ Requirements

Category Requirements
CDH Inputs Interaction History (IH), Product Holdings
APIs Standard Feedback APIs (clicks, impressions, conversions)
Deployment Lightweight (Laptop, small cloud instances, on-prem)
Security No client data sharing; internal only unless NDA in place
Governance KPIs to be defined early (ideally during sales cycle)

πŸ“ˆ Analytical Capabilities

Types of Metrics Supported

  • Engagement: CTR, Lift
  • Conversion: Conversion Rate, Revenue
  • Descriptive: Count, Sum, Mean, Median, Std, Skewness
  • Experiment: Z-test, Chi2, Odds Ratios
  • Model Scores: AUC, Precision, Novelty, Personalization
  • Exploratory: RFM, Distribution Skews, Percentiles

Visualization Types

  • Bar Polar Plot
  • Gauge
  • Treemap
  • Heatmap
  • Line & Bar Plots with facets and significance indicators

πŸ” CLV & RFM Analysis

CLV is treated as a strategic KPI and not a direct optimization target. The dashboard provides:

  • Baseline RFM scoring
  • Segmentation (At-risk, Top Spenders, Premium)
  • CLV Lift Measurement via:
    • Pre/Post intervention analysis
    • Control vs Test group comparisons
    • Visualization of spend/frequency shifts

🧠 Advanced Feature – β€œChat with Data”

  • Powered by OpenAI
  • Allows free-form querying
  • Ideal for use cases where standard dashboard filters don't suffice
  • Includes built-in code review and guided prompts

πŸ“¦ Distribution & Licensing