KPI Sync

Date: 2025-04-29 · Duration: 68.45999908447266 min Organizer: Nader Attendees: Ahmad, Mike, Nader

Summary

Card Decline Rate Insights: Reviewed metrics for Twilight Payment analytics; focus on two sandbox merchants (Crocs, Beverly Hills). Need to identify factors contributing to decline rates.
KPI Correlation Analysis: Ahmadreza presented analysis showing 85% correlation decline count (vs. 78% decline volume). Noted limitations in current analysis focusing on isolated features (Nader’s point).
Importance of Transaction Data: Michael stressed need for comprehensive response data (e.g., AVS, CVV). Emphasized value of complete data for understanding transaction failures beyond sandbox data.
Analytics Tile Architecture Alignment: Agreed on three-tier structure: Level 1 (basic KPI), Level 2 (KPI over time), Level 3 (detailed breakdown with multiple visualizations). Michael’s suggestions to simplify medium tile.
Implementation Strategy for V1: Decided on hard-coded visualizations for initial launch focusing on card type, transaction amount, and decline reasons. Future versions to incorporate ML for relevance.
Advanced Model Discussion: Explored potential for a multi-class classification model to predict specific decline reasons, aiming to refine understanding of decline causes and settings adjustments.
Action Items:
- Ahmadreza: Close ticket with full visualizations; list output graphs; convert findings to config structure.
- Michael: Lead next project phase.

Ahmadreza Abdoli Close the current ticket with a full visualization with card type and transaction distribution (53:37) List the four graphs that would be the output in a comment on the ticket (01:07:04) Convert findings into a config structure for frontend implementation (55:24)

Michael Wright Take responsibility for next phase of the project (01:07:41)