Context
A microfinance bank scaling digital credit across mobile and web.
Problem
Each customer segment had its own decisioning flow. Improving one flow did not improve the others, and growth was limited by whichever step was weakest.
System and insight
Customer journeyPreapprovalDecisioning and policyProduct routingCore bankingData warehouseCustomer messaging
Growth in credit is a product of three things: how many suitable customers can be reached, how much each can responsibly use, and how often they come back. Each was held back by a different part of the system.
Trade-offs
- One orchestration layer instead of many segment flows: more design up front and a shared dependency, in exchange for consistent decisions and reuse.
- Segment-based routing with parallel data-warehouse queries: faster for the customer, harder to coordinate in engineering.
- Policy engines, alternative data and machine learning together: more coverage than rules alone, and more to govern.
Intervention
- Architected a universal preapproval platform: one orchestration layer with parallel data-warehouse queries and segment-based routing.
- Designed credit decisioning that combines policy engines, alternative data and machine learning for offers and limit assignment.
- Defined the API and integration strategy for lending workflows and core-banking connectivity.
- Set up experimentation (A/B tests, product analytics, measurement frameworks) to tune conversion, risk and experience.
My role
Technical Product Manager at Renmoney. I led product strategy, architecture, requirements and delivery governance across engineering, data science, risk, compliance, marketing and operations, a group of more than 50 people. The engineering teams built and ran the systems.
Result
- Digital lending revenue and portfolio grew by more than 50%.
- Product launch cycles shortened from three months to three weeks.
- Funnel drop-off fell by 20% after a stronger KYC provider went in.
Learning
The problem wasn’t the feature. It was the system around it.