CredVantage: Underwriting the Unbanked MSME
India has tens of millions of micro and small businesses that banks find too small, too informal, and too risky to lend to. CredVantage built a lending model around alternative data — GST filings, UPI transaction flows, and supplier relationships — to underwrite borrowers the formal system ignored. This case studies the trade-offs between growth, default risk, and unit economics in MSME fintech lending.
The Problem
Traditional credit scores assume a paper trail that informal businesses simply do not have. The result is a vast credit gap: viable businesses unable to borrow because the system cannot see them.
Exhibit A
Alternative Data Stack
GST, UPI, and platform-review data were combined into a composite score that proxied for repayment intent and capacity.
The Model
CredVantage replaced the missing paper trail with digital exhaust: GST returns, UPI inflows, and platform reviews became the underwriting signal. Small-ticket, short-tenure loans kept risk contained while the data model learned.
Exhibit B
Loan Tenure vs Default
Shorter tenures showed materially lower default rates, letting the model gather repayment signal before extending larger lines.
The Tension
Faster approvals drive growth but also adverse selection. The hard problem is calibrating how much to lend, to whom, and how quickly to scale before defaults catch up to disbursement.
Exhibit C
Unit Economics
Contribution margin per loan turned positive only after the customer’s second or third cycle, making retention the core lever.
Discussion Questions
- 1.Is alternative-data underwriting durable, or does it degrade as borrowers learn to game the signals?
- 2.How should CredVantage balance growth velocity against portfolio quality?
- 3.What is the right cohort to subsidise during the learning phase?
- 4.Does retention-driven profitability create a lock-in moat or a fragility?
- 5.How would a rate-cycle shock change the calculus?