Case study, BNPL · Modeled
First contact from 3 days to 4 hours.
Modeled on a BNPL operator with about 10,000 active loans and roughly ₱180M in monthly issuance. Pre-default early warning, with AI voice and SMS handling about 70% of Tier-1 outreach, so human agents only touch the accounts that need a person.
- Operator
- BNPL, ~10k active loans
- Monthly issuance
- ~₱180M
- Delinquency
- 6.1%
- Collections team
- 24 agents
The problem
Volume outran the team.
Volume scaled faster than the collections team could hire. New-cohort delinquencies land before anyone can call. The per-account economics are tight, so every minute of agent time counts.
What we’d plug in
What we'd plug in.
The AI carries the routine outreach and the model decides who needs a human, so the existing 24-person team works only the accounts where they add the most.
Pre-default early warning
The model flags accounts before they go delinquent, so outreach starts while the balance is still small.
Autonomous Tier-1 outreach
AI voice and SMS handle about 70% of first-line contact on their own, across the channels each borrower responds to.
Human-required routing
Agents only touch the accounts the model flags as human-required, which is where their time pays off.
Settlement optimizer
For each account, the model weighs recovery against the probability the borrower accepts, and recommends the offer that maximizes the two together.
The modeled numbers.
Modeled at current volume, on PH market data, not a signed customer.
- 22% → 35%
- Recovery rate on delinquencies
- 3 days → ~4 hrs
- Time to first contact
- −38%
- Collections headcount needed
- +13 pts
- Recovery-rate lift
- < 2 mo
- Payback period
Want this modeled on your numbers?
Send us your monthly issuance and delinquency rate. We will model your recovery lift, headcount efficiency, and payback, then walk you through it on a 20-minute call. No pitch.