Audit. Prove. Bill on the lift.
The wedge is measurement: merchants already pay outcome fees to recovery agents, and none of those invoices says how much of the outcome the agent caused. Start there.
Market context
~70% of online shopping carts are abandoned on average (70.22% aggregate of 50 studies, 2006-2025).
In May 2026 UPI processed ~23.2B transactions worth ~₹29.9 trillion, and carried ~83% of India's digital-payment volume in 2024.
COD is returned far more than prepaid: ~24-26% RTO on COD vs ~2-10% prepaid; a ₹1,000 RTO order costs a brand ~₹285 while earning nothing.
Land first
- Already paying outcome fees
Merchants on a gateway recovery agent or a per-recovery vendor - the invoice audit is an immediate, owned insight.
- COD-heavy categories
Fashion and beauty, where RTO-gated COD and margin-gated offers matter most.
- Enough volume for a read
Roughly 150 or more eligible failures a day: at a 10-point lift from a 20% baseline, a 10% holdout reads in about two weeks.
Who buys, and the line for each
Stop paying outcome fees on customers who would have paid anyway.
Recover the high-intent drop-offs, and prove the lift before scaling spend.
Every contact and every suppression is logged with its rule; the AI cannot say what the rules forbid.
A defensible measurement story for outcome-priced recovery agents.
Pricing hypotheses
The merchant pays only on revenue the holdout says the outreach caused. Aligns the vendor with the merchant and survives a CFO's audit.
Simpler to buy, and the fallback when a merchant's volume is too small for a powered holdout in a reasonable time.
Launch sequence
- 1Shadow mode on one merchant's existing agent
- 2Holdout + guardrails for one powered read
- 3Publish the incremental number to the merchant
- 4Move the fee onto the measured lift
- 5Open to more rails and categories
Market statistics as cited, dated at source: Baymard, NPCI, Unicommerce and others. The volume threshold comes from the holdout planner in the control room.