Four points behind

A product case question from a product manager interview, set at Kavach Pay, a fictional b2b payments / risk company.

The question

Skills this case tests

A starting metric tree

Authorisation rate = (1 − pre-authorisation block rate) × issuer approval rate

How to approach this case

Clarify the metric before the gap: does the authorisation rate include payments Kavach stops before the issuer sees them, and is the competitor benchmark measured the same way? Agree the fraud limit in basis points up front. Then split the four points between what happens inside Kavach and what happens at the issuing bank, and diagnose each part separately. For Kavach's own rules, ask how accurate they are; for issuer declines, look at the decline types and how Kavach responds to each. Segment by merchant category, since risk and approval rates vary widely. A one-quarter goal favours levers you can prove quickly on held-out traffic. A strong close sequences two or three levers with sized gains, shows fraud stays under the limit, and states the precision-versus-recall trade-off.

The full method for mixed diagnose-and-decide questions · A worked example answer

After the mock interview you get scores on five skills, what to fix with quotes from your answers, and the ideal path for this case.