Consumer Credit

A lender usually arrives with the data spread across an origination platform, a decisioning engine fed by bureau data, a servicing system and a collections dialler, with marketing spend sitting in a fourth place again. Each team trusts its own extract, so the approval rate risk quotes, the funded volume finance books and the acquisition cost growth defends never quite reconcile — and the monthly contribution number is rebuilt by hand every board cycle from whichever version won the argument that month.

We land the events behind all of it — application, decision, funding, payment, contact, arrears status — in your own BigQuery, recorded once with the time each happened, and rebuild the loan-book position back as far as the data allows. That history is usually the unlock: vintage cohorts and roll rates that never existed become reportable, and the leaks show themselves. A typical one is a channel with a healthy approval rate and quietly worse vintage loss than the book average, which no departmental report was ever built to catch because origination and collections are measured by different people.

Once it is live the credit committee works from cohorts that are current rather than a month old, Consumer Duty outcome reporting regenerates itself with a full audit trail behind every figure, and risk, finance and growth argue about what to do rather than about whose number is right. The affordability and arrears views update on the same definitions the board pack uses, so the decision and the evidence for it come from one place.