Approval rate was the product

Volume had been flat for three quarters against rising traffic. The commercial team was buying more leads; the risk team was defending the loss rate. Both were right and neither was solving it.

We replayed twelve months of declined applications against outcomes. One bureau field was missing on forty-four percent of thin-file applicants and was being treated as an adverse signal rather than as absent.

We rebuilt the scorecard with an explicit missing-data treatment, added an affordability step for that cohort, and set limits low on first draw.

Approval rate up twenty-one percent, loss rate up eleven basis points against a forecast of forty, and cost of acquisition down because the same traffic converted.