Projected revenue lift from revised routing
The problem
Borrowers could apply for a loan without knowing which lender was most likely to approve them. For the business, a poor match also meant a missed opportunity to earn revenue from an application.
What I analyzed
I examined approval patterns across three lenders. Alongside credit scores, I tested indicators such as disposable income, loan-to-income ratios, and prior credit events. Credit score was the strongest approval signal in the analysis; the lenders differed in how selective they were.
The decision
I grouped borrowers by credit score and compared expected revenue per application: approval probability multiplied by lender payout. The model favored the more selective lender for high- and mid-score groups and the more permissive lender for lower-score applicants.
The recommendation
Routing each segment to its highest-value lender produced a projected 21.9% revenue lift without adding traffic. I recommended a real-time matching approach that could be updated as approval patterns and payouts changed.
The limit of the result
This was a modeled opportunity. The result depends on the approval and payout assumptions in the supplied data; it does not establish what a live rollout would achieve.