The second value: test before changing a decline route.

Measure agreement, disagreement, coverage and later outcomes on the lender cohort before a second value changes any lending decision.

A second AVM should not be sold as an automatic way to approve more loans. It is another estimate that can be tested where the current route produces uncertainty or sends cases to review.

Different models can produce different confidence outputs on the same property. That makes the overlap a useful evaluation question, not proof that either model is correct on an individual case.

Where the overlap lives

Choose the segment from lender evidence rather than a market anecdote. It might be a property type, geography, price band or confidence band where the current workflow sends enough cases to review to justify a test.

Run the second view alongside the first, keep the existing decision policy in control, and measure coverage, agreement, disagreement and later reference outcomes. The lender then decides whether any disagreement rule is safe and useful.

A second value creates another signal. Only a lender-run evaluation can show whether it creates a better route.

This is not a rip-and-replace

Ecosystem standardisation is real. If your brokers, your funders and your board all expect one brand, we understand the gravity of that, and we are not asking you to fight it. The second value sits behind the incumbent as an arbitration layer, not in front of it as a replacement.

IF current_route = REVIEW -> record second_view
COMPARE coverage, agreement and later reference outcome
CHANGE routing only after governance approval

How to test it

The honest test is a bounded side-by-side on a lender-controlled cohort. State the current route, select the segment, define the reference outcome and record the result by confidence band. Start without moving customer data where possible; agree a secure route in writing only if the question requires it.