The merchant is in the room and you are not
Every structural problem in point-of-sale home-improvement lending follows from this one fact.
The loan was underwritten against a scope. Then the scope changed, and nobody told the lender.
A lender underwrites a project against a scope and an amount. Work begins. The scope changes — because construction always changes — and the change is agreed verbally, or by text message, or in a note on a clipboard. The loan is now supporting work nobody underwrote.
This is the most under-managed risk in home-improvement lending, and it is under-managed because it does not look like risk. It looks like a project management detail.
The intuitive control is a threshold: bring the lender in on any change above some amount. It fails in the obvious way. A threshold applied per change order catches one large change and misses a series of smaller ones that sum to more.
The threshold has to be cumulative against the original agreed total. Then the sequence cannot walk past a limit written to catch the single event.
One change order is a fact of construction. Twenty is a business model. A per-order threshold cannot tell the difference.
The deeper fix is treating a change order as a new agreement rather than an edit. If amending the scope overwrites it, you have destroyed the thing the borrower originally approved and you can no longer show what they agreed to. If it creates a new version, both survive, and the chain from the original to the current state is reconstructable at any date.
That also solves the practical problem for the contractor: work already completed and evidenced carries forward, rather than needing to be re-documented against a new scope.
Ask for the distribution of cumulative scope growth across your book. If nobody can produce it, that is the finding.
Pick a closed project. We will show you the document you would hand a regulator. If it does not answer the question, nothing else matters.