Construction and rehab draw fraud is not a theoretical risk for private lenders — it is a recurring monthly event. Most of it is small, opportunistic, and hard to prove without the right data. Some of it is organized, repeat-offender behavior across multiple lenders.
This is a working guide to the draw fraud patterns we see most often, the signals that catch each one, and how a verification layer like LendlyX changes the economics of fraud for the bad actors targeting your capital.
The six most common draw fraud patterns
1. Work-not-completed draws
The borrower or contractor submits a draw for scope items that have not actually been completed. Photos may be from the prior draw, from a different property, or from a marketing site.
Signals: reverse-image match against prior draws, EXIF date earlier than draw window, GPS outside the property polygon, scope item already drawn in a prior cycle.
2. Photo recycling
The same photos appear in draws across multiple cycles, often with slight crops or filters applied to defeat naive duplicate detection.
Signals: perceptual hash match against prior submissions, pixel-level near-duplicate detection, model-based content embedding similarity.
3. Scope inflation
The budget contains line items that do not match the actual scope of work — fictitious permits, padded labor, materials at retail that were sourced wholesale.
Signals: budget cross-check against market data, line-item comparison against contractor invoice patterns, cross-document inconsistency between scope, invoice, and lien waiver.
4. Sequence fraud
Draws come in a sequence that does not match physical reality — drywall before framing, finishes before mechanicals, exterior before structure.
Signals: sequence-aware decisioning that scores each draw against the prior draw and the construction phase, not just the line item.
5. Cross-tenant photo reuse
A contractor or borrower active across multiple lenders submits the same evidence to all of them.
Signals: consortium-level fingerprinting (where available), photo provenance across the LendlyX network, repeat-actor signals on the borrower or contractor record.
6. Screen capture and AI-generated media
The submitted "photo" is actually a screen capture from another app, a download from a listing site, or an AI-generated image of a fictional condition.
Signals: screen-capture detection on the capture stream, AI-generation classifiers run on every upload, EXIF anomaly detection, capture-session device fingerprint.
What lenders get wrong about draw fraud
The most common mistake we see is treating draw fraud as a photo problem. It is a decisioning problem. Catching one fake photo only works if the decisioning layer routes the draw correctly, blocks the wire, and produces an audit trail you can rely on six months later.
The second most common mistake is trusting a third-party inspection as the fraud control. An inspector who shows up confirms the inspector was on site. It does not confirm the borrower's submitted evidence is real. The two are independent controls.
What an effective control stack looks like
For a private lender funding construction or rehab draws, the effective stack is:
- Capture-time verification — fraud signals run on the borrower's upload, not after the fact.
- Scope-aware decisioning — every draw scored against scope, sequence, budget, and prior draws.
- Reviewer evidence panel — verdicts surfaced with the underlying signal, not as a black-box score.
- Immutable audit trail — every signal, override, and verdict logged with request IDs.
- Optional in-person inspection — used as an exception control on flagged draws, not as the default.
How LendlyX fits
LendlyX runs the first four layers as fraud detection infrastructure for private real estate lenders. We integrate with your existing LOS and your existing inspectors — we don't replace either. We replace the unverified-photo, manual-review, hope-for-the-best model that draws on most private lending desks today.
Where to go next
- Private Lending Fraud Detection — the full fraud category and what we cover.
- Draw inspection software — how software-led inspections compare to in-person.
- For Lenders — verification, draw decisioning, and the reviewer workflow.
