Source-linked underwriting: why every figure needs an audit trail

True revenue is not total deposits

The most common source problem in bank statement analysis is also the most basic: gross deposits are not revenue. A busy business account mixes customer receipts with transfers between the business's own accounts, financing proceeds, refunds and reversals, and the occasional owner contribution. Count all of it and the numbers flatter the deal; a merchant who just took a second advance can look more profitable the day the new funding lands, which is exactly backwards.

In an illustrative book (one deal's complete file), June deposits of $128,450 less $15,610 in non-revenue items leaves $112,840 of true revenue across 248 transactions. Each of those numbers should carry its evidence: the statement period, the document it came from, and the deposits that were excluded along the way. Keep the document name, statement period, and relevant page or transaction close to each finding so the original context is easy to revisit.

Exclusions are where judgment enters, so they deserve the most visibility. A transfer between the merchant's operating and savings accounts is not revenue, and neither is the deposit of a new advance or loan. When items like these are set aside, the exclusion list is part of the answer. A reviewer who can see what made up the $15,610 can agree, or push back, in one look instead of rebuilding the month from scratch.

Every figure keeps its source

The same standard applies beyond revenue. Average daily balance, negative balance days, and the recurring debits that signal existing positions each feed a screening rule or a conversation with a funding partner, so each needs the same treatment. With positions especially, the evidence is the ACH descriptor itself: the funder name and cadence sitting in the transaction list, which is often the first sign of stacking before it shows up anywhere else. Context should travel with the work, too: a finding shared with a colleague should carry its proof with it, instead of sending them back to the inbox or the original PDF. If a screening rule flags a deal over monthly revenue or negative balance days, the reviewer should be able to see which transactions drove the result.

This matters beyond your own desk. Funding partners and auditors do not just ask what the number was; they ask how it was reached. A figure with its sources attached answers that question while the deal is still moving. A figure without one sends the whole file back for rework, usually at the worst possible moment.

Unclear extractions get flagged, not guessed

Some documents are incomplete, scanned poorly, or simply ambiguous. A deposit can look like revenue until someone recognizes it as a transfer between the business's own accounts, and a blurred line on a statement can flip a balance by a digit. Treat unclear extraction results as questions for review rather than filling in the gaps, and keep the exception visible where the next underwriter will actually see it. The goal is not to slow the deal down; it is to make sure the slow part happens once, in the open.

A visible flag gives a reviewer something specific to check, which is far more useful than a clean-looking number nobody verified. The ambiguous cases are usually the same handful: a round-number deposit that could be customer revenue or a capital injection, a statement with pages missing, a credit line that appears on the report but nowhere in the bank feed. Each is resolvable, but only if the question survives long enough to reach someone who can answer it.

Naming what is unclear is faster than discovering it later, and it keeps small ambiguities from becoming assumptions baked into the final decision. It is also what lets a deal survive scrutiny weeks later, when a funding partner asks how a number was reached.

A review that stays traceable, step by step

Put into practice, a source-linked review follows the same shape on every deal:

1. Pick the figure to verify: true revenue, average daily balance, debt payments, or a flagged screening result.
2. Open the source document behind it and confirm the statement period and page.
3. Check the exclusions: which deposits or transactions were set aside, and why.
4. Review recurring debits. Fixed daily or weekly payments to a funding company usually mean an existing position; a new one appearing mid-statement is worth a stacking question.
5. Flag anything ambiguous instead of guessing, and leave the flag where the next reviewer will see it.
6. Record the conclusion next to the source, so the file keeps both the answer and its evidence.

None of this adds much time. It removes the second pass, the one that happens later, under pressure, when a number no longer has its paperwork.

Data, results, and judgment stay separate

Sound underwriting keeps three layers distinct: source information pulled from the documents, calculated results like true revenue or average daily balance, and the interpretation an underwriter writes on top. That separation helps reviewers understand how a summary was assembled and where their judgment is needed. A clear explanation should make the underlying documents easier to explore, not harder.

When each layer can be inspected on its own, the team can agree with a conclusion, or see exactly where it needs a second look. Interpretation still matters; it just needs to be labeled as judgment instead of blended into the data underneath it. The underwriter stays the decision maker, and the sources stay one click away, ready to defend the call.

How this looks in Lendquo

Every deal in Lendquo is built on a source-linked bank sheet, where each figure traces back to the statement and transactions behind it, and ambiguous results can be flagged for your team to review instead of being guessed. When a question comes up, your team can ask questions of a deal in plain English and get an answer grounded in the same documents, with the sources still attached. The result is a book that can be reviewed, questioned, and approved from one record.