Most of a proof of cash is unremarkable. A deposit matches an invoice, a payment matches a bill, and the work is volume rather than difficulty.
The findings live in four places, and each one can be wrong in both directions — too aggressive and you erase real cash, too permissive and you invent it.
1. Inter-account transfers
A business with six bank accounts moves money between them constantly. Operating to payroll on Thursday. Savings to operating when a large payment clears. Location account up to a concentration account nightly. None of that is revenue and none of it is expense. But on the statements it looks identical to both: a credit here, a debit there.
Getting it wrong upward: miss the pairing and every internal transfer becomes revenue. On a business sweeping to a concentration account daily, that can inflate apparent receipts by a multiple. This is the most common way an unreviewed reconciliation produces a number that is not just wrong but absurd.
Getting it wrong downward: match too eagerly on amount and date and you will collapse a genuine customer payment against a genuine vendor payment that happened to be the same size on the same day. Real cash movement disappears.
The discipline is that a transfer requires both legs to be identified, in accounts known to be controlled by the same party, with the decision recorded and reversible. “It looked like a transfer” is not a finding.
2. Owner and related-party activity
Money that goes out to the owner and comes back is not a loan repayment cycle unless somebody says it is. Distributions, contributions, personal expenses run through the business, rent paid to an entity the seller owns, a family member on payroll who does not work there.
None of this is necessarily improper, and most of it is the raw material of the add-back schedule. But it has to be identified as related-party before anyone can argue about whether it normalizes. A proof of cash that quietly nets owner draws against owner contributions has removed the very thing the buyer is paying to see.
3. Merchant processor settlement
For any business taking cards — retail, restaurants, convenience, dental, veterinary, most consumer services — the deposit will never equal the sale. Three reasons, compounding:
- Fees come out first. The deposit is net. Gross revenue is the deposit plus the discount rate, and the discount rate varies by card type.
- Settlement lags. Friday’s sales arrive Monday or Tuesday. At a month boundary that shifts revenue across the cut-off.
- Batching. Several days can arrive as one deposit, or one day can split across two.
Reconciling to gross revenue requires the processor’s settlement detail. Sellers frequently do not have it to hand, and a diligence process that accepts net deposits as revenue will understate the top line and misstate the margin. At a month end it will also move revenue into the wrong period, which is how a business appears to be growing when it is not.
4. Cash-basis timing and the cut-off
A large share of lower middle market businesses run cash basis or a hybrid, with a CPA posting accruals once, in December. That produces two problems in the same file. Within the year, revenue and expense sit in the month money moved rather than the month work happened. At year end, a single adjusting entry reshapes twelve months at once.
This is why a cumulative reconciliation over the diligence period is close to useless. Total it all up and everything ties. Break it by month and you can see the December that absorbed the year, the January that was empty, and whether the seller’s trailing twelve months conveniently begins the month after a bad quarter.
What this means for tooling
Every one of these four is a place where a high automated match rate is actively misleading. A system that pairs on amount and proximity will report beautiful numbers while collapsing transfers it should not have and inventing revenue it should not have.
The useful tests are not about match rate:
- Can you see which items were treated as transfers, and undo it?
- Is related-party activity flagged as such rather than netted away?
- Does the output reconcile to gross revenue, or to net deposits?
- Are proofs presented per account, per month, or only in total?
Our platform matched 99.4% of 29,823 transactions across 36 bank accounts on a recent trailing-twelve-month engagement. The number we point to is not that one. It is the $32 difference at a single location that was still flagged at the end — because a system that resolves everything has not found anything.
More on what to test before you buy: how to evaluate proof of cash software. For the SBA context, see what an SBA cash proof must contain.