Ask a software platform how its payments program is doing and you usually hear a version of the same answer: fine. Revenue is growing, merchants are live, the processor sends a report every month, and the number at the bottom looks reasonable. So nobody looks harder.
Then we reconcile the actual files. Across the platforms Straata has audited, roughly half were not being paid correctly by their processing partner. Not underperforming. Not leaving optimization on the table. Simply not receiving what their own contract says they are owed.
Half of platforms are not being paid correctly. That is before you ask whether they are being paid well.
Half of platforms are not being paid what they are owed.
This is not a story about aggressive optimization or exotic levers. It is a story about arithmetic. A residual or revenue-share payment is supposed to follow a contract: a defined split, on defined volume, net of defined costs. When Straata rebuilds that math from the raw files, the paid number and the owed number frequently do not agree.
The gaps are rarely dramatic on any single line. A few basis points here. A fee category that should have been shared and was not. A tier that was supposed to reprice on a volume threshold and never did. Individually they are invisible. Summed across a book over a year, they are real money.
None of it shows up on the report the processor sends, because that report is the processor’s version of the math, not an independent check of it.
Correct has a definition. Most platforms have never checked it.
Being paid correctly is a testable claim, not a feeling. It means:
- The residual or revenue share matches the contract’s split, on the actual settled volume, to the cent.
- Every fee the contract entitles you to share in is actually being shared, and every cost deducted is one the contract allows.
- The repricing tiers, volume thresholds, and rate steps you negotiated actually fired when the conditions were met.
- Transactions qualified at the interchange level they were eligible for, rather than silently downgrading.
Each of these is verifiable against the raw data. None of them is verifiable against the summary the processor hands you.
Confirming you were paid correctly means rebuilding that calculation from the transaction and contract evidence, independently. The statement and the proof are not the same document.
The system is built so you cannot check the math yourself.
Three things keep this invisible.
Fee complexity
A single statement can carry 15 to 30 distinct fee lines across interchange, assessments, and processor markup, and the US interchange system alone runs to roughly a thousand rate permutations. Verifying it by hand is not a spreadsheet task. It is a data-engineering task.
Single-tenant reporting
Each processor reports only its own slice, in its own format. A platform running two or more processors has no common view, and no processor has any incentive to build one that shows where it is underpaying.
No independent baseline
Even a diligent finance team, handed clean numbers, has no reference for what correct looks like, because the reference lives in the contract and the raw settlement data, not the summary on top of it.
Being paid correctly is the floor. It is not the ceiling.
Suppose you clear that bar. Every dollar the contract owes you arrives. You are still only at the starting line, because correct is not the same as competitive.
The question that decides what a payments program actually earns is a performance question, and it is measurable. Straata benchmarks it against a defined set of metrics:
- Take rate — what you actually earn on the volume you process.
- Effective partner cost — what the relationship costs you, all-in, in basis points.
- Attachment rate — how much of your addressable base is actually on payments.
- Pass-through cost and average merchant fee — whether your pricing and cost structure are competitive for your size and vertical.
For each, the relevant comparison is not last month. It is where you sit against comparable platforms of similar size in the same vertical: below market, at market, or above it. Most platforms have never seen that comparison, because they have never had a benchmark set to see it against.
There are two ways to leave money on the table. Most platforms are doing both.
The first is accuracy: money you are already owed and not receiving. That is a recovery problem, and it is often the fastest to fix, because the contract is on your side.
The second is performance: money you could be earning if your take rate, attachment, and cost structure were at market-leading levels rather than wherever they happen to be. That is an optimization problem, and it is usually the larger of the two.
A platform that has only ever seen the processor’s monthly summary cannot tell the two apart, or even confirm it has the first one under control. That is the gap a diagnostic is built to close: reconcile the raw files to confirm you are being paid correctly, then benchmark the program against the metrics that decide whether you are being paid well. The Report Card grades both.
