Short answer
Credible evidence of a provider overcharge is a claim the provider can check and cannot dismiss. It has four properties: it is deterministic, so there is one correct expected number rather than a model's estimate; it traces to the exact contract clause that priced the charge; it sits at the level of the single transaction rather than a monthly blend; and it is reproducible, so the provider can re-run it and reach the same result. Evidence with all four survives the provider's own re-check, which is the only test that matters. Evidence missing any of them reads as an opinion, and an opinion is what gets a dispute waved away. This is the standard that evidence has to meet, rather than the list of things to gather.
Why most fee complaints fail
Most overcharge complaints fail before anyone looks at the merits, because they arrive as an assertion rather than a proof.
The common one is the effective-rate complaint: your blended cost went from one percentage to a higher one, so something must be wrong. That is a real signal that something changed, and it is not evidence. A blended rate mixes every transaction, every fee type, and every change in your own mix into one number, so a provider can explain the move a dozen ways that have nothing to do with an error. You cannot point to the charge that was wrong, because a blended rate has no single charge in it. The provider agrees the number moved, disagrees that it proves anything, and the conversation ends.
The pattern holds for any evidence a provider can re-explain: a spreadsheet total, a gut sense that fees feel high, a quote from a competitor. None of them name the specific charge, the specific rule it broke, and the amount. Credible evidence does all three, which is what makes it hard to dismiss.
The four properties of credible evidence
It is deterministic
A credible claim has one correct expected number, computed the same way every time. The provider charged a certain amount. Under the contract, the charge should have been a certain other amount. The difference is the discrepancy. There is no range, no estimate, no assumption a reviewer can dispute, because the expected figure is computed from the contract, not modeled. A variance a reviewer can argue with is not proof. A variance produced the same way twice is.
This is the line between evidence and opinion. A model that estimates what you probably should have paid produces a number nobody else can reproduce, so the provider is free to reject it. A deterministic expected charge produces a number the provider can compute themselves and arrive at the same place.
It traces to the contract clause
The expected number has to come from somewhere the provider agreed to. Credible evidence names the specific term that priced the transaction: this card type, under this clause, at this rate. When the evidence points to the exact language in the agreement, the provider cannot claim a different rule applied, because the rule is quoted back to them from their own contract. When it cannot, the dispute becomes a disagreement about which terms govern, which the provider usually wins by default.
Building that link is its own task, because a contract prices conditions rather than transactions. Resolving the clause that applies to a given charge is the work that makes the evidence traceable.
It sits at the transaction level
Credible evidence is about one transaction, or a set of individually named ones, and never an aggregate. A monthly total invites the provider to explain the total. A single transaction, with its expected charge and its actual charge side by side, leaves nothing to reinterpret. Either that charge matched the contract or it did not.
Transaction grain also makes the claim additive in the right way. Ten named transactions, each with its own traceable discrepancy, are ten proofs, and their sum is a number the provider can verify line by line. A single aggregate of the same amount is one assertion the provider can contest as a whole.
It is reproducible and dated
The last property is that someone else can run it and get the same answer. Credible evidence carries its inputs: the transaction, the charge, the contract version in force on that date, and the expected figure that version produces. Given those, the provider re-runs the comparison and lands where you did. The dated contract version matters, because a charge is only right or wrong against the terms that applied when it settled, and a dispute window is measured from that same date. Evidence that cannot be reproduced, or that cannot say which contract version it used, gives the provider room to question the whole basis of the claim.
Credible versus weak evidence
The same suspicion can be presented two ways. One gets a credit. The other gets a reply that thanks you for your feedback. The contrasts in this table are illustrative, meant to show the two forms rather than to state a real rate.
| Property | Weak evidence | Credible evidence |
|---|---|---|
| Basis | A blended effective rate rose over a period | One transaction's actual charge versus its contract-derived expected charge |
| Determinism | A modeled estimate of what you should have paid | A single expected number, computed the same way every time |
| Traceability | "Fees seem high" | Tied to the exact contract clause that priced the transaction |
| Granularity | A monthly or quarterly total | The individual transaction, named |
| Reproducible | The provider cannot re-run your spreadsheet | The provider re-runs the comparison and gets the same result |
| Provider's reply | Explains the change, closes the ticket | Verifies the line, issues the credit |
Why this is the standard that gets a credit
The reason these four properties matter is that the provider applies its own version of them before paying anything. A credit is money leaving the provider's side, so a dispute is reviewed by someone whose job is to reject claims that are not airtight. That reviewer runs one test: can I reproduce this discrepancy against our own records and our own contract, at the transaction level. Evidence built to pass that test gets paid. Evidence built to express a concern does not.
That is the job Bluefyn is built for. Bluefyn reconstructs the contract into an expected charge for every transaction, compares it to what was actually charged, and produces the discrepancy with its clause reference and its inputs attached, so the evidence is deterministic, traceable, per-transaction, and reproducible by design. It analyzes transaction and provider data. It never moves, holds, or custodies funds, and it does not file your dispute for you. It produces the proof that makes the dispute you file hard to refuse. For the steps of running that dispute once you have the evidence, see what to do when your provider overcharges you.
Frequently asked questions
What makes fee-overcharge evidence credible?
Four properties, together. It is deterministic, meaning there is one correct expected charge computed the same way every time rather than a modeled estimate. It traces to the specific contract clause that priced the transaction. It sits at the level of the individual transaction rather than a monthly blend. And it is reproducible, so the provider can re-run the comparison against their own records and reach the same result. Evidence with all four survives the provider's review. Evidence missing any of them reads as an opinion the provider can explain away.
Why is my effective rate not enough to prove an overcharge?
Because a blended effective rate contains no single charge to point to. It mixes every transaction, fee type, and shift in your own volume mix into one percentage, so when it rises the provider can attribute the move to a dozen causes that are not errors. It is a signal worth investigating rather than a proof. To turn the suspicion into evidence, you have to come down to the individual transactions whose charges did not match the contract, and show each one.
What is the difference between evidence and an opinion in a fee dispute?
An opinion is a claim the provider can re-explain. An effective-rate increase, a spreadsheet total, or a sense that fees are high all leave the provider room to offer an innocent account of the number. Evidence is a claim the provider can only verify or refute, not reinterpret: a named transaction, the contract clause that priced it, the expected charge that clause produces, and the actual charge, all reproducible. The test is whether the provider can re-run it and disagree with the method. If they cannot, it is evidence.
Does a provider have to accept transaction-level evidence?
A provider is not obligated to agree with any claim, but transaction-level evidence removes the usual grounds for rejecting one. When the discrepancy is named per transaction, traced to a clause in the provider's own contract, and reproducible against the provider's own records, there is no aggregate to reinterpret and no model to dispute. The provider is left checking whether the charge matched the agreed rule, which is a factual question with one answer. That is why this standard, rather than the strength of the complaint, is what turns a dispute into a credit.
