Inside the Benchmarking Engine: How AIME Decides a Fee Is Too High

Anyone can tell you your fees are too high. It is the easiest sales claim in the payments industry and it costs nothing to make. The hard part, and the only part a vendor will actually respond to, is saying which line is too high, by how much, and against what.

That is the job of the benchmarking layer. Once a statement has been turned into a clean ledger, every line goes through the same three questions.

Question one: is this even yours to negotiate?

The first split is between cost and margin, and it decides everything downstream.

• Interchange is set by the card networks and paid to the card-issuing bank. Your processor collects it and passes it on. It is a real cost and it is not negotiable line by line, although the rate you qualify for very much is. Interchange, Explained walks the mechanics.
• Assessments and network fees are the networks' own take. Also pass-through, also fixed, and also a favorite place to pad, because almost nobody checks a published table against a statement.
• Processor markup is the part your provider keeps. Every basis point of it is negotiable, and it is where most of the recoverable money sits.
• Ancillary fees are the monthly and per-event charges: statement fees, batch fees, compliance programs, gateway fees, equipment rental.

Telling a merchant their total cost is 2.9 percent is not useful. Telling them 2.1 points of it is network cost and 0.8 is markup, against a band of 0.25 to 0.45 for their profile, is a negotiating position.

Question two: what should this line cost?

Each line is compared against a benchmark band rather than a single number, and the band is chosen from the business the statement belongs to, not from an average of everybody.

A card-present restaurant with a 40 dollar average ticket and a B2B distributor with an 8,000 dollar average ticket have almost nothing in common in fee terms. Percentage-based costs dominate the first; per-transaction costs are a rounding error there and material for nobody. Benchmarking them against the same number produces confident nonsense in both directions.

Some lines have a hard published reference rather than a band. Network assessments are publishable numbers, so a statement charging above the published rate is not a matter of opinion. Those get flagged with the arithmetic shown, because they are the fastest wins in the entire audit and the least arguable.

Question three: what is actually recoverable?

The recoverable figure is deliberately conservative, and it uses one formula everywhere:

Two consequences follow, and both are on purpose. A line priced below benchmark never contributes a negative number that quietly offsets a genuine overcharge somewhere else. And the total is a floor rather than a projection: it is the money visible on this statement, not an annualized estimate with an optimistic multiplier bolted on.

Total figures are the sum of line-level findings, never a rate applied to volume. If a report claims 14,000 dollars a year, there are lines underneath it that add to 14,000 dollars a year, and each one names the charge, the benchmark and the difference.

Why the working is always shown

A savings report has one real job: to survive contact with the vendor who wrote the statement. That vendor has a relationship manager whose entire task, on receiving your report, is to find the weakest number in it.

So every finding carries its own evidence. Which line on which page. What it was charged. What the benchmark is and why that band applies to this business. What the difference comes to. A finding you cannot defend line by line is worse than no finding, because losing one argument costs you credibility on the twelve that were right.

What benchmarking cannot do

It cannot tell you whether a vendor is worth keeping. A processor charging above-band markup while running flawless settlement for a business with complex needs may be the right processor, and the correct outcome of the audit is a repricing conversation rather than a switch. That judgment is yours. The engine's job is to make sure you make it holding the numbers.

Frequently asked questions

Where do the benchmarks come from?

Published network and interchange tables where a rate is published, carrier and bank fee schedules where those exist, and the distribution of what comparable businesses actually pay across audited statements. Bands widen where the evidence is thinner, which is the honest way to express uncertainty rather than pretending to a precision nobody has.

Does a small business get benchmarked against enterprise pricing?

No, and that would be the fastest way to produce findings nobody can act on. Bands are selected by profile: card presence, average ticket, monthly volume and industry. A five-location restaurant group is compared against businesses that look like it.

Why is my markup high if my effective rate is fine?

Because a favorable interchange mix can hide expensive markup. A business running mostly regulated debit has low network cost, so a fat markup still lands at a respectable-looking blended rate. Separating the two is exactly the point of the first split above.

What happens after the findings?

The report is the input to a conversation with your existing vendor, not a switch order. And because rates drift back, the audit is the start of a monitoring loop rather than an event: see Continuous Monitoring.