TL;DR
- A card payment fee is four stacked charges, not one: interchange (to the issuer), network assessment/scheme fee (to Visa or Mastercard), gateway fee, and provider markup — only the last two are priced by agreements the merchant actually holds.
- US debit interchange is capped at 21 cents plus 5 basis points for large issuers under Regulation II — but a North Dakota court vacated that cap in August 2025, staying its own ruling pending an unresolved appeal.
- The UK Payment Systems Regulator found scheme and processing fees rose at least 25% in real terms between 2017-2023, costing UK businesses "at least £170 million" more yearly, with "little evidence" of a matching cost increase.
- A blended effective rate can't distinguish a legitimate scheme-fee increase from a card-mix shift or a coding error; only reconstructing expected cost per transaction against the contract in force on the settlement date can.
Short answer
A card payment fee is not one charge. It is four, stacked, and two of them are set by parties who are not in your commercial relationship at all.
Interchange goes to the bank that issued the card. The network assessment goes to Visa or Mastercard. The gateway fee goes to whoever moves the transaction data, and the provider markup goes to your acquirer. Those last two are priced by agreements you actually hold. The first two are pass-through costs your provider does not control and, according to the Payment Systems Regulator in the UK, usually cannot negotiate down either.
This matters because most merchants never see the four layers separately. They see one blended percentage. A blended percentage cannot tell you which layer moved when your costs rise, so it cannot tell you whether the rise was legitimate.
The four layers, and where each one is published
| Layer | Who sets it | Who receives it | Where the number is published |
|---|---|---|---|
| Interchange | The card network's schedule, or regulation where a cap applies | The bank that issued the cardholder's card | Visa and Mastercard publish US interchange schedules. In the US, debit from large issuers is capped by Regulation II (currently under appeal). In the EU, consumer debit and credit are capped by Regulation (EU) 2015/751 |
| Network assessment (called a scheme fee in the UK and EU) | The card network | The card network itself | Published by the networks, but far less completely than interchange |
| Gateway fee | Your gateway provider, under the contract you hold with it | Your gateway provider | Nowhere. No network, regulator or standards body publishes a schedule |
| Provider markup | Your own agreement | Your acquirer or payment service provider | Only your contract. It appears as its own invoice line under interchange-plus and interchange++, and vanishes into the headline rate under blended pricing |
Neither of the two layers you price has a published reference price anywhere. The gateway fee has no schedule to check it against, and the markup exists only in your own contract. On a blended contract the markup is not itemised anywhere; it is absorbed into a single rate you cannot decompose.
Layer one: interchange
Interchange is the fee your acquirer pays the cardholder's issuing bank. Visa is explicit that it is a transfer between financial institutions rather than a merchant charge: "Merchants do not pay interchange reimbursement fees; merchants pay 'merchant discount' to their financial institution." The interchange rate still determines most of what you are billed, because it is the largest input to that merchant discount.
In the United States, interchange splits into two very different populations.
Debit from large issuers is capped. Regulation II caps interchange for issuers with consolidated assets of $10 billion or more. The cap is 21 cents plus 5 basis points of the transaction value, plus a further 1 cent if the issuer certifies compliance with the Federal Reserve's fraud-prevention standards. Visa's own published schedule, with rates effective 18 April 2026, still lists the regulated debit rate as 0.05% + $0.21, with the extra cent available on certification. That figure has not moved since the cap took effect in the fourth quarter of 2011.
Everything else is priced by tier. Debit from smaller, exempt issuers carries no cap and is priced by merchant category. On the same schedule, card-present retail debit from an exempt issuer runs 0.80% + $0.15 and card-not-present runs 1.65% + $0.15. Downgrades cost more, and the schedule prices them as two separate programs: the EIRF downgrade rate runs 1.75% + $0.20, and the Standard rate 1.90% + $0.25. Consumer credit spans a wider band still. A card-present supermarket transaction on the lowest tier runs 1.18% + $0.05, and that rate covers traditional and rewards products rather than every card in the tier; premium consumer cards on the same tier price higher. Any consumer credit transaction that downgrades to non-qualified runs 3.15% + $0.10.
The gap between capped and uncapped is not academic. The Federal Reserve's most recent biennial report, published in December 2025 and covering 2023 data, found the average interchange fee on covered transactions was $0.24 over single-message networks and $0.22 over dual-message networks. The average on transactions exempt from the cap was $0.52. Two debit cards, identical to your customer, more than double the cost to you.
That is the first reason a blended effective rate is a poor diagnostic. Your card mix moves, and your average moves with it, with nothing in your contract having changed. We have written separately on why the effective rate is not evidence.
Outside the US the picture is simpler on paper. Regulation (EU) 2015/751 sets hard ceilings in the legislation itself. Article 3 bars any per-transaction debit interchange fee "of more than 0,2 % of the value of the transaction", and Article 4 sets the consumer credit ceiling at "0,3 % of the value of the transaction". Both have applied since 9 December 2015. Those caps cover consumer cards only. Commercial cards remain uncapped in the EU, which is why European merchant associations have begun lobbying to extend them.
Layer two: the network assessment
The assessment is the network's own fee. Interchange passes through the network to the issuer. The assessment stops at Visa or Mastercard.
It is small per transaction and large in aggregate. The Federal Reserve's report, published in December 2025, put total network fees on debit and prepaid transactions at $12.95 billion in 2023, up from $11.49 billion in 2021. Per transaction that averaged $0.129, which the report expresses as 0.28 percent of average transaction value. Acquirers and merchants carried 64.9 percent of that total, with issuers paying the rest.
Industry reporting places the headline percentage assessments in the same range: roughly 0.14% on Visa credit and 0.13% on Visa debit. Both percentages come from secondary sources rather than a fetchable network schedule, so treat them as indicative. The Federal Reserve's aggregate figure is the one to lean on: network-level fees run in the low tenths of a percent, and merchants and acquirers carry most of them.
The UK and EU call this a scheme fee. The Payment Systems Regulator describes the flow precisely: "Scheme and processing fees flow from merchants to card schemes via acquirers, who provide services to accept and process card payments on behalf of a merchant." The regulator treats scheme fees and processing fees as two related categories with separate definitions, which is worth knowing if you operate on both sides of the Atlantic and are trying to reconcile two invoices that use different vocabulary for overlapping charges.
Scheme fees are also where the sharpest regulatory finding of the last two years sits. In its March 2025 final report, the PSR found that average core fees as a proportion of transaction value had "increased in real terms by at least 25%" between 2017 and 2023, and estimated that UK businesses now pay "at least £170 million" more each year for scheme and processing fees than they did seven years earlier. Its verdict on the cause was blunt: "We have seen little evidence that fee increases are linked directly to any changes in relevant costs."
Layer three: the gateway fee
The gateway fee pays for the technology that transmits payment data securely between your systems and the processor. It typically covers transaction transmission, encryption of card data, and fraud screening. It is billed per transaction, as a flat monthly charge, or as both.
It is also the one layer in the stack that no authority even defines. Regulation II defines a base component. The PSR defines pass-through pricing. No card network, no central bank and no standards body has written down what a gateway fee is or what one should cost. Every definition in circulation is a vendor's own.
Treat that absence as information. When a layer has no published benchmark, your contract is the only reference point you have, and reconstructing what that contract says a transaction should have cost is the only way to check the charge.
Layer four: the provider markup
The markup is your acquirer's own margin, applied on top of interchange and assessments. It is priced by your agreement, and how visible it is depends entirely on your pricing model.
Interchange++ itemises everything. The PSR describes it as pricing "where acquirers bill the merchant exactly what the schemes charge acquirers." Interchange, scheme fee and acquirer margin appear as three separate lines. You can see each layer and check each one.
Blended pricing shows one number. A single rate applies regardless of card type, issuer type or channel, and the provider absorbs the difference between what it is charged and what it bills you. The PSR's own data covers merchants on standard blended contracts "accounting for 23% of transactions but also 95% of merchants" in the UK. The second figure is the one that matters here. Almost every UK merchant is billed this way, whatever share of processed value they represent.
Interchange-plus sits between the two, passing interchange through and adding a fixed markup, without necessarily breaking out the network fee as its own line.
One clarification on the taxonomy. The regulator's own glossary defines four pricing options, not three: IC+ pricing, IC++ pricing, standard pricing (the industry calls it blended), and fixed pricing. IC+ and IC++ differ by one thing, which is whether scheme fees pass through at cost alongside interchange. Anyone describing the market as three models is leaving out fixed pricing.
Which layers can you actually price?
Two of the four are pass-through, and there is direct regulatory evidence on how immovable they are.
The PSR found that acquirers "typically pass-through scheme and processing fees in full to merchants. They therefore have reduced incentives to resist fee increases from the schemes." On whether acquirers can push back at all, the finding was starker: the vast majority of acquirers the regulator engaged with said they "have been unable to negotiate reduced fees in the past five years".
Interchange is set by a published schedule or a regulatory cap. The assessment is set unilaterally by the network, and the acquirer standing between you and that network is rarely able to move it either. The gateway fee is priced by your gateway provider, and fixed by the contract you hold with it. The markup is set by you and your provider together.
The gateway fee and the markup are where commercial attention belongs. On a blended contract, only the gateway fee ever appears as its own line.
The negotiating-power finding above is a UK finding from a UK regulator. The Federal Reserve's biennial reports do not address US acquirer-to-network negotiating power directly, so we are not claiming the UK result transfers wholesale to the US market.
How to tell which layer a discrepancy came from
A rise in your effective rate is a symptom. Attribution is the work. These steps run in order, and each one narrows the field.
- Establish which pricing model you are on. Read the contract, not the invoice. If interchange, scheme fees and margin are not itemised separately, no amount of invoice reading will attribute a change, because the invoice does not carry the attribution.
- Separate regulated debit from everything else. The statutory cap is 21 cents plus 5 basis points, plus up to a further cent where the issuer certifies compliance with the Federal Reserve's fraud-prevention standards. Networks print that same cap differently: one schedule lists 0.05% + $0.21 and carries the extra cent in a footnote, another folds the cent in and prints 0.05% + $0.22. Check which convention your own schedule uses first. A one-cent gap can be that convention and nothing more. What the regulated rate never does is vary by merchant category, so regulated debit volume billed at category rates is a coding error, not a mix shift.
- Rebuild the mix before blaming the rate. Compare card type, issuer type and channel period over period. Exempt debit averaged $0.52 against $0.22 to $0.24 for covered debit in the Federal Reserve's 2023 data. A shift in that ratio alone will move a blended average with no rate change anywhere.
- Check the assessment against volume rather than against transactions. Assessments are largely ad valorem. If assessment charges rose faster than your processed value, the percentage changed, and the network publishes when it changes.
- Isolate the gateway charge. It has no benchmark, so it can only be checked against your own contract and your own transaction count. A per-transaction gateway fee billed against a transaction count that does not match your own records is a discrete, provable error.
- Reprice the transactions and compare. Reconstruct what each transaction should have cost under the contract version in force on the day it settled, and compare that to what was billed. The difference is either zero or it names its own layer.
Step six is the only step that produces evidence rather than suspicion. It is also the step nobody does by hand at volume, which is why the first five steps usually end in a conversation about averages instead of a dispute about a charge.
What changed in 2025 and 2026
Four developments are live. Three are unresolved, and the fourth has only just hardened into binding directions. Any piece that states these numbers as permanent is wrong.
The US debit cap is under legal challenge. On 6 August 2025 a federal district court for the District of North Dakota held that the Federal Reserve had gone beyond its statutory authority when it adopted Regulation II, and vacated the fee-cap standard. The court then stayed that vacatur pending the Board's appeal to the Eighth Circuit, in its own words "to prevent interchange transaction fees from becoming a completely unregulated market". The appeal remains active. The Board filed its opening brief in December 2025, the merchant plaintiff, Corner Post, responded in February 2026, and the Board's reply brief was reported in April 2026. The Eighth Circuit heard oral argument in May 2026 and has not yet ruled. The cap is enforced today, but it is enforced under live litigation rather than as settled law.
The Federal Reserve's own proposal to lower the cap has never been finalised. In October 2023 the Board proposed cutting the base component from 21 cents to 14.4 cents, the ad valorem component from 5 basis points to 4, and raising the fraud-prevention adjustment from 1 cent to 1.3 cents. The Board's Regulation II page still describes this as a proposal. The Visa schedule effective 18 April 2026 still prices the original numbers.
US credit interchange has a preliminary settlement. On 9 June 2026 a federal judge granted preliminary approval to a revised $38 billion Visa and Mastercard antitrust settlement covering more than 12 million US merchants, under which swipe fees would fall by 0.1 percentage points for five years and the standard consumer card rate would be capped at 1.25% for eight years. Preliminary approval is not final approval. Retailer groups have filed objections and the matter is unresolved.
The UK has stopped consulting and started directing. In July 2026 the PSR published its final decision, PS26/1, and gave each scheme two specific directions under section 54(3)(c) of the Financial Services (Banking Reform) Act 2013. The ITC Direction governs the fee information the schemes must give acquirers, and the Pricing Governance Direction requires them to show the evidence behind their pricing decisions. A third remedy, on regulatory financial reporting, went out for consultation in May 2026 and has not been directed.
The practical consequence is the same in every case. The published rate you verify against is a moving reference, and a fee check is only as good as the contract version and rate schedule in force on the day the transaction settled.
Why standardised statements have existed since 1986 and still do not solve this
None of this is a new complaint. The Association for Financial Professionals dates its own service codes to 1986, and says they "have been recognized as the standard for identifying balances and charges that appear on account analysis statements". That standard carried into the ANSI X12 EDI 822 account analysis transaction set, and onward into the ISO 20022 camt.086 billing message released in July 2012.
Forty years of standards for describing bank charges, and merchants still cannot tell which layer moved. Standards fix the format of the statement. They do not verify the arithmetic inside it. Verification requires the contract, the transaction, and a rebuild of what one should have cost the other.
What you can verify, and what you can only accept
Four layers, four different parties, one number on your invoice.
Interchange is capped only for large-issuer debit in the US and for consumer cards in the EU. Everywhere else it is whatever the network's schedule says. The assessment is the network's own price, and it has been moving: in the UK, average core fees for scheme and processing services rose by at least 25% in real terms as a proportion of transaction value between 2017 and 2023, with, in the regulator's words, little evidence of a matching cost increase. Those two are not yours to price. The other two are: the gateway fee, which has no published benchmark to check it against, and the markup, which on a blended invoice exists only in the contract that set it.
Bluefyn reconstructs contract pricing and checks fees transaction by transaction, so that expected and actual can be compared with evidence rather than estimated from an average. Bluefyn never moves, holds or custodies funds. It analyses transaction and provider data.
An average tells you your costs went up. Only a rebuild tells you which layer did it, and whether it was allowed to.
Frequently asked questions
What are the four layers of payment processing fees?
Interchange, the network assessment, the gateway fee and the provider markup. The issuing bank keeps interchange. The network keeps the assessment, called a scheme fee in the UK and EU, for the use of its rails. Your gateway provider bills the gateway fee for carrying transaction data to the processor, and your acquirer keeps the markup as its margin on top of the other three. Two of the four are priced by contracts you signed: the gateway fee and the markup.
What is the current Regulation II debit interchange cap?
21 cents plus 5 basis points of the transaction value, plus 1 cent where the issuer certifies compliance with the Federal Reserve's fraud-prevention standards. That has been the cap since the fourth quarter of 2011, and Visa's schedule effective 18 April 2026 still prices it. The Federal Reserve proposed lowering it in October 2023 but has never finalised that proposal, and a district court vacated the standard in August 2025 while staying its own ruling pending appeal.
What is the difference between interchange and an assessment fee?
Interchange is paid to the cardholder's issuing bank. The assessment is paid to the card network itself. Interchange is the larger of the two. In the UK and EU the assessment is called a scheme fee, and the Payment Systems Regulator treats scheme fees and processing fees as distinct categories rather than as one charge.
Which payment processing fees can I actually negotiate?
In practice, only your provider's markup and your gateway charge. Interchange is capped by regulation or fixed on a network schedule you have no part in writing. The assessment is the network's own price, and in the UK the Payment Systems Regulator found that the vast majority of the acquirers it engaged with had been unable to negotiate those fees down over a five-year period. That is a UK finding from a UK regulator. It has no US counterpart: the Federal Reserve's biennial reports do not examine acquirer-to-network negotiating power at all.
Why did my effective rate rise when my contract did not change?
Because the effective rate is a volume-weighted average across every interchange tier your transactions touch. A shift in card type, issuer type or channel moves the average on its own. Federal Reserve data for 2023 shows debit from exempt issuers averaging $0.52 against $0.22 to $0.24 for covered debit, so a change in that mix alone is enough.
What is interchange++ pricing?
A model in which the acquirer bills you exactly what the networks charge it, itemising interchange, scheme fee and acquirer margin as three separate lines. Blended pricing collapses all of it into one rate and absorbs the variance internally. It is the only common model under which the acquirer's own margin is a number you can read rather than infer.
