Managing interchange and merchant fees. 6 chapters and a final quiz.

Chapter 1. Building your cost base.

The cost base is the transaction dataset a merchant uses to negotiate acquiring, plus the handful of metrics derived from it. The payments manager assembles it before the first meeting, from an extract supplied by the incumbent acquirer. A rolling 12 months is the minimum, and both parties must be able to verify every line. A negotiation built on a headline rate and a hunch ends in a cosmetic discount, calculated on a card mix nobody has measured. You build the cost base once, then refresh it every quarter, because a card mix shifts faster than a contract.

What to ask your acquirer for, and in what format

• A transaction-level extract, not a summary report: a monthly aggregate blends together exactly the dimensions you are trying to see
• A rolling 12 months at minimum, to absorb seasonality and promotional campaigns
• A machine-readable format: CSV or a standard settlement file, never a 30-page PDF
• The unique transaction identifier (ARN or its local equivalent), the only way to reconcile the statement, the batch log, and the remittance advice
• The card’s country of issue, derived from the BIN by the acquirer: this is what moves a line inside or outside a cap
• The exact card product (consumer debit, consumer credit, commercial, prepaid), not just the brand
• The authorization result, declines included: a declined transaction still costs money and shows up in no fee rate
• The interchange program code applied by the network, when the acquirer reports it: it names the rate schedule actually used

-- Minimum columns expected in the extract:
--   arn, processing_date, gross_amount, currency, brand, card_product,
--   issuer_country, channel, auth_result,
--   interchange, scheme_fees, margin, ancillary_fees

-- 1. Overall effective MDR, in basis points
SELECT ROUND(10000.0 * SUM(interchange + scheme_fees + margin + ancillary_fees)
             / SUM(gross_amount), 1) AS effective_mdr_bp
FROM transactions
WHERE auth_result = 'CAPTURED';

-- 2. The same calculation by segment: this is the table you bring to the negotiation
SELECT card_product, issuer_country, channel,
       COUNT(*)          AS txn_count,
       SUM(gross_amount) AS gross,
       ROUND(10000.0 * SUM(interchange)  / SUM(gross_amount), 1) AS interchange_bp,
       ROUND(10000.0 * SUM(scheme_fees)  / SUM(gross_amount), 1) AS scheme_bp,
       ROUND(10000.0 * SUM(margin)       / SUM(gross_amount), 1) AS margin_bp
FROM transactions
WHERE auth_result = 'CAPTURED'
GROUP BY card_product, issuer_country, channel
ORDER BY gross DESC;

-- 3. The cost nobody counts: declined attempts
SELECT COUNT(*)            AS declines,
       SUM(ancillary_fees) AS fees_billed_on_declines
FROM transactions
WHERE auth_result = 'DECLINED';

The effective MDR is total payment-related fees divided by gross volume collected over the same period. It is the only figure that compares across offers, provided both calculations use the same scope in the numerator. Acquirers calculate it differently. Dispute fees belong in the numerator, and so do monthly minimums. Express the result in basis points, never as a percentage rounded to two decimal places, because rounding erases the very order of magnitude you are trying to measure. A 10-basis-point gap on 50 million collected is worth 50,000 currency units a year, and that calculation alone tells you how much effort the negotiation deserves.

Chapter 2. Unpacking a line and tracing the gross-to-net gap.

Unpacking a line means breaking down the gap between the amount submitted for settlement and the amount received in the bank account. The exercise rests on a reconciliation check: interchange, scheme fees, margin, and ancillary fees, added together, must reproduce that gap to the cent. Whatever part of the gap the sum fails to cover is called the residual. You cannot guess where a residual comes from by its size. Have the acquirer name it in writing, line by line.

LINE A — e-commerce, consumer debit card issued in the acquiring region
  gross submitted .............. 100.00
  interchange billed ........... 0.20         20.0 bp
  scheme fees billed ........... 0.06          6.0 bp
  acquirer margin .............. 0.25         25.0 bp
  total billed ................. 0.51         51.0 bp
  net settled .................. 99.49
  CHECK: 100.00 - 99.49 = 0.51           the gap is fully explained

LINE B — same transaction, same acquirer, blended contract at 1.10%
  gross submitted .............. 100.00
  fee billed ................... 1.10        110.0 bp
  net settled .................. 98.90
  CHECK: the gap is explained, but NOTHING can be verified
  Implied margin = 110.0 - (20.0 + 6.0) = 84.0 bp
  ... valid only if the segment's actual interchange and scheme
      fees really match line A. This is an estimate, not a reading.

LINE C — the line that triggers a written request
  gross submitted .............. 100.00
  interchange + scheme + margin . 0.51
  net settled .................. 99.31
  UNEXPLAINED RESIDUAL ......... 0.18         18.0 bp outside the breakdown
  Possible causes: cross-border fee, network program fee,
  dispute adjustment, carryover from a prior month. Four causes,
  four contacts, four fixes. None can be inferred from the amount.

A basis point is one hundredth of a percentage point, and it is the unit for measuring the components of a fee. A rounded percentage hides gaps worth tens of thousands of currency units a year on ordinary volume. Convert at extraction: divide the fee amount by the gross amount, then multiply by 10,000. A pivot table then produces the same measure by brand, by card product, and by channel, which exposes the segments where the margin drifts. Run over 12 months, the exercise reveals what no single monthly statement shows: a margin that varies from one segment to the next, with some of those gaps explained by no rate schedule constraint at all. That table is the exhibit you bring to the negotiation, because the monthly statement shows none of this dispersion.

Extracting the implied margin from a blended contract

• Isolate a homogeneous segment in the extract: one brand, one card product, one country of issue, one channel.
• Look up that segment’s interchange rate in the network’s published schedule, or in the legal cap where one exists.
• Estimate the segment’s scheme fees from an unblended quote obtained elsewhere, or from an interchange++ statement the same acquirer issues for another scope.
• Subtract those two components from the blended rate charged: what remains is the acquirer’s implied margin on that segment.
• Repeat segment by segment. The implied margin on a blended contract varies widely across segments, and that dispersion is what you negotiate.

Chapter 3. Finding the regime that applies to each transaction.

An interchange cap applies to a transaction, not to the merchant accepting it. Two identical sales, made in the same minute on the same website, can therefore fall under two different regimes and carry completely unrelated interchange. The applicable regime depends on the card itself: where it was issued, which product it is, and which network it runs on. Neither the merchant’s country nor the cart size plays any part. So the sorting happens line by line, in the extract.

• Count the lines with no usable country of issue: above a few percent, no cap check is possible, and the acquirer should be told in writing.
• Recalculate interchange on a sample of 100 capped lines: the rate must come out at or below the cap.
• Measure the share of commercial cards: it is the first explanation for an effective MDR higher than the quoted rate suggested.
• Track scheme fees as a monthly series, in basis points: the UK’s Payment Systems Regulator found that “core” scheme and processing fees have risen at least 25% in real terms since 2017, costing at least £170 million a year (PSR, MR22/1.10, March 6, 2025).
• Date every rate schedule change: a regulatory reduction has an effective date, and you verify it was applied on the following month’s statement.

The country of issue is often missing from the first file you receive. Ask for the issuer’s country, derived from the BIN, reported per transaction and tied to the line’s unique identifier. The acquirer has this data, since it uses it to apply the network’s rate schedules. A refusal to provide it is therefore not a technical constraint, and it tells you something about the state of the commercial relationship.

Segmentation produces a cost map. A single merchant often sees four regimes coexist in one month: intra-region consumer cards at the cap, commercial cards outside the cap, inter-regional cards under the antitrust schedule, and prepaid under the local rule. No rate negotiation reduces the last three. Reducing them depends on the mix, and so on acceptance, routing, and the checkout flow. The line between what you negotiate through the rate and what you reduce through the mix sets the order of the work.

Chapter 4. Setting up least-cost routing.

Routing is the choice of which network carries a card transaction. It is configured both at the acquirer and on the merchant’s payment gateway. Unless the merchant asks, the acquirer’s default setting applies, and that default was not chosen with the merchant’s interests in mind. Three families of regimes coexist worldwide, distinguished by who makes the choice: the cardholder, the acquirer, or the issuer when the card is issued. The action to take depends on the family, and it changes from market to market for the same company.

Routing decisions turn on three variables. The first is the total cost of the transaction, network fees included. The second is the authorization rate, which varies from one network to another on the same cards and by time of day. The third is how fraud is handled, with the burden of proof and the dispute rules that come with it. The cheapest network is therefore not always the right one, and a test that measures only cost will always pick it. Lost approvals show up six months later, in a decline rate nobody connects to the routing change anymore. A 10-basis-point gain wiped out by one lost point of authorization rate is a net loss.

The switchover protocol, in six steps

• Measure the baseline over four full weeks: total cost by brand, authorization rate by brand, fraud rate by brand.
• Turn on the new routing for a representative subset: one region of stores, or one BIN range online.
• Run the test for at least four weeks, with no other configuration change in that scope.
• Compare all three variables before and after, on the same scope, not just the headline rate.
• Read the brand mix on the following month’s statement: that is the proof the routing actually took effect.
• Roll it out, then return the measurement to a quarterly cycle, because a rate schedule revision can reverse the trade-off.

Three written requests frame the whole process. The first asks for the routing rule actually applied, brand by brand and channel by channel. The second asks for the configured order of preference, with a list of exceptions and the reason for each. The third asks for the emergency failover procedure that applies when a network goes down at peak time. A verbal answer leaves no record, and it cannot be produced if the settings are disputed later.

Chapter 5. Surcharging where allowed, steering everywhere else.

A surcharge is an extra amount a merchant charges a customer for paying with a particular instrument. It is one of three levers on the cost of acceptance. Steering shifts volume to a cheaper payment method, and a minimum amount keeps small purchases off an instrument whose fixed fee makes them unprofitable. Surcharging is banned in much of the world. Whether it is legal depends on the market, the card product, and sometimes the state where the customer lives, so check legality before running any numbers.

• Never exceed the cost actually incurred on the surcharged instrument: the network cap is a ceiling, not a flat fee to apply.
• Disclose the amount before the customer pays, on screen or in store, not at the confirmation step.
• Show the surcharge as a separate line on the receipt and the invoice, so it remains identifiable if there is a refund.
• Refund the surcharge pro rata on a partial refund, or the customer ends up paying a fee on a canceled sale.
• Document how the cost incurred is calculated and update it with every rate schedule revision: this is the evidence a network asks for in an audit.

The cost of a surcharge shows up in the conversion rate, which it lowers. A customer who finds an extra fee on the last screen sometimes abandons the cart, and abandonment varies widely with the amount shown. Test on a sample before any rollout, with conversion rate as the output metric. The calculation weighs the fees saved against the margin lost on abandoned carts over the same period. A drop in conversion cuts revenue, and that loss appears nowhere on the payment invoice.

Steering without surcharging

Chapter 6. Negotiating with numbers.

An acquiring negotiation means getting several bidders to price the same transaction file. Timing matters as much as the numbers. An acquirer approached three weeks before the contract renewal knows no migration can happen in that window. Leverage depends neither on the volume processed nor on the quality of the relationship. It rests on having a credible alternative, and that credibility comes from a timeline that leaves room to migrate. Twelve weeks is enough, provided the cost base already exists and does not have to be built in a rush.

Putting a number on the ask

INPUT DATA
  V  = annual volume processed
  M  = measured effective MDR, in basis points
  M' = effective MDR of the competing offer, priced ON THE SAME FILE

GROSS ANNUAL GAIN = V x (M - M') / 10,000

Illustrative example: V = 40,000,000; M = 118 bp; M' = 104 bp
  gross annual gain = 40,000,000 x 14 / 10,000 = 56,000

SUBTRACT BEFORE DECIDING
  technical switchover cost .......... integration, testing, re-tokenization
  conversion loss .................... during the ramp-up period
  internal project cost .............. person-days x loaded cost
  exit fees and notice ............... remaining contract term

TAKEAWAY: a gap of a few basis points generally does not pay for a
migration. It pays for a renegotiation, which is not the same project or
the same risk. The number does not tell you what to do; it tells you which
project gets funded.

• Effective MDR over the past 12 months, in basis points, with the calculation method spelled out.
• The mix by card product, showing the share of volume outside every cap.
• The inter-regional share, which measures exposure to the highest rate schedules.
• Unit authorization cost multiplied by the number of attempts, declines included.
• Annual cost of disputes, including cases won.
• Projected volume over three years, with the growth assumption stated: this is the number that buys you the tiers.

Price monitoring continues after signing. Every quarter it repeats the same metrics, with the same method, on the same extract, and comes down to three checks. The first measures drift in scheme fees, in basis points. The second checks that regulatory reductions were passed through on their effective date. The third looks at the mix, whose evolution can make the pricing grid obsolete. A gap caught within the quarter gets fixed by amendment; the same gap discovered at contract renewal will have been paid for three years.