Content authored by practitioners with experience at Amazon India, Intuit QuickBooks, and the Tata Group. Meet the team →
A single blended MDR quote — Razorpay 2%, PayU 2%, Cashfree 1.95%, PhonePe 1.95% — conceals the per-network spread between zero-MDR UPI, ~1.5% Visa/Mastercard credit, and 2.95-3.5% Amex and Diners. The merchant sees one deducted percentage on the settlement file and cannot tell whether Amex and Diners were separately and correctly priced. Either the merchant is cross-subsidising premium-network volume through low-cost UPI, or the gateway is under-recovering on the premium share and reclaiming silently in a later cycle via reclassification, a rate revision, or a renewal true-up.
Per-network effective-rate reconciliation groups every settlement-file transaction by network (Visa debit, Mastercard debit, Visa credit, Mastercard credit, Amex, Diners, UPI bank account, RuPay debit, international), sums the fee deducted and the gross transacted value for each group, and computes fee divided by volume. The effective rate per group is then compared against the gateway's published per-network rate card and against the blended quote the merchant believes is in force. A NETWORK_EFFECTIVE_RATE_GAP variance is raised whenever the per-network effective rate exceeds the contracted blended rate plus a tolerance, with Amex and Diners isolated as a dedicated bucket for separate slab verification.
Per-network MDR rule set keyed on network (Visa, Mastercard, Amex, Diners, RuPay, UPI bank, international) and product tier (consumer, premium, commercial); blended-quote rule recording the contracted blended rate; NETWORK_EFFECTIVE_RATE_GAP variance class with 10-bps slab-delta tolerance; Amex and Diners flagged as a separate audit bucket regardless of share; GST-invoice matcher for the 18% line; refund-MDR retention flag and 90-day rolling recovery window.
A per-network effective-rate table per settlement cycle with fee, volume, effective rate, expected rate, variance basis points, and recoverable amount; a CFO-facing dashboard showing network mix, blended vs effective spread, and Amex/Diners isolated share month-on-month; a dispute-pack export per gateway with per-transaction Amex and Diners evidence and expected-vs-actual fee calculation; a GST-invoice reconciliation schedule against the gateway's tax-invoice line.
A travel OTA aggregator running ₹12 crore monthly card GMV across air, hotel, and package bookings reads a single line on the monthly gateway statement: “MDR — Cards — blended 2.15%.” The CFO knows the gateway publishes 2% for standard credit cards and 3% for Amex and Diners. The reconciliation team accepts the blended deduction, signs off, and moves on. The settlement net matches. The bank reconciliation closes. No exception is raised.
A per-network effective-rate audit on the same settlement file uncovers that Amex and Diners — together 5.5% of card volume — are absorbed into the 2.15% blended rate without separate billing. Visa and Mastercard have actually been billed at 1.95%; Amex effective is 2.85% (close to the 3% list slab, acceptable); Diners effective is 3.10% (above list). The gap appears mild on the surface. It is not. ₹66 lakh of Amex and Diners volume per month is being billed flat instead of at the higher premium slab the merchant contracted for, hiding ₹56,100 per month — ₹6.73 lakh annually — of cross-subsidy inside the blended line. This article walks through the detection technique, the worked example, and the dispute discipline.
Quick-Reference: Amex / Diners in a Blended MDR Setting
| Aspect | Detail |
|---|---|
| Amex domestic published slab | 2.95% to 3.5% — Razorpay, PayU, Cashfree all list 3% |
| Diners domestic published slab | 2.95% to 3.5% — Cashfree explicitly lists 2.95% |
| Visa / Mastercard credit (consumer) published | 1.4% to 2.5% — gateways list ~2% |
| UPI bank-account (network MDR) | 0% — Section 269SU Income-tax Act + Section 10A PSS Act |
| Effective date of zero-MDR on UPI | 1 January 2020 |
| Blended-rate published quotes | Razorpay 2%; PayU 2%; Cashfree 1.95% (1.6% promo); PhonePe 1.95% (“Free” promo) |
| Network type | Amex and Diners are three-party closed-loop; Visa/Mastercard are four-party open-loop |
| Relevant RBI circular | DPSS.CO.PD No.1633 / 02.14.003 / 2017-18 (debit MDR cap only) |
| Credit-card MDR cap (regulatory) | None — uncapped, contractually negotiated |
| GST on the fee | 18%, separate line, ITC-recoverable; never folded into MDR % |
Network rates current to 2026-06-23. Visa/Mastercard credit and Amex/Diners ranges reflect negotiated spreads, not regulated caps. The only RBI-capped instrument is non-RuPay debit.
What does “blended MDR” actually mean, and where does the leakage hide?
A blended rate is a single percentage applied across all qualifying transactions regardless of network. The gateway prices the blend by assuming a network mix — typically a heavy UPI tail, a Visa/Mastercard credit middle, and a small Amex/Diners head. When the rate card publishes a 2% blended figure, the gateway has implicitly priced for, say, 5% Amex and Diners share at ~3%, 60% Visa and Mastercard credit at ~2%, and 35% UPI / RuPay debit at ~0%. Average that on volume and the all-in cost-of-acquiring is somewhere near 1.5-1.7% — and the gateway charges ~2% blended to retain its spread.
The leakage hides in two places. First, where the merchant’s actual Amex and Diners share exceeds the priced share, the gateway is under-recovering on the premium tail. Three contractual mechanisms then come into play: retrospective reclassification (re-categorising a tranche of transactions in a later cycle as premium and billing the differential as a settlement adjustment); a notice-period rate revision invoking the carve-out clause that exists in nearly every gateway agreement; or an opaque true-up at the next contract renewal. The merchant experiences the first as an unexplained settlement adjustment line, the second as a “rate change” letter, and the third as a renewal that comes in much higher than the published headline.
Second, where the merchant’s actual Amex and Diners share is below the priced share, the merchant is cross-subsidising — paying a higher blended rate than its true network mix would justify. A UPI-heavy OTT or D2C merchant on a flat 2% blended quote with effectively zero Amex and Diners share is over-recovering for the gateway, by design.
Neither outcome is visible from the settlement file unless the merchant computes per-network effective rates.
How do you compute a per-network effective rate?
The per-network effective rate is total fees deducted on a network divided by gross transacted volume on that network, for a defined reconciliation period. The computation is mechanical and reproducible from the settlement file as long as the file carries a network column (or a BIN column from which network is derived).
The audit groups the settlement file by network. The relevant groups for an Indian merchant are: UPI bank account (zero expected), RuPay debit (zero expected), Visa debit (capped at 0.40% or 0.90% per the RBI 2017 circular), Mastercard debit (same cap), Visa credit (consumer ~2% contracted), Mastercard credit (consumer ~2% contracted), Amex (premium ~3% contracted), Diners (premium ~3% contracted), international card (2.69% to 3.5% plus forex). For each group, the audit sums the gross transacted value and the total fee deducted, computes fee divided by volume to get the effective rate in percentage terms, and compares that effective rate against (a) the published per-network rate the merchant believes the contract specifies, and (b) the blended quote the merchant believes is in force.
A network whose effective rate equals the blended rate has been priced flat. For Visa and Mastercard credit at the consumer slab this is acceptable when the consumer slab and the blended quote are aligned. For Amex and Diners it is almost never acceptable — their list rate is 2.95-3.5% across every Indian gateway, materially above any blended quote sub-2.5%. A flat-priced Amex or Diners line is therefore the most reliable single leakage signal in the entire merchant-fee dataset.
Worked example: a travel OTA aggregator at ₹12 Cr monthly card GMV
A multi-product travel OTA aggregator processing flight, hotel, and package bookings carries ₹12 crore of monthly card GMV through its primary gateway. The card mix is 4% Amex, 1.5% Diners, and 94.5% Visa/Mastercard. The contracted rate card specifies 2% on standard credit, 3% on Amex and Diners, and a blended 2.15% headline figure the merchant uses for monthly forecasting. The gateway has deducted 2.15% blended across the entire card volume for three consecutive months. Settlement net matches. No exception is raised.
The audit runs per-network effective rates on a 90-day window. The Visa/Mastercard effective rate computes at 1.95% — five basis points below the contracted 2% standard slab. Acceptable; the gateway is delivering on the consumer credit line. The Amex effective rate computes at 2.85% — close to the 3% list slab and within the 10-bps audit tolerance. Acceptable. The Diners effective rate computes at 3.10% — ten basis points above list, marginal but within tolerance for a single cycle. The critical finding is not in the rate column — it is in the carve-out test.
Amex and Diners are not separately invoiced. The settlement file shows them inside the blended 2.15% deduction line. Visa and Mastercard credit have actually been billed at 1.95%. Amex and Diners volume — ₹48 lakh Amex plus ₹18 lakh Diners equals ₹66 lakh per month — has been billed at the same 1.95% Visa/Mastercard effective rate that everyone else is on, not at the 2.80% effective rate the merchant’s own contract specifies as the Amex/Diners weighted slab. The 0.85 percentage-point gap on ₹66 lakh works out to ₹56,100 per month of fee that should have been deducted at a premium slab but was absorbed into the blended line.
₹56,100 per month is ₹6.73 lakh annually. The leakage is not the gateway over-charging — the gateway has under-recovered on premium volume by ₹56,100 per month and is sitting on the right to reclaim that through any of the three contractual levers above. The OTA’s exposure is the unannounced rate revision letter or the reclassification adjustment that will eventually arrive, at a moment the CFO has not budgeted for it. The right action is to renegotiate to a transparent per-network rate card today — at known cost — rather than absorb a back-dated reclassification at unknown cost later.
Detection technique: the per-network effective-rate audit, step by step
The audit runs monthly on the gateway settlement file. The data needed is a per-transaction grain with at least transaction-amount, network or BIN, fee-deducted, and refund flag. Most large gateways expose this in the settlement export; smaller gateways may need a contractual request for per-transaction fee itemisation.
First, group the file by network. Where the file does not carry a network column, derive network from the first six digits of the BIN against a network identifier table (Visa starts with 4, Mastercard with 51-55, Amex with 34 or 37, Diners with 30, 36, or 38, RuPay starts with 60 / 65 / 81 / 82, JCB with 35). For UPI, derive from the instrument code in the settlement file.
Second, sum gross transacted volume and total fee deducted per network. Exclude refunded transactions from the volume base if the gateway has retained MDR on refunds (a separate leakage pattern documented in the cluster — refund MDR retention is industry-standard and not by itself a defect, but should be tracked separately from the network audit).
Third, compute effective rate per network as fee divided by volume in percentage terms. Round to two decimal places — basis-point precision is appropriate at audit grain.
Fourth, compare each network’s effective rate against the published rate card the merchant has signed for that network. The minimum benchmark is the RBI 2017 circular cap for non-RuPay debit (0.40% / 0.90% with per-transaction ₹200 / ₹1,000 caps) and the zero-MDR mandate for UPI bank account and RuPay debit. The merchant-specific benchmark is its contracted per-network slab for credit, Amex, Diners, and international.
Fifth — and this is the Amex/Diners-specific check — verify that Amex and Diners appear as separate effective-rate lines and are not absorbed into a blended figure. If their effective rate matches the Visa/Mastercard credit effective rate to within 10 basis points, the gateway is not separately pricing them — flag the contract for renegotiation regardless of the cycle’s direction.
Sixth, compute the monthly recoverable or exposed amount: Amex and Diners volume multiplied by the gap between the effective rate billed and the contracted premium slab.
Run the per-network effective-rate audit on your own settlement file
Enter your monthly card GMV, the network mix (Amex / Diners / Visa / Mastercard / UPI share), and the blended rate the gateway is currently deducting. The calculator returns the effective rate per network, the Amex/Diners cross-subsidy or under-recovery, and the annualised leakage figure your CFO needs to bring to the gateway renegotiation conversation.
Open the MDR Effective-Rate Calculator →What does a clean per-network rate card look like, and how do you negotiate to it?
A clean rate card lists every network the merchant accepts as a separate line item — Visa debit, Mastercard debit, RuPay debit, Visa credit consumer, Visa credit premium, Mastercard credit consumer, Mastercard credit premium, Amex, Diners, RuPay credit on UPI, PPI on UPI, UPI bank account, international Visa/Mastercard, international Amex — with a single percentage against each. A blended rate may still exist as a forecasting convenience but the per-network rates are the contractual truth.
The negotiation lever is the audit itself. Once the per-network effective-rate table is in hand, the merchant has three positions to argue from. First, the Amex/Diners volume is small enough relative to total card volume that pricing it at list (3%) and the rest at the existing blended slab will leave the merchant’s total fee essentially unchanged — the gateway loses nothing by switching to a transparent card, and the merchant gains audit visibility plus protection against future reclassification. Second, where the Visa/Mastercard effective rate is below the contracted consumer slab (as in the worked example, 1.95% vs 2% contracted), the merchant has evidence the gateway is willing to underwrite the consumer line below contract — there is room to formalise that as a renegotiated consumer-credit slab. Third, the merchant carries a credible threat to route Amex and Diners through a different acquirer that does itemise the premium slab — most large gateways will concede on transparency before they concede on price.
The dispute discipline is separate from the negotiation. Where the audit surfaces a Diners effective rate above list (3.10% vs 3% list, in the worked example), or any network’s effective rate above its contracted slab by more than the 10-bps audit tolerance, raise a dispute against the specific settlement cycle within 90 days. Gateways generally process retrospective adjustments where the merchant produces the per-transaction evidence and the contracted rate card — the 90-day window matches the typical acquirer settlement cut-off, beyond which the gateway cannot itself recover from the network.
Continue reading in this cluster
This article is part of the eight-pattern merchant-fee leakage series. The companion pieces extend the per-network discipline to adjacent slabs and recovery scenarios:
- MDR fee reconciliation against contracted gateway rates — the cornerstone framework for reconciling deducted MDR against the contracted rate card per cycle.
- Premium card misrouting to the 3% slab — BIN audit — Pattern #4: signature, infinite, and rewards cards auto-routed to the premium slab without per-transaction itemisation, and a 90-day BIN audit recovery.
- RuPay credit-on-UPI 2% surcharge above ₹2,000 — Pattern #2: the high-cost cell disguised as “UPI” — RuPay credit on UPI carries ~2% interchange above the ₹2,000 ticket break, materially different from bank-account UPI.
For the full cluster index see the merchant-fees insight hub and the payment gateway reconciliation money page for the broader settlement-reconciliation framework that this audit fits into.