Chargeback Ratio Denominator Mistakes Merchants Make

Already covered elsewhere — not repeated here Net-of-refunds sales All-network blending Authorised, not settled Dollar-weighted ratios Wrong merchant-ID level The five-minute denominator check FAQ

Already covered elsewhere — not repeated here

To avoid a near-duplicate page, this guide deliberately omits four denominator errors that existing guides already explain in full, and links to them instead:

All sales vs card-not-present settled transactions (Visa VAMP): VAMP's denominator is settled card-not-present transactions only; card-present volume must be split out. Covered in the Visa VAMP threshold guide.
Counting each fraud case once (Visa VAMP): a fraud report that later becomes a dispute enters the numerator twice — once as a TC40 fraud report and again as a TC15 dispute. That double-count is how the formula works, not a spreadsheet error. Covered in the Visa VAMP threshold guide.
Wrong month's transactions (Mastercard): Mastercard divides this month's first-presentment chargebacks by last month's transactions, so a quiet sales month raises next month's official ratio. Covered in the chargeback ratio guide and the Mastercard ECM/HECM guide.
Wins counted as removals: winning a dispute recovers revenue but does not remove the dispute from the month's count — neither network recalculates for representment outcomes. Covered in the winning-does-not-fix-your-ratio guide.

What follows is only the uncovered ground: five errors no existing guide on this site treats as denominator errors.

1. Net-of-refunds sales

The error: dividing by sales minus refunds — "we sold 10,000, refunded 800, so our denominator is 9,200." The networks divide by gross settled transactions, not by sales net of refunds. Subtracting refunds shrinks the denominator and inflates your ratio, so a merchant doing the honest-looking net calculation reports a worse ratio than the network computes — and may spend money fixing a problem that does not exist at that size. Worked example: 120 disputes over 10,000 settled transactions is 1.20%; over 9,200 net-of-refund transactions it is 1.30%. Same disputes, a tenth of a point of phantom ratio. Keep refunds where they belong: refunds are a customer-service and revenue number, and a refund issued before a dispute posts keeps an event out of the numerator — but a refund never enters the denominator arithmetic. (The related point that refunding after a dispute posts does not remove it is covered in the VAMP guide; this section is only about the denominator side.)

2. All-network blending

The error: one denominator for all card brands — "we processed 20,000 card transactions, so our ratio is disputes over 20,000." Each network measures only its own volume. Visa's VAMP denominator is Visa settled card-not-present transactions; Mastercard's ECM denominator is Mastercard transactions from the prior month. Blending all brands into one denominator dilutes the ratio for the network that is actually in trouble and hides the brand where the problem sits. Worked example: a merchant with 120 Visa disputes on 10,000 Visa transactions (1.20% Visa basis) and 5,000 Mastercard transactions with no disputes reports 120 ÷ 15,000 = 0.80% blended — comfortably under every line, while the Visa-basis figure is the one Visa actually scores. Always compute one ratio per network, on that network's transactions only, and never present a blended figure to an acquirer as if it were a program ratio.

3. Authorised, not settled

The error: counting authorisations, attempts or orders as transactions. The denominators are settled (cleared) transactions — Visa's TC05 is settled card-not-present volume, and Mastercard's denominator is processed transactions. Authorisations that were declined, voided before capture, or never captured never settle, so they never enter the denominator; counting them inflates the denominator and flatters the ratio. This bites hardest for merchants with high decline rates or pre-authorisation flows (hotels, rentals, trial checkouts): a 15% decline rate counted as volume understates the ratio by roughly that same share. Worked example: 120 disputes over 10,000 settled transactions is 1.20%; over 11,500 authorisations including 1,500 never-settled attempts it is 1.04% — under the line on paper, over it in reality. Pull the settled count from the processor settlement report, not from the checkout or gateway attempt log.

4. Dollar-weighted ratios

The error: dividing dispute dollars by sales dollars — "$12,000 disputed on $1,000,000 sold is 1.2%, so we are fine." The network ratios are count-based, not dollar-weighted: disputes (and, for VAMP, fraud reports) divided by transactions. A dollar-weighted figure answers a different question — revenue at risk — and it can be badly misleading in both directions. A merchant selling $5 items with 150 disputes on 10,000 transactions has a 1.50% count ratio (on the ECM line) but a tiny dollar ratio; a merchant selling $2,000 items with 10 disputes on 10,000 transactions has a 0.10% count ratio but a large dollar ratio. The first merchant is the one facing a monitoring program, and the dollar calculation says the opposite. Track both numbers, but never compare the dollar figure with a program threshold — thresholds are counts over counts.

5. Wrong merchant-ID level

The error: measuring at the wrong level of the merchant hierarchy — a portfolio total when the network measures per merchant ID, or a single store's figure when the acquirer scores the portfolio. The programs score the merchant ID being measured, and the count gates (1,500 events for VAMP, 100 chargebacks for ECM) apply at that level: aggregating ten small MIDs into one portfolio figure can manufacture a count gate that no single MID trips, while splitting one business across MIDs can hide a ratio that trips on the combined book. The direction of the error depends on which level your acquirer actually scores, which is why the first question is always what level the acquirer's count is computed at — get it in writing. One change makes this urgent rather than academic: under GMAP from April 1, 2027, Mastercard's ECM, HECM and EFM monitoring shifts from the merchant ID to the sub-merchant ID wherever one is populated (acquirer-summary, confirm with your acquirer — see the GMAP switch guide). A marketplace seller measured on a blended portfolio today may be measured per sub-merchant tomorrow, and the ratio at that level can look entirely different.

The five-minute denominator check

Before you compare any ratio with any threshold, run this check on the month in question:

Per network, not blended: one Visa figure on Visa volume, one Mastercard figure on Mastercard volume.
Settled only: settlement report counts, not authorisations, attempts or orders.
Gross, not net: do not subtract refunds from the denominator.
Counts, not dollars: disputes and transactions as counts, not amounts.
Right level: the merchant ID (or, from April 2027, sub-merchant ID) level your acquirer scores — confirmed in writing, not assumed.
Right month: same month for Visa, prior month for Mastercard — the two errors covered in the companion guides, repeated here only as a checklist line.

Then run the corrected figures through the Chargeback Ratio Calculator and the Merchant Monitoring Program Checker. If the corrected ratio still crosses a line, the problem is real and the remediation is the numerator: descriptors, alerts, cancellation flow and reason-code fixes — the prevention work in the monitoring programs guide. If the corrected ratio falls back under the line, the problem was the denominator, and the fix cost nothing.

FAQ

Should I subtract refunds from my transaction count before computing my ratio?

No. The networks divide by gross settled transactions. Subtracting refunds shrinks the denominator and inflates your ratio — 120 disputes over 10,000 settled transactions is 1.20%, but over 9,200 net-of-refund transactions it is 1.30%.

Can I combine Visa and Mastercard volume into one ratio?

No. Each network measures only its own volume: Visa scores Visa settled card-not-present transactions, Mastercard scores Mastercard transactions from the prior month. A blended figure hides the brand that is actually in trouble.

Do declined or voided authorisations count in the denominator?

No. Only settled (cleared) transactions count. Declines, voids and never-captured authorisations never settle, so counting them inflates the denominator and flatters the ratio.

Is my ratio dispute dollars divided by sales dollars?

No. Network ratios are count-based: disputes (plus fraud reports for VAMP) divided by transactions. A dollar-weighted figure measures revenue at risk, not program position, and must never be compared with a program threshold.

At what level is my ratio measured — store, merchant ID, or portfolio?

At the merchant ID your acquirer scores — ask for it in writing, because aggregating or splitting MIDs changes both the ratio and whether a count gate trips. From April 1, 2027, Mastercard's ECM, HECM and EFM monitoring shifts to the sub-merchant ID where populated (acquirer-summary, confirm with your acquirer).