Every dispute you defend in the prior modules also gets counted. The card networks run portfolio-level surveillance on your dispute and fraud rates, and once you cross a published line, your acquirer inherits fines, scrutiny, and pressure to put you on a remediation plan or offboard you entirely. For a subscription business this matters more than for a single-purchase merchant, because recurring billing manufactures dispute volume on a schedule. The same card that worked last month can trip a "subscription canceled" or "not recognized" dispute this month, and that drip feeds straight into the ratio the network watches.
This module is about the meter, not the dispute. The sibling course covers how the four-party system resolves a single chargeback. Here we care about the count, the denominator, and the timing that decides whether you stay invisible to the network or land in a program with your name on it.
How the meter actually works
Both networks compute a ratio: disputes over sales, with a minimum count gate so low-volume merchants are not flagged on noise. The two details that trip people up are which denominator and which month.
Mastercard: lagged denominator, two gates
Mastercard's Excessive Chargeback Merchant (ECM) program divides chargebacks received this month by sales transactions from the prior month. The lagged denominator is the trap. If you ran a launch promotion in March and disputes from those new subscribers land in April, your April chargebacks are measured against a denominator that does not include April's (possibly larger) sales. A growing book can flatter the ratio, but a book that just stopped acquiring can make it spike.
ECM enrollment requires both a count and a ratio: 100 to 299 chargebacks in a single month and a chargeback ratio at or above 1.5 percent. Crossing both gates in a single month flags you for monitoring; Mastercard applies the ECM designation and fines only after you exceed both thresholds in two consecutive months. The High Excessive tier (HECM) requires 300 or more chargebacks and a ratio at or above 3 percent. You must clear both the count gate and the ratio gate. One hundred chargebacks at 1.2 percent does not enroll you, and 1.8 percent on 60 chargebacks does not either.
Visa: VAMP folds fraud and disputes into one number
Visa retired the old Dispute Monitoring Program (VDMP) and Fraud Monitoring Program (VFMP) and consolidated them into the Visa Acquirer Monitoring Program (VAMP), which became the single live program through 2025 and 2026. The VAMP ratio is broader than the old dispute-only math: it combines TC40 fraud reports and TC15 disputes in the numerator, divided by TC05 settled transactions.
That combination matters for subscriptions. A fraud alert (TC40) and the chargeback that follows it can both land in your numerator if you do not refund fast enough to head off the dispute. The same bad transaction can be counted twice.
As of April 1, 2026, the merchant VAMP ratio threshold is 1.5 percent, tightened from the 2.2 percent that applied from mid-2025. Visa also runs an enumeration ratio targeting card-testing and automated attacks, with a 20 percent threshold, and merchants over threshold are charged $8 per dispute above threshold in a flagged month. Acquirer-level thresholds sit far lower (around 0.50 percent "above standard"), which is why your acquirer cares about your number long before you personally cross 1.5 percent.
A worked example
Take a DTC subscription box doing 40,000 Visa sales a month, steady, no growth spurt.
At a 1.0 percent VAMP ratio you are generating roughly 400 combined TC40 plus TC15 events a month. Comfortable. Now you push a free-trial-to-paid conversion that adds 6,000 first-time rebills, and a slice of those customers forgot they signed up. Say 300 of them dispute as "subscription canceled" or "don't recognize," on top of your baseline 400.
Numerator: 700. Denominator under Visa is current settled transactions, call it 46,000. Ratio: about 1.52 percent. You just crossed the April 2026 line on a single conversion campaign, and the fines attach for that month.
Run the same event through Mastercard's lagged math and it can look worse, because the 300 spike-month chargebacks are divided by the prior month's smaller sales count, not the inflated current month. The denominator that would have rescued you is a month too late.
Why subscriptions cross the line differently
Three structural features of recurring billing feed the meter, and each maps to an earlier module.
Involuntary churn and aggressive dunning (modules 2 and 3) push retries against cards that have already failed, and a retried charge on a card the customer considers dead is a chargeback waiting to happen. Card lifecycle gaps (module 4) mean a card that was updated by the issuer but not by you bills a stale credential, which reads to the cardholder as an unexpected charge. And the recurring nature itself means "I forgot I was subscribed" disputes arrive monthly, not once.
The defensive levers live in the modules around this one. Account updater and pre-dunning notices cut the disputes before they count. Clear billing descriptors and proactive cancellation flows reduce "don't recognize" volume. Compelling Evidence 3.0 (module 8) can get qualifying friendly-fraud disputes reclassified so they fall out of your fraud count, which directly lowers the VAMP numerator.
Staying out, in practice
Watch the ratio weekly, not monthly. By the time a network notice arrives you are already a month or more into the data, and a remediation plan gives you a defined window to get back under both gates.
Manage the count gate, not just the ratio. Mastercard's dual requirement means a high-volume month can absorb more disputes before the count gate trips, so a sudden drop in new sales can enroll you even if customer behavior did not change.
Refund before the dispute when the economics allow it. On Visa, a fast refund can stop a TC40 from maturing into a TC15, keeping one event out of the numerator instead of two. For a subscription business pushing thousands of rebills, the cost of a handful of pre-emptive refunds is almost always less than crossing a threshold and inheriting fines plus acquirer scrutiny.
The takeaway: the dispute you win still counts. Monitoring programs measure volume and ratio, not who was right, so the work that keeps you out of them is upstream of representment. Fix involuntary churn, keep credentials current, and refund the borderline cases early, because the cheapest dispute to defend is the one the network never logs against you.