Two ratios, two denominators, one portfolio.

Fictional scenarios and synthetic data. Reviewed 6 November 2026.

Stilling Bank sends Northbank Pay two monitoring figures for Brightwater Tickets. One looks uncomfortable; the other looks ordinary. Before choosing which one to believe, Ines asks for the underlying event populations and reporting periods.

What is actually being decided.

The desk needs to explain the numbers and agree a response with the acquirer. A ratio alone does not establish programme status: the applicable network rules, region, entity, exclusions, count criteria and reporting extract matter. This article teaches reconciliation, not a live schedule of thresholds or assessments.

Separate the numerator from the story.

Visa’s published VAMP overview describes a count-based card-not-present metric combining fraud reports and disputes against settled transactions. It also identifies exclusions whose effect depends on the timing of the data extract. Preserve raw events and the adjustments actually recognised in the acquirer’s report rather than subtracting every refund or alert locally.

A refund, a dispute resolution and a fraud-report adjustment are not interchangeable records. Nor should the two numerator components be silently deduplicated by transaction unless the applicable methodology requires it. A transaction-level investigation and a programme event count can answer different questions.

A worked month, with explicit assumptions.

For this synthetic reconciliation, the acquirer confirms 40,000 eligible settled transactions and 660 eligible events after its recognised adjustments. The ratio is 660 / 40,000 = 1.65%. If both counts triple, 1,980 / 120,000 remains 1.65%. The ratio is unchanged; event count and commercial exposure are not.

That arithmetic does not establish whether either month enters a programme. The team needs the applicable current rules and any relevant count criteria before making that statement. No fee is inferred from the example.

At portfolio level, 1,980 events out of 13,200 represent 15% of the numerator. The merchant’s 120,000 transactions out of 2,400,000 represent 5% of volume. This concentration can explain why an acquirer asks questions even when a merchant’s individual classification is still being reconciled.

Put the reporting month on the page.

A previous-period denominator produces a different number from a same-period denominator. Suppose the agreed monitoring calculation uses 240 chargebacks and 12,000 transactions from its specified comparison period: 240 / 12,000 × 10,000 = 200 basis points. With 20,000 transactions as the denominator, the result is 120 basis points.

These examples show the effect of denominator choice, not a declaration that a merchant meets a current Mastercard tier. Confirm the exact ECP calculation and classification with the acquirer’s current Mastercard materials. Preserve the month labels so a seasonal volume change cannot masquerade as a change in underlying customer behaviour.

Enumeration has another denominator.

Visa’s published overview defines its enumeration ratio using authorisation attempts, including approved and declined attempts. Dividing enumeration events by settled sales is therefore the wrong denominator for that definition. Keep this operational traffic measure distinct from the fraud-and-dispute ratio.

A way to work it.

1. Name network, region, entity and reporting period.
2. Obtain the actual numerator and denominator records.
3. Show recognised adjustments separately, with extract dates.
4. Reproduce the acquirer’s number before proposing a correction.
5. Confirm the current classification criteria and any assessment directly.
6. Attach an action plan with an owner, review date and measurable outcome.

Take it to the desk.

Could another analyst reproduce both ratios from the component records without asking which month you meant?

Continue with dispute lag and cohorts and working with your acquirer.

Fictional scenario. Synthetic data. No real customer records.