Metrics and diagnostics
How to Calculate Your Shopify Chargeback Rate: Count and Value Explained
Calculate a count-based chargeback rate by dividing a defined number of chargebacks by a defined population of eligible payment transactions, then multiplying by 100. Track disputed monetary value separately. Orders, payment transactions, disputed cases, and currency amounts are different units, and mixing them can produce a plausible-looking percentage that answers the wrong question.
For a Shopify merchant, the safest report contains two distinct views: the processor-reported monitoring metric and an explicitly named internal analysis. Reconcile the two before using either to assess a change in performance.
Write the calculation contract first
For a monthly operational event ratio, specify the numerator as distinct eligible chargeback events received in the reporting month. Specify the denominator as chargeback-eligible payment transactions processed in that month, using the same store, provider, timezone, and eligibility scope. The purchase that produced a current chargeback may have happened in an earlier month.
Shopify documents a monthly count calculation using chargeback-eligible transactions. It also distinguishes a standard rate from an NDRP-inclusive rate, with the latter used for Shopify Payments account monitoring. Keep those labels intact when recording the displayed values. Shopify's rate definitions.
An order count is not a safe substitute for eligible transaction count. Some orders may be unpaid, use other payment methods, or involve multiple payment events. Build the denominator from the applicable transaction definition rather than treating every row of an order export as a qualifying card payment.
Your calculation contract should contain:
| Field | Required specification |
|---|---|
| Population | Store, payment provider, methods, and eligibility rules |
| Numerator | Event types included, deduplication rule, event timestamp |
| Denominator | Eligible transaction type and processing timestamp |
| Period | Start, end, timezone, and extraction date |
| Program treatment | Whether automatically resolved disputes are included |
| Display | Percent precision and source of the official comparator |
Keep a zero denominator as “not calculable.” Do not display 0%, which would imply measured absence of disputes in a valid transaction population.
Work through count examples
In a hypothetical month, a store receives twelve qualifying chargebacks and processes 2,400 eligible transactions. Its internal monthly event ratio is 12 divided by 2,400, multiplied by 100, or 0.5%.
Suppose the store also has 600 orders paid through a payment method outside this report's eligibility scope. Adding them to the denominator would yield 0.4%, but that percentage would no longer represent the stated population. The lower number would be a reporting error, not an operational improvement.
Now suppose three of the twelve chargebacks are later won. Keep the filed-event count at twelve for this calculation. Shopify states that winning does not remove a dispute from its chargeback rate. Shopify's chargeback monitoring guidance.
For a separate hypothetical NDRP-inclusive illustration, assume the applicable definition adds four distinct automatically resolved disputes to those twelve cases. With the same 2,400 eligible transactions, the inclusive ratio would be 16 divided by 2,400, or approximately 0.667%. This arithmetic illustrates why two legitimate metrics can differ; the actual included events must come from the applicable provider definition.
Do not transfer this inclusion rule into a Visa monitoring calculation. Network programs have their own specifications. An internal Shopify comparison does not establish network eligibility, exclusions, or thresholds.
Keep purchase cohorts distinct from monthly events
A purchase-cohort analysis asks a different question: what proportion of eligible transactions processed in a purchase period subsequently produced a qualifying dispute within a stated observation window? Both numerator and denominator refer to that purchase population.
In a hypothetical cohort of 2,400 transactions, eight have linked disputes after the chosen follow-up. Its observed cohort proportion is approximately 0.333%. That can differ from the 0.5% event ratio above because the current month's twelve incoming disputes include older purchases.
Neither calculation is inherently wrong. The error is displaying them under the same name or comparing cohorts with unequal follow-up. Label the first “monthly incoming chargebacks / monthly eligible transactions” and the second “purchase-cohort disputed transaction proportion, observed through [date].” State whether multiple episodes on one transaction are counted once or separately.
Track value exposure alongside frequency
A count rate describes frequency, not the amount at stake. Suppose the twelve hypothetical cases total USD 3,600 while eligible processed value is USD 180,000. The separate disputed-value ratio is 2%. Label it as an internal value measure, not the chargeback count rate.
Use the actual disputed portion when a case challenges only part of a payment. Keep fees, later recoveries, and accounting adjustments outside this gross disputed-value numerator unless the report explicitly defines another measure. Do not add EUR amounts directly to USD amounts. Show each currency separately or describe the conversion method.
A store with fewer disputes can still have greater value exposure if the cases concern larger purchases. Displaying count, count rate, disputed amount, currency, and average disputed amount makes that difference visible without confusing it with a profit calculation.
Reconcile before drawing conclusions
When your figure differs from the provider's display, compare date boundaries, timezone, refresh time, eligible payment methods, automatic-resolution treatment, repeated event rows, and historical updates. Reconcile raw counts before comparing rounded percentages.
Publish a short reconciliation note: “Internal result [value]; displayed provider result [value]; explained difference [events or population]; remaining unmatched items [count]; owner [role].” If a material difference remains unexplained, keep the internal figure provisional. Accurate labels and a traceable denominator are more useful than a polished percentage whose meaning changes from one meeting to the next.
Review dispute amounts and reasons together in Lower Chargeback.
Related reading in this collection:
- Design Internal Chargeback Alerts from Your Own Baseline
- Why a Falling Sales Month Can Distort Your Dispute Dashboard