ReturnsIntel began after years of seeing monthly SKU return rates above 100% in a Shopify workflow. At first glance, that seems impossible. A business cannot return more of July's units than it sold in July.

The calculation became clear once we examined the dates. If a store sells two shirts in July and processes three returns in July from earlier orders, dividing three by two produces 150%. The arithmetic is correct. The label is the problem.

Both calculations can be useful, but they answer different questions. The event-month ratio can help with workload or cash-flow reconciliation. The cohort rate is the stronger product-performance metric because the numerator and denominator refer to the same units.

Shopify's sales reports document that a sale appears on the order date while a sales reversal appears on the date it is processed. Its finance documentation gives the same pattern: net sales can appear positive on the order date and negative when a later return or refund is processed.

Shopify also warns that a selected date range can include only one side of a transaction. Its Product Insights documentation explicitly notes that net units can be negative when more items are returned than sold during the period.

• Sales reversals are broader than physical returns. Refunds, cancellations, order edits, and other adjustments can affect sales reports without representing an item sent back.
• Returns and refunds are separate. Treating every refund as a physical return can inflate the numerator.
• Requests and completed returns are mixed. A return lifecycle includes requested, approved, processed, closed, canceled, refunded, exchanged, and restocked states.
• The newest cohort is incomplete. Recent sales have not had the same opportunity to be returned as older sales.
• The same unit is counted twice. Return-provider, Shopify Return, and Refund evidence may all refer to one event.
• SKU history is incomplete. Missing, changed, or reused SKUs can break variant-level attribution.

• Choose the SKU or product family and the original sale month.
• Count units sold in that cohort.
• Attach item-level returns and attributable refunds to those sold units.
• Deduplicate evidence that refers to the same returned unit.
• Apply a consistent observation window and hide incomplete cohorts.
• Compare the resulting rate with returned-unit volume and observed financial exposure.

A properly deduplicated unit-level cohort return rate should not exceed 100%. If it does, inspect event attribution, quantities, exchanges, and denominator completeness.

Next, use the full SKU return-rate calculation guide or print the Shopify Return Leakage Checklist.

Sources: Shopify sales reports, finance reports, analytics discrepancies, and Product Insights.