A year of refunds, in one report.

Timing of each refund against its shipment and delivery scan. Same $845,222, split once. The 30-day return window is the store's own.

August 2026 was the highest month of the year at $108,401, 48% above the 12-month median of $73,137. November and December carried the order peak (8,673 and 7,733 orders) at a lower refund rate.

The customer had the goods and disputed anyway. About $67 is still open and can come back if the chargeback is answered with evidence. Settled chargebacks are history and never count as open money.

Each refund was rated medium or high risk on the patterns behind it, not on checkout fraud analysis. They are reviewed one at a time in Activity.

Orders refunded before shipping are cancellations, not returns. Refunds after shipment, where abuse can happen, come to $494,861.

Together this cohort accounts for $110,392 in refunds. Review these customers before the next order ships.

19 address clusters where several accounts ship to one address. The largest: 9 accounts on one Wyoming address. A 5-account cluster in Texas matched at full confidence.

Ranked by dollars. The top customer filed 13 refunds; the next four filed 3 to 10 refunds each, $2,280 to $3,167.

The store refunds 3.2% of orders. Disposable-email orders refund 24 times more often.

28 of 178 orders carrying the code saw a refund, against 4.8% across 48,020 orders without it. $13,825 is the amount concerned, not money recovered.

53% of chargebacks were filed as fraudulent. The evidence that answers them: delivery scan, address match, prior orders from the same customer.

This customer averages a risk score of 60 across refund events. The report shows a masked account number; the customer profile shows the full history.

Checkout fraud analysis asks one question: was this a stolen card. Abuse that starts after the parcel ships is invisible to that check, which is where these chargebacks came from.

One customer who charged back has placed one more order since, worth $259. Review the record before another order ships.

Refunds most of their orders, ships to a flagged address, refunds a lot: the worst matches all three. A far stronger signal than any single pattern, and the first customers to watch before their next order.

Loyal customers here average 1.2 orders at $283 each. Had these refunders stayed, about $437,675 in lifetime revenue. An estimate, not a loss.

Product gaps separate a fit or quality problem from customer behaviour. Context, not a verdict.

Bundles refund more often than single accessories. Worth one conversation with the product team.

The store average is 0 per 1,000. Two chargebacks is a small count; the report says so rather than drawing a conclusion.

Damage in transit, late delivery or carrier claims are worth one conversation. It is context, not fraud.

Recording reasons at refund time makes every other pattern sharper. A missing reason is shown as a coverage gap, never assigned to an abuse category.

The audit behind this issue is frozen. Live monitoring runs on every new order, return and refund from installation onward. Money counts as recovered only once the bank returns it. Nothing is held unless the merchant enabled the rule. Next issue: Oct 1, 2026.