Follow one return, from arrival to recovered.

A return lands from Shopify, Trendyol, a CSV export, or a manual entry, and RefundMind structures it the moment it arrives — customer, item, order, reason, and channel attached, with a timeline that records every event from intake on.

From there the operator works a queue, not an inbox: cases sort by urgency, filter by channel, reason, status, or value, and carry a full audit trail. Nothing lives in a spreadsheet, and no decision is unaccountable.

A return is a choice, not a reflex: refund, exchange, store credit, or decline. RefundMind puts all four on the case and recommends the one your rules favour — with the rationale shown, so the operator sees why, not just what.

The operator confirms or overrides, and either way it is logged. Rules you write once apply across every matching case, and bulk decisions clear the similar ones in a single move.

When a case resolves as anything other than a full cash refund, that is recovery — and RefundMind books it in money. Per case, per channel, per period, refund-out sits against recovery-kept, so your recovery rate is a number instead of a feeling.

Recovery also closes the physical loop: once a unit is inspected, the operator dispositions it — restock, donate, or discard — so resalable stock goes back to the floor instead of a forgotten corner.

Work one case at a time and the patterns stay hidden — which customers churn value through returns, which SKUs fail most, which channels carry the quiet cost. Intelligence is the layer that reads what Cases, Decisions, and Recovery already produce and turns it into signal.

It is on the roadmap, and we name it that way — not as shipped. The preview shows the direction: a risk read on the customer, the pattern behind a case, and the action that would prevent the next return.