Return, refund, or warranty claim? Routing requests so defects don't poison your data
A routing decision tree for returns, refunds, and warranty claims, plus a worked example of how mis-routed defects distort rates, COGS, and reorders.
InventoryX is in development and not available today. ClearStock is the live Shopify app.
Every refund looks identical in your sales report. That is the trap. A customer who ordered the wrong color, a customer whose bike light died in week six, and a customer invoking your one-year warranty in month ten all arrive wearing the same clothes — a request to send something back. Refund them all the same way and your data can no longer tell remorse from defect from warranty. The cost compounds quietly: return rates you can't diagnose, reorder quantities built on phantom demand, and supplier claims you never file because the defect counts don't exist anywhere. Routing each request down the right path is what keeps the numbers truthful.
One exception up front: this article covers the three legitimate paths. Requests that follow a pattern — serial returners, wardrobing, empty-box claims — are a different discipline with different tools, covered in spotting return abuse patterns. Here we handle the legitimate ones and the mechanics of sending each down its own route.
The three legitimate paths, and where they differ
- Remorse return. The product works exactly as intended; the reason is the customer's — changed mind, wrong size, wrong color. It should end in a refund and a restock to sellable, and it is the type your return rate should actually measure.
- Defect return. The product failed early, within your return window. Refund or replace, on your shipping dime, and the unit gets inspected rather than resold.
- Warranty claim. The product failed after the return window but within the period your warranty promises — a commitment about how the product performs, not about whether the customer still wants it (FTC, A Business Guide to Federal Warranty Law, accessed November 2026). The remedy is repair or replacement per your document.
| Remorse return | Defect return | Warranty claim | |
|---|---|---|---|
| Trigger | Changed mind, wrong fit or color | Fails within the return window | Fails after the window, within the warranty period |
| Return shipping | Customer, commonly | You | You |
| Money direction | Refund | Refund or free replacement | Free replacement or repair |
| Where the unit goes | Back to sellable | Inspected; written off or held damaged | Inspected; written off; possible supplier claim |
| The record you need | A clean return rate | Defect count per SKU | Warranty claims per SKU |
The last row is the one merchants skip, and skipping it is what poisons everything downstream.
What mis-routing actually costs: a worked example
Illustrative setup: LIGHT-TR-02, a rechargeable trail bike light. Last month it sold 120 units at an illustrative $24 retail against a $9.20 unit cost. Reading the support emails afterward, the truth was 1 remorse return and 5 defective units — switches that died. The merchant refunded all six identically, no restock, no tags.
Recorded return rate: 6 ÷ 120 = 5%. True remorse rate: 1 ÷ 120 ≈ 0.8%. Three costs follow from that four-point gap:
- The diagnosis is wrong. A 5% return rate reads as "customers don't love this product," which points at photos, listing copy, or expectations. The truth — a 4.2% defect rate — points at the switch assembly and a conversation with the supplier. Marketing fixes can't fix hardware.
- Demand is understated. Recorded net demand is 120 − 6 = 114 units; true sellable demand is 120 − 1 = 119. Reordering to 114 leaves you about 5 units short each month. Over a 10-week reorder cycle that compounds to roughly 11 units — about $270 of estimated missed revenue at $24 retail (illustrative).
- Supplier recovery evaporates. Five defects at $9.20 is $46 a month recoverable through a debit memo against the supplier — if defect counts exist to support the claim. Buried inside ordinary refunds, that is roughly $550 a year, estimated, that nobody files for.
There is a margin version of the same problem: Shopify's native profit reports match refunds against cost-per-item (Shopify Help Center, accessed November 2026), so five defect refunds book as ordinary margin erosion — indistinguishable from a discounting mistake. And if the merchant had restocked those units instead, the failure compounds: a light that died once goes back on the shelf, sells again, dies again, and generates a second refund plus a worse review.
The routing decision tree
Three questions at intake, in order:
- Does the product function as intended? Yes → remorse path: refund, restock to sellable. No → question 2.
- Did it fail within the return window? Yes → defect return: refund or replace, you pay the label, inspect and write the unit off. No → question 3.
- Is it inside the warranty period, and does the failure match a covered defect? Yes → warranty claim, remedy per your policy. No → a goodwill decision, which you can grant but must log separately so it pollutes neither ledger.
| Route | Money | Inventory | Ledger to keep |
|---|---|---|---|
| Remorse | Refund | Restock to sellable | Return rate (clean) |
| Defect | Refund or $0 replacement | No restock; damaged or written off | Defect count per SKU + supplier claim |
| Warranty | $0 replacement or repair | No restock; written off | Warranty claims per SKU |
| Goodwill | Discretionary | Case by case | Separate goodwill log |
Wrong-size and wrong-color requests deserve a sub-route of their own: when stock allows, a swap keeps the sale, and what that does to your shelves is covered in the exchange-first inventory guide.
Processing each route in Shopify
- Remorse: refund with the restock option set, so the unit returns to sellable the moment the return is marked received (Shopify Help Center, accessed November 2026).
- Defect: refund without restock, or receive the unit and adjust it to a damaged, unsellable state. If the customer wants a working unit rather than money back, ship the replacement as a $0 draft order — the unit movement stays visible in order history without printing a fake revenue line.
- Warranty: the same mechanics as a defect replacement; what differs is which paragraph of your policy authorizes it and which ledger records it. The document that defines that promise is the subject of the warranty wording guide.
Three intake questions that force the route
- "What is the product doing — or not doing?" Function failure points at defect or warranty; "it doesn't fit" or "I changed my mind" points at remorse. This single question sorts the majority of requests.
- "When was it bought, and when did it fail?" Two dates, two windows: the return window and the warranty period. Both live on the order; ask before you promise anything.
- "Do you want money back, or a working unit?" Intent determines the money direction and whether you are shipping a replacement draft order or processing a refund.
Then the habit that makes all of it data: tag the order when you process it — defect-2026-11, warranty-2026-11 — so counts are pullable in bulk later. A routing decision that isn't recorded is just a conversation.
Where tooling stands
One note about what we build: InventoryX (in development) computes per-SKU demand and out-of-stock days from recorded sales, and it doesn't yet separate defect refunds from remorse returns — the routing ledger stays a spreadsheet habit for now. Until routing data lives inside the tool, the weekly buying list inherits whatever distortion your refunds carry; the tag habit above is the working mitigation.
Common questions
What's the difference between a warranty claim and a return? A return unwinds the sale, typically within a short window, ending in a refund. A warranty claim invokes a promise that outlasts the window: the product failed through a covered defect, and the remedy is repair or replacement. They differ in data too — returns should measure remorse; warranty claims should measure defect rates.
Should defective returns be restocked? No. A unit that failed once is likely to fail again, and restocking it converts one defect into a second refund and a worse review. Receive defective units into a damaged or written-off state, count them in your defect ledger, and pursue recovery from the supplier instead.
Do warranty replacements count as sales? Usually not as revenue. The common mechanic is a $0 draft order or a manual shipment, so units move while revenue stays flat (Shopify Help Center, accessed November 2026). What matters more is counting replacements per SKU — a rising replacement rate is the earliest signal of a component problem.
How do I calculate a defect rate? Defective units divided by units sold over the same period, per SKU — 5 defective against 120 sold is 4.2%. Build it monthly from tagged refunds and warranty claims. Once the rate exists per SKU, it supports supplier conversations and gives your warranty duration math a real input.
What is a remorse return? A return where the product works exactly as intended and the reason is the customer's: changed mind, wrong size, wrong color, a gift that missed. It is the type your return rate should measure, and the reason separating it from defects matters — mixing the two makes both numbers unreadable.
Next step
Tag today's first five refund requests using the decision tree, and open a two-column defect ledger — SKU, count — the next time a defect crosses your desk. If your policy never defined what happens after the return window, write that promise first with example warranty wording; if wrong-size requests keep arriving as refunds, steering exchanges instead of refunds covers what a swap program does to your stock.