Return fraud is a small-store problem too. Here's how to catch it without an enterprise budget

Return fraud stories usually come with enterprise numbers attached — organized rings, freight-scale losses, loss-prevention departments. Which lets small merchants tell themselves it’s someone else’s problem. It isn’t. If anything the incentives run the other way: fraudsters know a two-person Shopify store has no loss-prevention department, no fraud scoring, and no time to argue about a $60 hoodie.

You don’t need enterprise tooling. You need three habits and the discipline to keep them on busy weeks.

Know what you’re actually defending against

Most of what hits small stores is one of these:

• Wardrobing — buy, wear it to the event, return it. The item comes back smelling of one great Saturday night.
• The serial returner — a shopper whose relationship with your store is mostly returns. Each individual return looks fine; the pattern is the problem.
• Damage claims without damage — “it arrived broken” plus a demand for a refund, sometimes with a request to skip sending it back.
• The empty box / wrong item return — the tracking number says returned; the box says bricks.

Notice what these have in common: every single one works better when you pay out before physically inspecting what came back. That’s the thread to pull.

Habit one: proof before payout

Require a photo on damage claims. That’s it — that one requirement kills a surprising share of nonsense claims, because the effort of faking evidence exceeds the payoff of a small-ticket fraud. Honest customers don’t mind; they’re usually already annoyed enough at the broken item to photograph it enthusiastically.

The photo also protects you in the other direction: when the claim is real, you can approve it fast and confidently, sometimes without waiting for the item to ship back at all. Evidence speeds up the honest path and slows down the dishonest one. That’s the whole game.

Habit two: watch frequency, not individual returns

No single return tells you anything. The third return in thirty days tells you a lot.

The tricky part is that frequency is invisible in the default workflow. Each return request arrives as its own email or its own row, and unless you have an unusual memory for customer emails, shopper number 1015’s second and third returns look exactly like anyone else’s first. You have the data — it’s your own order history — but nobody’s surfacing it at the moment you decide.

Fix that however you can. A spreadsheet with a lookup works. A returns tool that flags repeat requesters right in the approval queue works better — Core and Growth in RefundShift do this from your store’s own return history, with no third-party “risk score” — but the principle matters more than the tool: the return frequency should be in front of you at the moment you review the request, not discoverable three tabs away.

A flag is not a verdict, to be clear. Some of your best customers return a lot because they buy a lot. The flag’s job is to make you look twice, and looking twice is usually all it takes.

Habit three: automate only behind hard limits

There’s a strain of returns software that sells “auto-approve rules” as a time saver — returns under $X, customers above Y orders, approve automatically. For a small store, an unguarded auto-approve rule is a published price list for stealing from you. Fraudsters find thresholds fast, and they stay politely under them. If you do automate, set a low amount cap and make return-frequency signals a hard stop that sends the request back to your queue.

Approving a return takes seconds when the evidence is in front of you: photos, order history, return count, amounts. The expensive part was gathering that evidence, not clicking the button. So buy or build the gathering, and keep the click for anything outside a narrow, guarded lane.

This also changes how the two bad outcomes feel. Approve a fraudulent return knowingly-carelessly and it stings; you’ll tighten up. Have software approve it silently without guardrails and you won’t even know which policy to fix.

What about the customer who’s just… difficult?

Not fraud, but adjacent: the return that’s technically eligible and clearly abusing your goodwill. The stained “unworn” dress. Here’s where a middle option earns its keep — instead of approve-or-reject, ask for more: another photo, the tags, the original packaging. A revision request costs an honest customer thirty seconds and costs a chancer their momentum. Most chancers quietly drop off.

Where store credit fits in

One more structural defense worth knowing: returns settled as store credit instead of cash refunds are simply less attractive to fraud. Cash extracted from your business spends anywhere; credit only spends at your store, on your margins, and leaves a paper trail attached to a customer account. Fraudsters want liquidity. Don’t give them any.

None of this requires software with an enterprise invoice. It requires evidence before money, patterns made visible, and either a human on the approve button or hard guardrails around a deliberately narrow exception. Small stores can do all three — often better than the big ones.