Ecommerce Returns Fraud: Fight It Like an Operator

Every seller has opened a return carton and found a brick, a rag, or a three year old unit where a new one should be. The first reaction is anger. The second should be process. Returns fraud in Indian e-commerce is not an occasional event. It is a steady leak that scales with your volume, and it responds to discipline, not outrage.

Know the taxonomy

Fraud has types, and each type needs different evidence.

• Empty box returns. The carton comes back with filler or nothing. Caught by weight, if you weigh.
• Item swaps. A genuine looking but older, cheaper or counterfeit unit replaces yours. Caught by serial numbers and QC codes, not by weight.
• Used item returns. The product comes back worn, washed or run. Common in appliances, footwear and electronics.
• Refund abuse. Claims of non delivery or damage designed to trigger a refund without any return at all.
• Wardrobing. Fashion’s oldest trick. Bought for the wedding, worn once, returned with the tag tucked back in.

Label every incident by type in your returns log. The mix tells you where to invest. A brand drowning in swaps needs serialisation. A brand drowning in wardrobing needs tag design and pattern flags. Without the log, you fight everything equally and win nothing efficiently.

Where the marketplaces stand

Marketplaces sit between you and the buyer, and their default position favours the buyer, because buyer trust is their product. But every major platform runs some form of seller protection claims process for returns that come back wrong. The mechanics differ by platform and change often, so check the current policy rather than relying on seller group folklore. What does not change is the principle: claims are decided on evidence, filed within a window, in the format the platform asks for. Sellers lose claims less because they are wrong and more because they file late, file thin, or never file at all.

Evidence discipline

Weigh everything, both ways

Record the packed weight of every outbound order against its order ID. Weigh every return before opening it. A weight gap is the cleanest, most objective evidence you can put in a claim. It is also cheap: one calibrated scale and one barcode scanner at each station, and a log that fills itself.

Unboxing videos at scale

One phone propped over the returns table does not survive volume. Build a fixed camera station above the returns bench, with the return label scanned into frame at the start of each clip. Store footage indexed by return ID for at least the length of the claim window. The video that wins a claim is one continuous shot: sealed carton, label, opening, contents. Cuts kill credibility, so design the bench such that nobody needs to pause the recording.

Serial numbers and QC codes

For electronics and anything above a value threshold you set, record serial numbers at dispatch. For unserialised goods, apply your own discreet QC stickers or coded marks placed where a fraudster would not think to look. A return that comes back without your code is a swap, provable in one photograph. This single habit converts the hardest fraud type to catch into the easiest one to win.

Design packaging against tampering

Prevention is cheaper than claims. Use tamper evident tape that tears instead of peeling. Seal high value boxes so opening leaves a visible mark. In fashion, use return tags that must be removed to wear the garment and cannot be reattached, positioned where they show in a mirror. None of this stops a determined fraudster. All of it raises the effort required, and most fraud is lazy. Raise the cost of cheating you above the cost of cheating the next seller.

The claim filing ritual

Treat claims like a production process, not a grievance. Every wrong return triggers the same checklist within twenty four hours: weight record pulled, video clip attached, photographs taken, serial or code mismatch documented, claim drafted in the platform’s own format, filed inside the window. Track outcomes monthly: filed, won, lost, value recovered. The teams that recover meaningfully run this as a weekly rhythm inside account health operations, the same way our Amazon India Account Management engagements run it. What gets filed on time gets paid. What waits for someone to feel like it does not.

Read the patterns

Fraud clusters. Analyse your returns by pincode, by customer, by SKU and by payment mode. A pincode running at double your average return rate deserves a rule: convert it to prepaid only, or add verification for cash on delivery orders. Repeat offenders ordering under different names surface when you match phone numbers and delivery landmarks. High RTO zones and high fraud zones often overlap, and both drag your unit economics at the same time. Most platforms give you enough raw data to build this in a spreadsheet. Review it monthly and turn the worst clusters into standing policy.

Absorb or fight: the decision rule

Not every incident deserves a war. Set a threshold based on item value and evidence strength. Below the threshold, absorb it, log it, and let pattern analysis catch the repeat offenders. Above the threshold, or wherever the evidence is clean, always file, because platforms notice sellers who document consistently. Fighting everything burns team hours on small losses. Fighting nothing teaches the ecosystem that you are free to steal from. The point of a threshold is that the decision becomes a rule, not a mood, and a junior teammate can apply it without escalation.

Build the system this month

Start with the cheapest layers. Week one: scales at outbound and returns, weights logged against order IDs. Week two: the fixed camera station and a folder structure for clips. Week three: the claims checklist, printed with owner and deadline above the returns bench. Week four: the first pincode and repeat customer analysis. Then measure gross return rate and recovered value monthly. Fraud never goes to zero. But a seller with weights, videos and codes stops being the easy target, and the leak shrinks to a number your margins can carry.