The Returns Fraud Agent

Seven customer accounts at one address in Zetland had returned 40 orders in a year. None of them had returned more than nine, so the rule the owner had just drafted would never have found them. It would have found one of his best customers first.

On a Tuesday in October, Tash opened a return at the warehouse in Alexandria. The order was a black linen midi dress, A$189, returned for change of mind inside the 30 days the brand allows. The dress in the bag was black and midi-length. It was also two seasons old, from another label, with the care tag cut out. The barcode on the swing tag had been peeled off a different garment and did not scan.

The account behind it had four returns in twelve months, which is ordinary. Priya, who runs customer service, searched the delivery address. Six more accounts shared the same unit in Zetland, under different first names. Each of the seven shared a mobile number or a card with at least one other, and those links chained all seven together. Between them, the seven accounts had returned 40 orders, and 11 of those parcels, when Tash went back through the photos she sometimes takes, held something other than what was sold.

That same week Daniel, who owns the business, had written a new rule for the returns policy: any account with more than 20 returns in twelve months loses free returns. Priya ran it against last year's orders. The first account it caught belonged to a customer in Mosman with 43 returns, who orders two sizes of almost everything, keeps one, and had kept A$6,900 of clothing in the year. None of the seven Zetland accounts came close to 20.

The brand, Tash, Priya, Daniel and their customers are a composite, built from the patterns we see in online fashion brands of this size, and the figures attached to them are design projections. The consumer law, the privacy rules and the published figures are real, and linked to their sources.

An online womenswear brand in Sydney, selling direct through its own Shopify store, with a warehouse in Alexandria. Turnover is about A$7.2 million a year from 46,000 orders, at an average of A$156. Ten people work there: Daniel, a buyer, a marketing lead, two in customer service, four in the warehouse and a part-time bookkeeper.

Change-of-mind returns are free within 30 days, for unworn items with tags attached, refunded to the original payment. That policy is a choice. Returns arrive through Shopify's customer returns, are checked at the warehouse, and are refunded by customer service once the parcel is received. About 21% of orders come back, in this design, which is 9,660 returns and about A$1.38 million refunded in a year. Most of them are a dress in the wrong size.

This Blueprint covers what customers send back. Our Invoice Matching Blueprint covered the other direction: what suppliers send in, and what they bill for it.

Every return looks like every other return. Tash's team processes about 40 parcels a day. Each is opened, glanced at and marked received. Nobody compares the barcode in the bag with the order, because there is no screen that would show the order at the bench. Photos are taken when something looks odd, and filed in a shared folder by date.

Shopify counts customers by email address. It holds one customer record per email. A household with seven email addresses is seven customers, each with a return count that looks normal. The links between them (one address, one phone, one card) are all in the store's own data, in fields nobody adds up.

The only tool to hand is a count, and a count hits the wrong people. A returns count rewards spreading the same behaviour across several accounts, and punishes the customer who buys two sizes because the size guide does not tell her which one fits. The heavy returners are mostly the brand's most loyal customers. In this design, of the 46 accounts with more than 20 returns in the year, 40 were profitable after return costs, and together they kept A$141,000 of clothing.

The law limits what a rule can touch. Under the Australian Consumer Law, a business does not have to offer refunds when a customer simply changes their mind, but a change-of-mind policy it has published is one it must honour. Faults are different. If a product has a major problem, the customer can choose a refund or a replacement, whatever the store's policy says (ACCC: repair, replace, refund, cancel). A sign or policy saying "no refunds" is unlawful (ACCC: small business rights and responsibilities). So a returns rule can only ever govern the free, voluntary part: change of mind, unworn, with tags.

New privacy rules cover decisions made with software. From 10 December 2026, businesses covered by the Privacy Act that use a computer program to make, or do something substantially and directly related to making, a decision that could reasonably be expected to significantly affect someone's rights or interests must say so in their privacy policy, under new transparency rules in the Australian Privacy Principles (OAIC: transparency for AI and automated decision-making; Allens summary of the OAIC guidance). Most small businesses with turnover of A$3 million or less are outside the Act (OAIC: small business). At A$7.2 million, this brand is inside it. A score that decides who keeps free returns is the kind of thing the rule is written for.

In the United States, retailers lost an estimated US$103 billion to fraudulent returns and claims in 2024, 15.14% of all returns, according to Appriss Retail's annual research. That figure covers all of US retail and includes return abuse and claims this brand would not see. Apply the design's assumptions to this brand's 9,660 returns last year, about 0.65% with the wrong item in the bag and about 1.82% worn or with tags removed, and it models something smaller and more specific:

63 returns where the item in the bag was not the item sold, refunded in full: about A$9,450.

176 change-of-mind returns worn or with tags removed, refunded in full under a policy that says unworn with tags: about A$26,400, and most of it not saleable again at full price.

That is about A$35,850 a year refunded outside the brand's own policy, roughly 2.6% of what it refunds. The Zetland household sits partly inside it: 11 of its 40 returns are among the 63. Its other 29 refunds came to A$4,530, and they were within the policy as written, so the design flags the household for review and saves nothing on them. A$35,850 is real money for a business this size, and still small next to what Daniel's rule puts at risk: A$141,000 a year in sales from 40 profitable customers.

The design is next. The design, the build guide, the running costs and the take-away are free to subscribers. Enter your email below and the rest of this Blueprint opens. New ones arrive every Tuesday and Friday.