The same dress, back again in a size 12
Monday morning, the returns bags from the weekend are piled by the packing bench. Your warehouse person opens them one at a time. A linen midi dress in a 12, reason 'too small'. The same dress in a 10, 'too tight across the bust'. A pair of wide-leg trousers in a 14, 'too long'. The same trousers in an 8, 'too long'.
Nobody is surprised. Everyone on the team knows the midi dress comes up small, because they have handled thirty of these bags. The shopper does not know, because the product page says 'true to size' in the same template text every style gets. So the next shopper orders her usual 12, and in ten days the dress is back on the bench, needing a steam and a new swing tag before it can be sold again.
Each return also costs you the outbound postage, the label, the packing and the time. For a lot of brands, the return label is free to the customer as well.
Why the lesson never reaches the product page
The problem is not that you lack information. Your returns portal or your inbox already holds hundreds of reasons. The problem is the shape of it.
- Return reasons are a single dropdown, 'did not fit', with no 'too small' or 'too big' split and no note of which part did not fit.
- Reasons sit against the order, not against the style and size, so nobody can pull them up by product.
- Free-text comments like 'gaping at the back' live in emails that only customer service reads.
- Product pages are built from a template when the style launches and are never touched again.
So the knowledge stays in people's heads. It leaves when a staff member leaves, and it never gets to the one person who needs it: the shopper about to pick a size.
What fit returns are costing the business
| Where it hurts | What you see |
|---|---|
| Margin | Two lots of postage and handling on an order that earns nothing |
| Stock | The right size sits in a returns bag while another shopper finds it sold out |
| Condition | Garments come back creased, marked or missing tags and need work |
| Customer | A first-time buyer who got the wrong fit often does not come back |
| Design | The factory makes the same pattern again next season |
None of these show up as a single line in your accounts. They show up as a returns rate that stays stubbornly where it is, whatever you try.
How we turn returns into fit guidance
We connect to your store, usually Shopify, and your returns portal if you use one, and build the missing link between the returns bench and the product page.
- We change the return form so a fit return asks two short questions: too small or too big, and where (length, waist, chest, hips, shoulders, sleeves). It takes the customer seconds.
- Each reason is stored against the style, colour and size, not only the order, so you can see the linen midi dress in one view.
- Free-text comments are read by a language model (for example OpenAI or Anthropic Claude) and sorted into the same fit areas, so 'gaping at the back' counts as a fit note rather than disappearing.
- A weekly report ranks styles by fit returns and shows the direction: this style runs small, that one is long in the leg.
- Your team approves a short fit note for each flagged style, such as 'Runs small in the bust, we suggest sizing up', and it appears on the product page next to the size selector through a metafield.
- The same notes go to your design or production lead as a list, so the next production run can be graded differently.
A person always signs off the wording. The system finds the pattern; your team decides what to tell shoppers.
A normal week once it is running
The returns bench works the same way as before, scanning each parcel in. The difference is on Friday, when the report lands. It says three styles have tipped over your threshold for 'too small'. Someone in the team reads the comments, agrees one fit note per style and publishes them in a few clicks.
Shoppers on those pages now see the note before they add to basket. Customer service has fewer 'what size should I get' messages about those styles, because the page answers it. And when the buyer sits down with the factory to plan the reorder, the fit feedback is already on the table instead of in someone's memory.
Edge cases we plan for
Some returns say 'too small' when the real reason is that the customer ordered two sizes and kept one. We keep those separate, because they are not a fit fault. A style with only a handful of sales can look alarming on a small sample, so the report shows the count next to the rate and your threshold can require a minimum number of returns before anything is flagged.
Styles also change between production runs. When a new batch arrives with a corrected pattern, the fit note is retired and the counting starts again, so an old problem does not haunt a fixed garment.
Does this sound like your returns bench?
- Your team knows which styles run small, but the product pages do not say so.
- Your return reasons are one generic 'did not fit' option.
- You cannot pull up returns by style and size without a spreadsheet afternoon.
- Fit comments are buried in customer service emails.
- The same styles come back for the same reason season after season.