Reducing returns in online fashion: the wrong size and how to avoid it

Someone returning a garment because “it’s not my size” or “it fits differently from what I expected” has not changed their mind: they bought without the information to choose. The moment a return is decided is not delivery day, it is the day of purchase — in front of a selector with S, M, L and a photo on a model who does not have their body.

That is why downstream measures — free returns, faster procedures — do not reduce returns: they make them less painful. To reduce them for real you have to step in before the cart, where the choice takes shape.

Almost every store has a size chart, and almost no shopper uses it. To use it they would need to know their measurements in centimetres, have a tape measure at hand, and then translate “waist 78” into a letter. Most skip the step and pick the size they “usually” take — which changes from brand to brand, and between a top and a pair of trousers.

• Nobody knows their measurements by heart. The chart starts from a figure the shopper does not have.
• Labels are not comparable. One brand’s M is another’s L; a 42 in trousers is not a 42 in shoes.
• The chart gives the size, not the fit. A garment can be the right size and hang in a way the shopper dislikes: that is a return too.

Cabina steps in exactly where the choice takes shape, and on the three problems above. The measurements are estimated from a photo the shopper uploads — or takes on the spot with their phone — and stay editable by hand: those who know them type them in, those who do not get them. The comparison is made against the size charts you uploaded, in centimetres, row by row, and the answer speaks the language of your selector: “M” for the top, “42” for the trousers, “40” for the shoe. If the measurements land between two sizes, it says so and recommends one, with a confidence percentage.

And the fit is visible: a generative model produces the image of the shopper wearing the garment — not a cut-out pasted on top, but a new photo in which the garment actually falls — with a before/after comparison. Whoever wants to see how a long coat or a dress looks on a body like theirs sees it before paying.

The recommendation is not a number to remember: where the page has a recognisable selector, the recommended size is preselected, and the “Add to cart” after the try-on starts with that variant. The step where the shopper puts the wrong size back by hand disappears.

The recommendation is only as good as the chart you give it. Three rules, explained in full in Size charts for automatic size recommendation:

• One chart per garment category — tops, trousers, dresses, outerwear, shoes — so that on trousers the recommendation is a number, not a letter.
• Real ranges in centimetres, taken from your supplier or measured on the garments, not copied from a “standard” chart that does not exist.
• Dimensions that do not matter at zero: for trousers, height may say nothing, and declaring it at random makes the recommendation worse.

You already know your return rate from your back office. To see the fitting room’s effect you have two more numbers in the dashboard: try-on sessions and how many run to completion. A simple way to read the result: compare the return rate of items enabled for try-on with that of items not enabled, over the same period.

The free plan gives 10 try-ons a month with no credit card: enough to enable the items with the most returns and watch what happens. Plans and costs are on the pricing page.