The $849.9 billion problem, and what video actually fixes

Returns are the quietest line item in a fashion store's P&L and often the most expensive. The product went out, the revenue booked, and then it came back: you paid to ship it twice, paid someone to inspect it, and in a good case you get to sell it again at full price.

That is the whole of retail. Apparel is considerably worse than the average. Category benchmarks put apparel returns somewhere in the 20 to 40% range and footwear in the 17 to 30% range, against roughly 19.3% for online sales overall. If you sell clothes, one in four or five items you ship is coming back, and you have almost certainly built that into your margin without ever attacking the cause.

Nearly all of it is one problem

Ask why, and the answer is boringly consistent. Fit and sizing is the single largest driver of apparel returns in every survey that measures it. The exact share depends heavily on how the question is asked: studies land anywhere from about 40% to as high as 70%, with most credible estimates clustering above half.

That range is wide enough that you should distrust anyone quoting you a precise figure, including us. What is not in dispute is the ranking. Fit is first, and nothing else is close. Damage, description mismatch and simple change of mind divide up what is left.

This matters because it tells you something specific about where the failure happens. A fit-driven return is not a fulfilment problem or a quality problem. It is an information problem, and it happened before checkout. The shopper formed an expectation from your product page, that expectation was wrong, and the return is the correction.

What a flat lay cannot tell you

A flat lay is a good photograph of an object. It shows colour, print, construction and, if it is well shot, fabric texture. What it cannot show is the thing that causes the return:

• Drape. Whether a fabric falls or holds its shape. A stiff cotton and a fluid viscose can photograph almost identically flat and behave nothing alike on a body.
• Proportion on a person. Where a hem actually lands. Whether a boxy cut reads as oversized or shapeless.
• Movement. How a skirt behaves when someone walks. Whether a knit clings.
• Scale. A measurements table is not the same as seeing an item on a body, and most shoppers do not read it.

Every one of those is a motion or a body property. The photograph is not underexposed or badly styled. It is structurally incapable of carrying the information, and adding a fifth angle of the same flat lay does not help.

The honest version of what video does

Here is where most vendor writing on this subject goes wrong, so let us be careful.

We cannot tell you that adding video will cut your returns by a specific percentage. We have not run that study, and the numbers circulating on marketing sites for this claim generally trace back to a vendor rather than a study. Treat any specific return-reduction figure you see, including from a competitor, as marketing until someone shows you the methodology.

What is defensible is the mechanism. If fit is the leading cause of apparel returns, and fit information is what static photography structurally cannot carry, then media that does carry it should reduce the gap between expectation and reality. A shopper who has seen the garment move on a body has a better-calibrated expectation than one who has seen it lying flat. Better-calibrated expectations mean fewer corrections.

That is a mechanism argument, not a measured result. It is also testable in your own store, which is a great deal more useful than someone else's benchmark.

How to actually measure it

You have the data to test this, and it costs nothing but patience:

• Pick 10 to 20 products with meaningful order volume and a return rate above your store average. Volume matters more than you think here; a product with 8 orders a month will take a year to tell you anything.
• Record their current return rate over a period long enough to include the return window, plus a few weeks.
• Add video to half of them. Leave the other half alone as a control. Resist the urge to change the copy, the price or the photography at the same time.
• Wait a full cycle, then compare the two groups rather than before-and-after on the same group. Seasonality will otherwise eat your result.

If it works on your catalogue you will see it, and you will have a number that is actually about your products rather than someone's aggregate.

Why this used to be impractical

None of the above is new thinking. The reason most small fashion stores have static product pages is not that nobody realised video helps. It is that video, historically, meant booking a studio, hiring a model, hiring a videographer, shipping samples, waiting weeks for an edit, and paying somewhere in the low hundreds per SKU. At those economics you shoot your hero products and nothing else, and the long tail of your catalogue, which is where the return rates are usually worst, never gets touched.

Generative video changes the arithmetic rather than the argument. When a clip costs a few credits instead of a few hundred dollars and arrives in minutes instead of weeks, the calculation stops being "which products justify a shoot" and becomes "which products would benefit", which is a much longer list.

Where to start

Not with your whole catalogue. The products worth doing first are the ones where the gap between the flat lay and the reality is widest: fluid fabrics, unusual cuts, anything oversized, and anything already carrying a return rate you dislike. We wrote a whole framework for choosing them in which products deserve video first.