Case: Small Apparel Brand on Wildberries Cuts Returns by a Third

Apparel is the biggest category on Wildberries and the hardest one for a foreign seller to get right. This representative story follows a small apparel brand on Wildberries selling leggings and casual dresses; the brand is a composite of real seller scenarios and the figures are generalized, but the problems and fixes are the ones we see week after week.

The brand and the problem

The brand was run by a two-person team sourcing from a Guangzhou factory: seamless leggings in eight colors and a line of jersey dresses in four styles. They had done well on a Southeast Asian marketplace and brought the same assets to Wildberries: studio shots on a plain background, size tags in S/M/L, and a short description translated into Russian.

Orders came in, but so did returns. Wildberries clothing typically sees high return rates because shoppers order several sizes to try on at a pickup point, and the brand was running well above the category norm, around 55% on leggings and over 60% on dresses. Every return meant logistics fees both ways and a product that had been handled. With about 9 orders a day and more than half coming back, the unit economics were close to zero.

Reviews told the story: «размер не соответствует» (size does not match), «на фото выглядит иначе» (looks different from the photo), «непонятно, какой рост» (unclear what height it is for).

Fix 1: a size chart Russian shoppers can use

The first change had nothing to do with design and everything to do with data. Russian shoppers think in Russian sizes (42, 44, 46, 48) and in centimeters, and for dresses they expect a model height reference. S/M/L alone is not enough.

The team measured every item flat, built a table with Russian size, bust, waist and hip in centimeters, item length and the height the fit was designed for, and had it rendered as a dedicated infographic card in Russian. The same values went into the structured attributes of the listing so the size filter would work. They also added a plain line to the description: «Модель на фото ростом 170 см, на ней размер 44».

Fix 2: lifestyle cards instead of a catalogue

A plain studio shot tells a shopper what the garment is; it does not tell her how it will look on her. The team generated a full card set for each style: a main image with the key benefit in Russian («не просвечивают, держат форму» for leggings), a lifestyle scene in a realistic setting, a macro of the seam and fabric texture, a card showing all available colors, and a card listing composition and care in Russian.

Fabric composition deserves its own note. Russian shoppers read it closely, and the legally required label needs percentages. Writing «92% полиэстер, 8% эластан» on a visible card did more for trust than any slogan.

Fix 3: let AI Autopilot choose the main image

The team had three candidates for the leggings' main image: a close-up of the fabric, a full-body lifestyle shot, and a studio shot with a large Russian headline. Nobody could agree, and guessing is expensive when the main image drives most of the clicks.

Instead of arguing, they switched on AI Autopilot, which rotates main-image variants on the live listing, collects impressions, clicks and orders for each, and keeps the winner once the difference is statistically meaningful. Over about ten days the lifestyle shot with a short Russian headline won on CTR by roughly 35% over the plain studio image. For dresses the result was different: the studio shot with the size reference in the corner won, probably because it answered the shopper's biggest fear up front. That is exactly why you test instead of assuming.

Results after six weeks

• **Returns** on leggings fell from about 55% to around 37%, and on dresses from over 60% to about 42%. Still high by other categories' standards, but close to the Wildberries clothing norm.
• **Daily orders** grew from about 9 to roughly 21, driven by higher CTR and a conversion rate that climbed from about 2% to around 3.5%.
• **Reviews** shifted. Size complaints nearly disappeared, and the average rating moved from 4.2 to 4.6, which in turn improved search position.
• **Net margin per order** roughly doubled because fewer return legs ate into it.

The total spend on new visuals for twelve styles was under $100, and the full set for a new color variant now takes about three minutes to generate.

What this means for your apparel listings

• Publish a Russian size chart in centimeters, with model height, as both an image and structured attributes.
• Show the garment on a body and in context, not only on a hanger or a mannequin.
• Put composition and care in Russian on a visible card.
• Prepare two or three main-image variants and let the data decide; the winner often differs by product type.
• Watch returns as closely as orders. For apparel they are the number that decides whether you make money.

For the broader picture, read what Russian shoppers expect from a card and our guide to A/B testing the main image with AI Autopilot.

Want to rebuild your own apparel cards the same way? Start a 7-day trial for $1, upload a photo of your bestseller, and have a Russian card set with a size chart ready before your next shipment leaves the factory.