Key takeaways
- At marketplace scale, the industry's 30% return rate is a P&L line, not a nuisance, and fit drives the largest share of it.
- Showroomprivé deployed Kleep Smart Sizing: per-garment recommendations from a shopper's own profile, via a quick questionnaire.
- Result: an 11% reduction in returns attributable to AI sizing, plus fewer bracketed orders and lighter reverse logistics.
- The mechanism was precision, a garment-specific answer shoppers trusted enough to order a single size.
- Single-digit percentages at large volume are substantial money: the ROI compounds with every avoided round trip.
The challenge
Showroomprivé operates at scale across a vast, fast-moving fashion assortment, flash-sales dynamics, many brands, constant catalogue turnover. At that volume, the 30% return problem is not a nuisance. It is a major line on the P&L, with reverse logistics, repackaging and restocking costs attached to a third of everything shipped.
The flash-sale model sharpens the problem further: assortments rotate quickly across many brands, each with its own sizing behaviour, so shoppers can rarely lean on familiarity. Fit and size drove a large share of returns, as they do across the industry. The opportunity was clear: help shoppers buy the right size the first time.
What they deployed
Showroomprivé added Kleep Smart Sizing, giving shoppers a size recommendation built from their own profile and matched to each garment's real measurements, not a brand-wide chart, but an answer specific to the item on the page.
Shoppers could get to a confident size through a short questionnaire rather than wrestling with a static chart, removing the guesswork that leads to bracketing. Critically for a high-velocity catalogue, the approach works garment-by-garment, so a new brand entering the assortment gets accurate recommendations without waiting to accumulate purchase history.
The result
- An 11% reduction in returns attributable to AI sizing.
- Fewer bracketed orders, which eases reverse-logistics load.
- More confident shoppers at the point of decision.
- A return on the deployment that compounds with every avoided round trip.
Why it worked
The win came from precision. A recommendation tied to the real garment, not a brand-wide range, gives shoppers a reason to trust it and order a single size. Trust is the operative mechanism, the psychology we unpack in Why Shoppers Don't Trust Size Guides, and it is earned by being specific and right, repeatedly.
At Showroomprivé's volume, even a single-digit percentage cut in returns translates into a substantial operational and financial saving: fewer collection legs, fewer inspection hours, fewer garments missing their selling window. Small percentages, large numbers.
What it signals for the wider market
Showroomprivé is one data point in a consistent pattern. ETAM's A/B test showed +5% revenue and −16% size-related returns; ba&sh's head-to-head against True Fit showed +88% add-to-cart; across deployments, returns typically fall 20 to 50% on assisted purchases.
The volume curve tells the adoption story: 17k Kleep sizing flows in 2023, 500k in 2024, 4.2 million in 2025, one every 5 seconds in 2026, across 300+ brands from Givenchy to Represent. Real-measurement sizing has crossed from experiment to infrastructure.
How did Showroomprivé reduce its return rate?
By deploying Kleep Smart Sizing: shoppers complete a short questionnaire and receive a size recommendation matched to each garment's real measurements. Returns attributable to sizing fell 11%, with fewer bracketed orders easing reverse logistics.
Does AI sizing work for marketplaces with many brands?
Yes, arguably best. Because recommendations are computed per garment from real measurements rather than from accumulated purchase history, new brands and fast-rotating assortments get accurate answers from day one, which suits flash-sale and marketplace models.
Is an 11% return reduction a good result?
At marketplace volume, yes: each avoided return removes two shipping legs, handling labour and restocking risk, so single-digit percentages compound into substantial savings. Deployments focused on assisted purchases typically see 20 to 50% reductions on that segment.
How long does it take to see results from a sizing deployment?
Returns data lags purchases by the length of the return window, so expect a clean read within one to two return cycles. Leading indicators, usage, add-to-cart, single-size ordering, move within weeks.
Conclusion
Showroomprivé's 11% return reduction shows what real-measurement sizing does at scale: it turns a structural cost into a managed one. Give shoppers a size they can trust, and the returns that fit anxiety creates simply stop happening.
