Return rates for online apparel have been a known crisis for more than a decade. Brands see 30 to 40 percent of what they ship come right back. The cost — in reverse logistics, restocking labor, packaging waste, and lost margin — runs into the billions. The industry has tried size guides, fit quizzes, detailed measurement charts, and customer reviews. None of it moved the needle.

Now a new generation of AI startups is trying harder. CNBC reports today that 17 AI-powered companies have raised more than $400 million in the last five years building what they call the answer to retail's "silent killers" — returns and fit uncertainty. What's changed is the underlying technology, which has become genuinely capable in ways it wasn't two years ago.

The Core Problem AI Is Attacking

The primary driver of returns isn't buyers' remorse or impulse purchases. It's fit uncertainty. Shoppers don't know whether a medium in one brand matches a medium in another, whether a garment runs loose or structured, or whether that dress will fall the right way on their specific body. They order two sizes, keep one, return the other — or return both.

According to industry analysis, retailers that deploy virtual try-on with accurate size recommendation see return rates drop 30 to 40 percent. Add a body-measurement layer and you get an additional 15 to 20 percent reduction in fit-related returns on top of that base. If those numbers hold at scale, this is a meaningful margin recovery story — not a feature.

What the Technology Actually Does Now

Platforms like Catches have built tools that let shoppers create a "digital twin" — a virtual representation of their body — and see how specific garments fit on that model. Business of Fashion noted that generative AI has dramatically improved the visual fidelity of these systems; the photorealistic rendering gap that made earlier versions look cartoonish has largely closed.

Other startups are building in the background, on the backend sizing infrastructure rather than the consumer-facing interface. Their value proposition: plug into a retailer's existing product catalog, ingest garment measurements, and serve size recommendations so accurate that the "just order two sizes" behavior becomes unnecessary.

Google took its own virtual try-on technology mainstream this spring, making it accessible directly within product search results across Google platforms. When Google normalizes a technology, the window for startups to own it exclusively narrows fast.

The Commercial Logic

Zalando, the European fashion marketplace, has been perhaps the most aggressive early adopter, using virtual try-on at scale and reporting return rate reductions that validate the technology's commercial case. What's useful about Zalando as a test case is its scale — the results aren't from a boutique with highly curated sizing but from a platform handling hundreds of brands with inconsistent sizing conventions.

The $400 million in startup investment makes sense against this backdrop. If you can demonstrably cut a 35% return rate to 22%, you're not selling a feature — you're selling a margin improvement that shows up directly in the P&L. For an apparel brand doing $50 million in online revenue, that math can justify significant technology spend.

The more interesting competitive pressure is on what happens to mid-tier brands and multi-brand marketplaces that don't invest. If the brands deploying AI try-on are capturing the confident buyers while their competitors are absorbing the "order two" traffic, return rates become a competitive moat, not just an operational cost.

What's Still Unresolved

Adoption is the gap. Most of the brands that need this technology haven't deployed it yet. Industry tracking suggests that despite the technology's maturity, the majority of fashion e-commerce still relies on static size guides and customer reviews to drive fit decisions. The implementation friction — integrating with existing product data, photography standards, and checkout flows — remains a real barrier for mid-size brands without strong technical teams.

That's ultimately the market opportunity the $400 million is chasing: not the Nordstroms and Zalandos who can build their own, but the tens of thousands of brands selling on Shopify and wholesale marketplaces who can't.