You’re a Medium. Except When You’re Not.
Try this experiment. Walk into Zara, pick up a medium t-shirt, and note how it fits. Then walk into H&M and grab another medium. Different fit. Now try the same at Nike. Three brands, three completely different garments, all labeled "M."
This is not a minor inconvenience. It is a structural failure of the fashion industry, and it costs everyone money.
How We Got Here
Standardized clothing sizes were first introduced in the US during the 1940s, based on a study of around 15,000 women. The data was flawed from the start, as the sample was not representative. By the 1980s, the US government dropped its voluntary sizing standard entirely. Brands were free to define sizes however they wanted, and they did.
Then came vanity sizing. Brands discovered that customers feel better buying a smaller number. A size 8 from 1970 is roughly a size 4 today. The waist measurement didn’t change, just the label.
Regional differences add another layer. European, Asian, and American sizing systems use different base measurements. A French 40 is roughly a UK 12, which is roughly a US 8. "Roughly" is doing a lot of work in that sentence.
$25 Billion in Wasted Effort
In the US alone, size-related returns cost retailers about $25 billion per year. That figure covers shipping, processing, repackaging, and the roughly 30% of returned clothing that never gets resold at full price. For online-only brands, return rates sit between 30% and 40%, with sizing as the top reason.
Each return burns money. A $50 item costs the retailer $10-15 to process the return. Multiply that across millions of orders and you have a problem that eats directly into margins.
Customers lose too. They order two or three sizes, keep one, send back the rest. It wastes time and creates friction that pushes people back to physical stores.
Solution 1: AI Size Recommendation
The most effective fix right now is AI-powered size recommendation. Here is how it works in practice.
The customer inputs basic data: height, weight, age, and fit preference (loose, regular, or tight). The system cross-references this against purchase and return history from thousands of other customers with similar profiles. It then recommends a specific size for that specific product.
The results are significant. Brands using AI size recommendation report a 50-60% reduction in size-related returns. The system gets smarter over time because every purchase and return adds data to the model.
Companies like Fit Analytics (acquired by Snap), True Fit, and Sizely already offer this as a plug-in for e-commerce platforms. Integration typically takes 2-4 weeks.
Solution 2: Virtual Try-On
Virtual try-on gives customers a visual confirmation of how a garment will look on their body. Using a photo or a 3D avatar based on the customer’s measurements, the technology overlays the clothing item and shows how it drapes and fits.
This is not a replacement for size recommendation. It is a complement. The AI tells you "get a size L," and the virtual try-on shows you what that size L actually looks like on your frame. Together they cover both the rational and the visual side of the buying decision.
Solution 3: Better Size Charts With Real Measurements
This one is surprisingly simple but most brands still get it wrong. A good size chart lists actual garment measurements in centimeters: chest width, body length, sleeve length, shoulder width. Not "fits sizes 8-10," which tells the customer nothing.
Adding a reference model ("Model is 175 cm, wears size M") helps anchor the information. Including flat-lay measurement diagrams removes ambiguity about where exactly the measurement is taken.
What Brands Should Do Right Now
The sizing problem will not disappear on its own. Brands waiting for a universal standard will wait forever. Here is a practical starting point.
First, add AI size recommendation to your product pages. The ROI is immediate and measurable through reduced returns. Second, publish detailed measurement charts for every product, not generic brand-wide charts. Third, explore virtual try-on for high-return categories like dresses and outerwear. Fourth, collect fit feedback after delivery and feed it back into your recommendation engine.
The technology exists today. The brands that adopt it first will keep more customers, process fewer returns, and spend less on reverse logistics. The ones that don’t will keep paying the $25 billion tax.