Why 40% of clothes bought online get returned

Somewhere in a warehouse right now, a worker is unpacking a dress that fit perfectly. On the size chart, anyway. In the fitting-room mirror of the buyer's bedroom, it didn't. Back in the box it went, joining an estimated 30–40% of all clothing bought online that makes the return journey. For in-store purchases, that figure is under 10%. The difference isn't buyer's remorse. It's that when we shop online, we're asked to guess, and the game is rigged against us.

The vanity sizing spiral

In 1958, the US government published a voluntary commercial standard for women's clothing sizes, giving each size number a defined set of body measurements. It was revised in 1970 and withdrawn entirely in 1983, by which point most brands had long since stopped following it. In the decades since, sizes have drifted steadily downward in a phenomenon researchers call vanity sizing: as average bodies grew, the numbers on the labels shrank to flatter us. A women's size 12 from 1958 is roughly a size 4–6 today. The label stopped being a measurement and became a marketing decision.

The result: a "medium" is not a fact about a garment. It's a brand's opinion about who they want wearing it.

Every brand speaks its own language

Even setting vanity sizing aside, brands genuinely cut for different bodies. Zara designs to a slim continental European block (narrower shoulders, shorter torso). Nike cuts athletic and runs footwear about half a size small. American heritage brands cut generously through the chest and seat. None of them is wrong; they're just designing for different people while using the same size labels.

So the shopper who is a confident M at H&M orders an M at Zara and gets something they can't button. They conclude, reasonably but wrongly, that their body is the problem. Their body was never the problem. The label was.

Size chart chaos within a single brand

Here's the part that surprises even seasoned shoppers: the same brand often doesn't fit like itself. A brand's denim line, its youth line, and its main collection are frequently developed on different fit blocks, in different factories, on different continents. H&M's Divided line famously runs a full size smaller than its basics. Two "size 10" dresses from the same retailer can differ by more than an inch in the bust, well within normal manufacturing tolerance.

Size charts, meanwhile, describe the body the garment was designed for, not the garment itself, and they rarely mention ease, stretch, or cut. A chart can be technically accurate and still lead you to the wrong size.

What all those returns actually cost

Returns feel free, but somebody pays. Retailers spend an estimated $20–$30 to process a single returned garment: shipping, inspection, repackaging, restocking. Industry analyses put the annual cost of returns in the hundreds of billions of dollars, and a significant share of returned clothing is never resold at all: it's landfilled or incinerated because processing it costs more than it's worth. The environmental bill (double shipping, packaging waste, destroyed inventory) lands on all of us.

And shoppers pay twice: once in the prices that absorb all that processing, and again in the quiet tax of time: ordering three sizes to try at home, queueing at the drop-off point, waiting on refunds.

The fix isn't better size charts. It's no size charts.

Every failure above has the same root cause. A size label is a lossy, brand-specific compression of the only data that matters: the measurements of the garment and the measurements of you.

That's the premise we built Attirel on. You measure yourself once, with a guided tutorial. Attirel compares your actual dimensions against each product, factoring in how the brand runs, the garment's cut and stretch, and feedback from how past purchases fit you. Then it puts a fit score with a plain-English confidence label on every item, at every store you shop. When the data is thin, it says so, instead of letting you guess.

The 40% return rate isn't a shopper problem. It's a translation problem, and translation happens to be exactly what software is for.