SHEIN tested fashion in 100-unit batches, not forecasts

Traditional apparel manufacturing commits to large production runs months before a garment reaches a store, based on forecasts of what will sell — buyers and designers guess, at scale, which colors, cuts, and styles a season's customers will want, and the entire industry accepts massive unsold-inventory writeoffs as an unavoidable cost of that guess being wrong as often as it is. More forecasting sophistication reduces the error rate somewhat, but the fundamental problem — committing capital to a prediction before any real customer has actually bought anything — doesn't go away.

SHEIN's '小单快返' ('small order, fast return/replenishment') model treats every new design as a live market test rather than a forecast: suppliers are contracted to accept minimum orders as small as 100–200 units, a design is produced at that tiny scale, photographed, and put up for sale online immediately, and the digital sales data from that small real batch — not a forecast, not a focus group, actual purchases — determines within days whether to reorder it at larger scale or drop it.

SHEIN tested fashion in 100-unit batches, reordering only real hits instead of betting on forecasts.

Small batches made real-market tests as cheap as guessing.

Actual purchase data beat any forecast as a signal.

More tests per dollar found more genuine hits.

Fast reorder cycles kept inventory writeoffs low.

When demand is unpredictable, shrink the test until it's cheap enough to run many real trials instead of betting on one forecast.

SHEIN scaled to an estimated $23 billion in 2022 sales and roughly $32.5 billion in 2023, adding between 2,000 and 10,000 new product listings daily at peak — a pace of style-testing volume unmatched by traditional retailers using forecast-driven production — while its valuation rose from roughly $5 billion in 2019 to as high as $100 billion in 2022 on the strength of the model's efficiency.