What is size prediction in fashion?
Size prediction uses data and machine learning to recommend the size that is most likely to fit a specific customer for a specific product.
In short
Size prediction is an AI-based method that recommends which size a shopper should choose for a particular garment. It analyses information such as the customer's past purchases and returns, body data they provide and the product's measurements, then suggests the size most likely to fit.
How does it work in practice?
A size prediction tool usually appears on the product page as a find my size feature. The customer may answer a few questions about height, weight, body shape or preferred fit, or the system may use their history with the retailer. The model compares this with garment measurements and with how similar customers rated the fit of the same product.
Data sources often include:
- Product measurements and grading rules from the tech pack.
- Purchase and return history, especially returns marked as too small or too large.
- Customer input such as body measurements or reference brands.
- Fit feedback from reviews and post-purchase surveys.
Why does it matter for fashion businesses?
Fit problems are a frequent reason for online returns, and returns are costly to handle and often cannot be resold at full price. Better size advice can lower these costs and improve customer confidence. Aggregated fit data also gives product teams and wholesale brands evidence on whether a style runs small or large and whether the size curve offered to retailers matches demand.
How is AI changing it?
Machine learning allows models to learn from large volumes of fit outcomes rather than relying only on static size charts. Newer approaches use body scans from phone cameras or combine size data with style preferences, so the tool can distinguish between customers who like a loose fit and those who prefer it fitted.
Common pitfalls
- Inconsistent garment measurements across suppliers and seasons.
- Cold start for new customers and new styles without history.
- Misread returns, since not every size-related return is a fit problem.
- Privacy concerns when collecting body data.
Frequently asked questions
How does size prediction reduce returns?
It steers customers towards the size most likely to fit, reducing the need to order several sizes and send back the ones that do not fit. The effect depends on data quality and how many shoppers use the tool.
Can wholesale brands benefit from size prediction?
Yes, indirectly. Fit insights from retail channels can inform grading, size curve recommendations for retailers and product development for future seasons.