Fashion Tech
Deep Dive
How Size Engines Work
Size recommendation engines combine multiple data sources: customer body measurements (from profiles, past purchases, or body scanning), product technical specifications (garment measurements by size), fit preference data (loose vs. fitted), and collaborative filtering (what size worked for similar body types). Machine learning models continuously improve accuracy as they process more purchase and return data.
Business Impact
Effective size recommendation reduces returns by 20-40% and increases conversion rates by 10-15%. For a fashion e-commerce brand processing $100M in annual revenue with a 30% return rate, a 10-percentage-point reduction in returns can save $3-5M annually in logistics costs alone — plus the margin recovery from items that can be resold at full price instead of being returned and marked down.
Implementation Challenges
Size recommendation faces challenges including inconsistent sizing across brands, limited garment measurement data, the subjective nature of fit preferences, and the cold-start problem (no data for new customers). The most successful engines build large databases of garment measurements and invest in gathering explicit fit feedback from customers to continuously train their models.
OSF Perspective
OSF considers size recommendation one of the highest-ROI technology investments in fashion e-commerce. It simultaneously improves customer experience, reduces operational costs, and contributes to sustainability by preventing the enormous waste associated with fit-related returns.
Related Terms
Body Scanning | Virtual Try-On | Return Rate | Conversion Rate
Notable Brands
True Fit, Fit Analytics (acquired by Snap), 3DLOOK, Sizely