Nike Fit uses computer vision to cut sizing-related footwear returns

Background

Online footwear retail faces a persistent sizing problem: customers order multiple sizes, return what does not fit, and generate reverse logistics costs that erode margins. Nike estimated that roughly one-third of its online returns were size-related, and that 60% of customers were wearing an incorrect shoe size at the time of purchase. Traditional size charts and customer self-measurement were inadequate at scale.

What Was Implemented

• Nike Fit integrated into the Nike mobile app, using the smartphone camera as a scanner
• Computer vision captures 13 data points mapping the full morphology of both feet
• Machine learning models translate foot-scan data into per-model size recommendations (accounting for variation by silhouette)
• AI improves recommendation accuracy over time as the foot-morphology database grows
• Deployed globally as a consumer-facing feature

Results

Nike Fit was designed to reduce sizing-related returns and increase customer confidence at the point of purchase. At launch, Nike cited a baseline problem of 500,000+ annual sizing-related customer service calls and estimated ~60% of consumers wearing wrong sizes. The book's claim of "up to 60%" returns reduction and secondary-cited figures of a 28% reduction in online returns are both unverified — not found in primary sources fetched . Nike has not publicly disclosed post-launch return-rate data. The qualitative benefit — reduced sizing anxiety, fewer wrong-size purchases — is directionally supported by the technology's design.

Lessons

• Sizing is a root-cause driver of online apparel and footwear returns; AI-powered measurement addresses it at the moment of purchase
• Foot-morphology databases built at consumer scale become proprietary competitive assets over time
• Vendor-reported metrics at product launch should be verified against post-deployment disclosures before treating them as established results