How Sizing Recommendations Work on Zalando

Zalando SE Sizing

See how our Sizing team uses data to create a more personalized shopping experience.

It’s no secret that Machine Learning (ML) and Artificial Intelligence (AI) are changing the fashion industry. At Zalando, customer-focused improvements are being driven by machine learning and enabled by AI in multiple ways. One way algorithms are building innovative products at Zalando is through sizing recommendations.

Across any fashion e-commerce assortment, about 10% of articles have sizing issues.


This is especially true when dealing with multiple brands and manufacturers as sizes often vary from brand to brand.


To solve this challenge, the sizing team at Zalando has developed an algorithm to identify these articles and advise customers on whether they should buy one size up or one size down before they make a purchase.

Zalando SE Sizing

Additionally, the algorithm can identify when a customer is purchasing a brand which he or she has not purchased before and let the customer know if shoes or clothing from that particular brand are often either larger or smaller than what customers might typically expect.

The key to building such a recommendation engine is, of course, access to significant amounts of data. Our algorithm uses two sets of data: customer return data and in-house fittings

Return Data

We collect return data when customers return a product claiming that it is too large or too small. Our return centers record this data and submit it for collection in our database.

Zalando SE Sizing

In-House Fitting Data

Through our in-house process, articles are physically tried on by trained fitting models and judged in terms of size. This helps to identify sizing issues in articles as soon as they are added to the assortment, compared to the data from returned articles which is subject to the delay incurred through the time required from going online to purchase to return.

Additionally, the algorithm can identify when a customer is purchasing a brand which he or she has not purchased before and let the customer know if shoes or clothing from that particular brand are often either larger or smaller than what customers might typically expect.

Providing a Size Recommendation Increases Customer Confidence

When customers are confident that the items they are purchasing will fit, they buy more frequently (increasing conversion for brands) and they buy more. Additionally, the size flag reduces size-related return rates (currently by 1%), since customers can purchase the right size the first time.

Zalando SE Sizing