• Fashion United/ Regina Henkel

Fewer returns: How Zalando plans to eliminate fitting issues from shopping online


Online retailer Zalando wants fewer customers to return the clothes they purchased due to fit problems. About one third of all returns are due to clothes being too small or too large, and that, according to Zalando, could be optimized. If customers would receive the right size right away, they would be happy and Zalando would save costs. It would also be good for the environment because fewer goods would be out on the roads. As an IT group, Zalando has many ideas on how to create more accuracy between consumer and product when it comes to fit - but most of them are still future dreams. Until they are ready to be implemented, Zalando uses surprisingly conventional methods. We talked with Stacia Carr, director of engineering/sizing at Zalando, about their plans.

Why are fit problems so relevant for Zalando? What are the costs of fit-related returns?

In general, returns are not a problem for Zalando. We see them as an important part of the promise to our customers. If they don't like an item, they can return it, the same way they can put an item back on the shelf in an offline store. Through improvements in the online customer experience, however, the number of size-related returns can be reduced. While an offline environment allows customers to try on articles, online shopping includes the extra step of returning an item. This additional step requires that we examine how we can provide the customer with a size and fit recommendation in order to avoid unnecessary returns.

Internationally, where do most articles get returned because of fit issues?

One third of all our returns are due to size in our 17 markets. We do not share this information at market level.

Consumers are used to convenience. How do you encourage your customers to put more effort into size selection when ordering a product in different sizes is easier and free?

For customers who want to order several sizes and send most of them back, this option will continue to be available free of charge; this is an important part of our customer promise. We would like to advise customers, however, to improve their experience. Through our sizing team, we do this by consultations on size and fit. If we can convince our customers that our sizing consulting services offer added value to their experience, then we hope that they will be able to get the right size with their first order. We hope this will give them the confidence to discover new brands and looks.

Other online retailers are already restricting their free shipping. Is this also an option for Zalando?

Like many other providers, we have introduced a minimum order value in some markets. This is currently not the case in Germany but in the UK where it is 19.90 pounds.

In recent years, there have been various ideas about how to reduce the number of fit-related returns. Previously, the problem was comparing customer data with product data. How do you get the measurements of the clothes and the customers?

Our consultation for current sizes depends on a deep understanding of the products that make up Zalando's extensive range. We offer our customers size consultations for about 80 percent of our selection, in two ways. The first one is a size marker that informs the customer whether certain article sizes are larger or smaller than standard sizes. This way, a customer can make the choice to order one size bigger or smaller. The second form of consultation is a personalized size recommendation, for which we actively recommend a certain size based both on our understanding of the product and on the customer’s purchase history and the preferences in size and brand.

In the future, we will also research new technologies such as visual adjustment. The key to this are a customer’s body measurements and the willingness to provide us with this information. We explore methods such as the use of smartphones to create 3D images of customers and to enable them to choose avatars that match their profile. But all this is still in the future. The goal is to leverage our expertise in product data with a better understanding of our customers to offer them a better customer experience.

How does Zalando find out the measurements of a product? Are they supplied by the manufacturers or other data suppliers?

In our experience, measurements given by the manufacturers are not always reliable. There are many reasons for this, for example, if the material of the end product differs from the one of the sample. That's why our team has fit models test those 80 percent of articles that are offered to find out if they are made to measure or not. These products are tried on and measured. All this contributes to a better understanding of our products. We are also researching technologies here, such as a computer vision, to determine measurements and whether an object is likely to have size-related problems or not, based on a machine-learning algorithm.

First attempts have been made to measure the body on the basis of an extensive series of measurements of people in different countries and to derive fit data from it. Do you work with such projects?

We work with many different teams throughout the entire fashion industry that research sizes and related parameters. At the moment we wouldn't publicly announce any such collaboration.

Just measuring the body and clothing is probably not enough to define the right fit in clothing, especially skinny fits. How do you solve the problem?

Our team of fit models helps us with that. We have about 30 models who try on a variety of clothes from our selection: Shoes, jeans, dresses, shirts and many other articles. They then decide whether an object has the right dimensions or is actually slightly larger or smaller than the specified size. Our models know how skinny fit articles should fit and give us the necessary input so that we can decide if we want to give a size recommendation for the article or not.

Are there any solutions for calculating the stretch effect of fabrics and to integrate it into fit models?

Stretch depends strongly on the fabric composition; we are investigating different ways to obtain data on the composition of materials and to derive rules from them. It is still too early to talk about these experiments.