AI Is Reshaping Zalando From Within

Artificial intelligence at Zalando is no longer a stand-alone technology experiment, but part of the way the platform connects customers with products and content. The company uses it in recommendations, product discovery, size selection, virtual fitting, marketing content production, and conversational shopping. Gilbert Kreijger, Zalando Communications Manager, who speaks to ICTbusiness Media and ICTbusiness.info about this transformation, says the company describes the process as increasingly precise “matchmaking” between customers and its assortment.

According to Zalando, this approach has already increased the number of items added to shopping bags by 13%. At the same time, its Size & Fit solutions reduced returns related to incorrect sizing by more than 8% during 2025. On the platform itself, as much as 90% of marketing content is now created with the help of generative AI. This has reduced campaign production time from six weeks to a few days while increasing content output by 70%. More than 10 million customers used Zalando Assistant in the first quarter, and between January and June the company recorded a 63% increase in high-value interactions. At ABOUT YOU, SCAYLE STUDIOS generated more than 80,000 outfits in just 30 days, with production costs around 90% lower and time-to-market more than 95% faster. Alongside the accelerating adoption of AI, Zalando emphasises human oversight, data protection and the need for EU AI Act rules to be implementable within realistic timeframes.

How does Zalando use artificial intelligence to improve product discovery and personalise the customer experience, and what results are you already seeing?

Our goal is to connect customers with exactly the product or content they will love – we call this matchmaking. Artificial intelligence (AI) plays a key role in achieving that goal.

The Zalando app offers an AI-powered discovery feed that brings together Zalando’s most powerful experiences: from curated boards and personalised product recommendations, to shoppable video and livestreams, as well as inspiring campaigns and high-quality content from brands, creators and Zalando itself. All of this is brought together in a single personalised feed that is easy to browse and tailored to the customer’s own interests.

We also offer AI-powered Size & Fit Solutions: our size advice recommends the right size by taking into account, for example, an individual customer’s previous purchases, preferred sizes and user feedback. Zalando’s Virtual Fitting Room uses AI-based technology to create a personalised 3D avatar based on a customer’s unique body measurements, allowing them to see how different sizes might look on their body shape.

Zalando’s matchmaking precision has already led to a 13% increase in the number of items added to shopping bags. Size & Fit solutions help customers choose the right size in advance, which reduced returns related to incorrect sizing by more than 8% in calendar year 2025.

How does Zalando Assistant fit into the personalised customer experience, and what measurable impact is it having on customer engagement?

Our AI-powered Zalando Assistant offers a personalised way to discover and interact with Zalando’s assortment. It understands context and your style: it is fully connected to the customer profile and account history. If you are browsing jeans and open the Assistant, it knows that you are looking at jeans and can offer specific advice on flared versus straight-leg styles, even without you asking.

The Assistant also enables customers to add items to their shopping bag and proceed to checkout directly from the Zalando Assistant interface itself. A customer can ask the Assistant to find an item that complements one already in their shopping bag.

In the first quarter of the year, more than 10 million customers used Zalando Assistant. Over the past quarter, it continued its evolution from a conversational tool into a proactive lifestyle companion. Between January and June, Zalando recorded a 63% increase in high-value interactions (for example, marking an item as ‘liked’ or adding it to a shopping bag). New releases introduced web search for trend-driven queries, deeper integration with customer support, and visual discovery through image uploads.

How extensively is GenAI already used in Zalando’s content production, and how do you ensure quality, accuracy and human oversight?

As much as 90% of Zalando’s marketing content on the platform itself is now generated using artificial intelligence. This means that most of the content you see when you open the Zalando website or app and browse the assortment – banners, highlighted trends, campaign visuals and teasers – is created using generative AI. This does not apply to product detail pages or advertising outside Zalando, but to editorial and marketing content on the platform itself, in other words our ‘inspiration layer’.

Of course, you will still see large human-led productions on the platform, such as our spring/summer campaign with Lily Collins. But while those major hero campaigns set our global seasonal direction, the remaining 90% of our on-platform content is focused on local relevance. It is fast-growing, trend-led content that keeps your everyday feed fresh. This has reduced campaign production time from six weeks to just a few days and increased content production by 70%.

Content created with the support of artificial intelligence is produced within a defined brief, follows the same brand, legal and content guidelines as any other Zalando material, and is reviewed by our creative and editorial teams before publication. People remain responsible for product accuracy, styling, brand presentation and the final visual result. The goal is not to automate creative judgement, but to use technology within a controlled production process that enables teams to create more relevant content without lowering quality standards.

What measurable impact has SCAYLE STUDIOS had on e-commerce content production, costs and speed to market?

For ABOUT YOU, SCAYLE STUDIOS has already become the backbone of e-commerce content production, with around 90% of production taking place through AI-based workflows. In the past 30 days, more than 80,000 outfits have been generated, with production costs around 90% lower and time-to-market more than 95% faster.

How does Zalando protect customer data and trust as it develops and deploys artificial intelligence?

Our approach to artificial intelligence is guided by a clear principle: technology should empower the customer, not take control away from them. Zalando processes customer data in accordance with all applicable data-protection laws at both national and EU level. We continuously discuss how to move at the speed of technological development while maintaining the highest ethical standards in the world. We are committed to our standards and will never sacrifice customer trust for a short-term benefit (‘quick win’).

How will the new EU AI Act requirements affect Zalando’s AI processes, and what do you see as the biggest practical implementation challenge?

We support the goal of increasing the transparency of AI-generated content. In the recently published guidance on the Artificial Intelligence Act (EU AI Act), we welcome the fact that the final guidance related to Article 50 also, to some extent, shows progress compared with the draft published by the European Commission in May, making it a more practical and useful tool. However, the start of application of the guidance on 2 August 2026 left companies with only 10 working days to act.

While we appreciate the recognition that the definition of ‘deep fake’ material depends heavily on context, which far better reflects the real-world application of the law, such a short implementation period makes detailed case-by-case assessments almost impossible. In addition to requiring significant legal, technical and operational adjustments within organisations, these context-dependent requirements carry a risk of inconsistent labelling of similar or identical types of content across the market. Despite companies’ best efforts to comply with the rules, these context-dependent requirements carry a risk of inconsistent labelling of similar or identical types of content across the market, despite companies’ best efforts to comply with the rules.

Given the exceptionally short implementation period, companies face significant practical challenges in adapting internal processes, governance frameworks and consumer-facing disclosures in time.

We therefore encourage the Commission to ensure that future guidance remains closely aligned with the letter of the Artificial Intelligence Act and that it provides a more realistic timeframe for implementation. This would help ensure consistent and effective enforcement of transparency obligations, while at the same time achieving the Commission’s goal of increasing consumer trust without creating an unnecessary compliance burden.