Zalando: a generative AI shopping assistant at scale

Sector: Fashion e-commerce · Region: Germany, Austria, UK, Ireland

Case digest compiled by BAICA from public sources, July 2026

Context

Zalando is one of Europe's largest online fashion platforms, operating across more than 20 markets. Fashion discovery is a hard search problem: customers often know the occasion ("a wedding in Santorini in July") rather than the product attributes a keyword search needs. In late 2023, Zalando became one of the first major European retailers to put a generative AI assistant in front of customers at scale.

What was deployed

The Zalando Assistant, launched in beta around December 2023, is a conversational interface built on large language model technology. Customers describe what they need in natural language and the assistant translates that into product recommendations, holding context across the conversation. By mid-2024 it was live in Germany, Austria, the UK and Ireland (Zalando Corporate, June 2024). Zalando paired it with Trend Spotter, an internal AI tool that tracks emerging fashion trends weekly across six European fashion capitals using engagement signals — searches, likes, add-to-carts — rather than sales data alone.

Reported results

According to Zalando's own June 2024 disclosure, more than 500,000 customers had engaged with the assistant in roughly its first six months. Two behavioural findings stand out. First, queries put to the assistant were about three times longer than queries typed into the standard search bar. Second, conversations averaged about four back-and-forth exchanges (Zalando Corporate). Zalando has not published revenue or conversion figures attributable to the assistant, and this digest does not estimate them.

What to learn

  • Conversational interfaces surface intent that keyword search never captures. Three-times-longer queries are three times more signal about what the customer actually wants.
  • Launch narrow, in beta, in selected markets. A staged rollout let the team improve context-tracking based on real customer feedback before expanding.
  • Honest metrics first. Note what was measured publicly at this stage: engagement and query behaviour, not revenue claims. Early-stage AI projects should be judged on learning metrics before commercial ones.

Source: Zalando Corporate Technology, "Inspiring and empowering customers with AI-powered experiences", June 2024.

Add your case to the library.

Failed projects teach as much as successful ones, and we publish both. Cases are anonymised on request and fact-checked before publication.