Shopping searches on generative AI grew 4,700% in one year

Shopping-related searches on generative AI platforms grew 4,700 per cent between July 2024 and July 2025, according to the State of Fashion 2026 report co-published by Business of Fashion and McKinsey & Company. Nearly a quarter of global consumers now use generative AI as their primary discovery starting point, whilst 41 per cent of consumers say they trust generative AI search results more than paid search ads. For fashion retailers, the implications extend beyond discovery to fundamental questions about visibility, product data architecture and traffic acquisition.

How much traffic are fashion retailers receiving from AI platforms?

The shift from experimental to structural is measurable in referral data. According to SimilarWeb, ChatGPT accounted for 16 per cent of Zara's and 8 per cent of H&M and Aritzia's inbound traffic between June and August 2025. 53 per cent of US consumers who used generative AI for search in Q2 2025 also used it to help them shop.

Adobe Analytics reports AI-sourced traffic to US retail sites grew 393 per cent year-over-year in Q1 2026, with AI-referred visitors converting 42 per cent better than all other traffic by March 2026. That conversion advantage represents a reversal: a year earlier the same AI traffic converted 38 per cent worse. Adobe also indicated that consumers arriving at US retail sites from AI platforms engage 14 per cent more than their counterparts, spend 59 per cent more time on site and bounce 33 per cent less, and add items to their cart 28 per cent more than other shoppers.

What is generative engine optimisation and why does it matter for fashion?

GEO ensures content is visible to and optimised for generative AI models. Unlike traditional search engine optimisation, which focuses on ranking in result pages, the process of optimising a generative engine for e-commerce involves organising product data so that AI assistants recommend and mention your product range, and differs from e-commerce SEO in that it targets the structured feed and page data that the model reviews before deciding on a product.

These shifts make AI chatbot responses the new SEO, according to McKinsey. Whilst search engines remain dominant, driving around 80 per cent of all search traffic, the rapid rise of AI-driven discovery highlights a fundamental change in how people find and evaluate products. LLM referral traffic in fashion and sport grew roughly twentyfold year on year through 2025, and during the 2025 holiday cycle, 96 per cent of fashion citations on Gemini went to non-retailer sources.

How should fashion retailers approach product feed optimisation for AI?

If product pages are the least machine-readable surface, and the feed is what both Google and OpenAI are instrumenting for conversational answers, the product data feed is where to start. The technical requirements differ from traditional shopping feeds.

• Product feeds enriched for AI with detailed, structured data including titles, descriptions, specs, attributes, and images that generative tools can parse and recommend, and granular product attributes such as compatibility, material, dimensions, and lifecycle tags so AI can confidently match your product to shopper intent
• Feeds plugged into Google Merchant Center with clean GTINs, taxonomy, and rich attributes
• Reviews and ratings in your feed, as the ChatGPT feed specification accepts product review count and average rating as feed-level fields
• Pricing synced across your feed, site, and all marketplaces so AI agents do not encounter mismatches, schema markup on every product page, and shipping and return policy fields

Deploy schema on every product page including Product, Offer, AggregateRating, BreadcrumbList, FAQPage, and Organization mapped to clean JSON-LD that agents can crawl. AI tools rely on normalised, literal product data to match results to user intent, improving relevance with more personalised product recommendations.

Which fashion retailers are deploying AI shopping tools?

22 million people visited the Zalando Assistant in 2026, up from just over 2 million in the first two years. Zalando reports that interactions with the assistant last three times longer than conventional searches and average more than two questions per conversation. European online fashion retailer Zalando unveiled its first AI shopping agent that will allow customers to buy products from other retailers through the Zalando app, with Zalando planning to launch AI-assisted checkout on its Zalando Extended platform in 2027, allowing customers to buy products from other online retailers without leaving the Zalando app, with AI agents helping to complete purchases.

At Zalando, the State of Fashion 2026 report says, generative AI reduced image production costs by 90 per cent. Revolve has reported significant increases in customer engagement from its experiments with generative AI-powered search.

What operational changes does this require from fashion commerce teams?

To ensure their products are visible to and favoured by AI models, brands must rethink their digital marketing and e-commerce infrastructures, where semantically rich data and API-accessible content will be critical to success. More than 35 per cent of executives report already using generative AI in areas such as online customer service, image creation, copywriting, consumer search, or product discovery.

The shift requires coordination across previously separate functions. AI systems accumulate information from editorial coverage, Reddit discussions, creator commentary, product reviews, structured commerce data, social media posts, and wider authority signals to generate conversational answers. ChatGPT and Claude lean on editorial such as Vogue and Business of Fashion, Reddit appears in the top five citation sources for all five major engines, and Wikipedia acts as a bridge node across the editorial-leaning models.

Unlike search engines that return ranked links, large language models like ChatGPT, Gemini, and Perplexity synthesise information and surface products based on contextual relevance, structured content, and third-party signals like reviews and editorial mentions. This means traditional SEO and paid search budgets may need reallocation. Adobe data shows AI-sourced traffic to US retail sites grew 393 per cent year over year in Q1 2026, and by March 2026, AI-referred visitors converted 42 per cent better than non-AI traffic.

What are the risks of not optimising for AI discovery?

Invisibility is the primary risk. A product page can look flawless to a human and remain invisible to an AI agent if the underlying data is sparse or inconsistent, as discovery in AI commerce is governed by data quality, not page design. Shoppers are increasingly willing to make purchases from AI-based recommendations, as seen in the 84 per cent growth in AI-driven revenue per visit on US retail sites between January and July 2025.

The industry is grappling with weak consumer sentiment and the rise of so-called agentic commerce, in which AI agents can handle the entire purchasing process on behalf of shoppers, from comparing prices across retailers to completing payment. In the years ahead, autonomous AI shopping agents may even act on consumers' behalf, completing tasks from monitoring prices to buying products. Fashion retailers without machine-readable product data risk exclusion from these automated purchase flows entirely.