Whether asking for advice on what to wear for a specific occasion or the best silhouette to suit their body type, more customers are turning to AI chatbots for shopping than ever before.
While AI-driven traffic is often estimated to account for less than 1 percent of global web visits, traffic from AI chatbots to fashion brands and retailers is growing fast. Referral traffic from AI to US retailers increased 760 percent year on year in November 2025, according to research by Adobe, which also found that shoppers who arrive via AI are more likely to complete a purchase than those coming from other channels. Some brands like Zara and H&M get a meaningful share of referral traffic from AI, accounting for 15 percent of Zara’s referral traffic and 8 percent of H&M’s between February 2025 and January 2026, according to SimilarWeb.
There is no doubt that this is changing the top of the marketing funnel for fashion brands. But where the picture is less clear is exactly how and why brands are showing up in AI large language models (or LLMs), where the rulebook is different from traditional search and users can get a different result each time — even with the same prompt.
BoF Insights Brand Pulse, a new tracking tool to measure fashion brand performance in the age of algorithms and AI, developed by BoF’s data and advisory team BoF Insights in partnership with Quilt.AI, deploys a proprietary methodology to assess how LLMs understand, rank and prioritise fashion brands, as one part of a wider measure of brand performance.
For this analysis, we studied the high street segment, because comparability can be high and loyalty low for high street brands, encouraging customers to turn to AI when looking for a particular product, rather than go straight to their favourite brand.
We assess fashion brand performance in LLMs along four distinct dimensions. These reflect how ChatGPT, Gemini and Claude recall, select and prioritise brands when responding to shopping-related prompts. The results reveal the top ranked high street brands by LLM performance across the four dimensions.
Our findings indicate two different paths to maximise brand performance within LLMs: brand positioning that is broadly useful or sharply defined.
Download ‘Which High Street Fashion Brands Are Winning the AI Discovery Battle?’ to learn more.