The Maze: Fashion's new front door is already concentrating traffic. Across 11 retailers tracked by RETAILBOSS from January through June 2026, SHEIN averaged 750,000 monthly referrals from AI platforms. H&M reached 451,000 and UNIQLO 402,000. That makes SHEIN's lead roughly 1.7x H&M's. It is a striking distribution signal, but not a victory lap: the source does not disclose geography, platform mix, attribution rules, or conversion.
The leaders occupy a different scale from the rest. The top four — SHEIN, H&M, UNIQLO and Mango — generated about 1.88 million of the 2.18 million monthly visits across this selected set, or 86%. Fashion Nova followed at 119,800. Then the curve collapsed: Aritzia reached 46,600, boohoo 42,000 and House of CB just 15,000. This is not a gentle ranking. It is a winner, two large challengers, one middle player and a long tail fighting for inclusion.
AI compresses discovery before the click. Search engines can show dozens of links; assistants tend to synthesize a short answer and name a small number of brands. That changes the scarce asset from page rank to recommendation eligibility. Similarweb measured ChatGPT referrals jumping 157.7% after more prominent brand links appeared on May 7, 2026, while homepage referrals rose 354.7%. Roughly 60% of those visits then landed on homepages. The assistant increasingly sends a shopper to the brand's front door, not to one keyword-matched product page.
Machine-readable product knowledge is becoming acquisition infrastructure. A model has to connect assortment, fit, price, availability, returns, reviews and brand positioning before it can recommend a retailer with confidence. Adobe found U.S. retail homepages averaged 75% machine readability, category pages 74% and product pages only 66%. A large paid-media budget cannot repair fragmented product information inside an AI answer. Retailers need structured first-party content and consistent third-party evidence so the same brand can be interpreted reliably across prompts.
Traffic leadership is not commercial leadership. The RETAILBOSS ranking omits its method and does not include Zara, so it cannot represent the complete fashion market. The January-June average also straddles ChatGPT's May referral-design change, mixing two channel regimes. Semrush estimates AI traffic grew 66% in 2025 but still represented only 0.14% of total web visits. The channel is fast, small and volatile. Operators should keep three scorecards separate: whether the brand is named, whether the answer sends a visit, and whether that visit produces revenue.
Why it matters: SHEIN's advantage shows that AI discovery has moved beyond an SEO experiment, but the next management mistake is obvious: treating an immature referral metric as a sales metric. Fashion retailers should audit what assistants say, repair the product and policy information machines cannot parse, and connect AI referrals to landing pages, conversion and revenue. The prize is not simply appearing in an answer. It is turning a compressed recommendation into a measurable customer journey.
Sources: Malte Karstan / RETAILBOSS | Adobe Digital Insights | Similarweb | Semrush
