For most of the history of online shopping, discovery meant a search results page, a marketplace grid or a social feed. Each showed many options and let the shopper choose. AI assistants work differently: they read on the shopper's behalf and return a short, reasoned list. In 2025 and 2026 that behaviour moved from experiment to habit, and fashion and beauty are among the categories where it is growing fastest.
1. AI is now a measurable traffic source
Referrals from ChatGPT, Perplexity, Gemini and similar tools are still a small share of total retail traffic, but the growth rates are steep:
- Adobe measured a 693% year-over-year increase in traffic from generative AI tools to US retail sites over the 2025 holiday season, and forecasts a further 130% rise for the 2026 season.
- Similarweb reported AI referrals growing 278% in fashion and 312.5% in beauty year over year (June 2025 to May 2026, worldwide).
- Similarweb's State of Ecommerce 2026 found that a brand recommended by an AI assistant gets a two-to-one advantage over competitors, and that shoppers who research with AI also use classic search 89% of the time.
- In Adobe's 2026 survey, 77% of shoppers who had used AI for online shopping said it made them more confident in their purchase, and 69% said they were less likely to return the item.
The last two points matter as much as the traffic. AI-assisted shoppers arrive further down the funnel, having already compared options.
2. Shopping is moving inside the assistant
Assistants are no longer only sending traffic. They are adding product cards, comparisons and checkout:
- ChatGPT shows shopping results with product details, and in September 2025 OpenAI introduced Instant Checkout for participating merchants, built on the open Agentic Commerce Protocol developed with Stripe.
- Google has added shopping features and agentic checkout to AI Mode, building on its Shopping Graph and merchant product feeds.
- Perplexity offers shopping answers with product cards and in-app purchasing through partners.
Availability varies by country and merchant, and the details change quickly. The direction does not: the assistant increasingly owns more of the journey.
3. What changes for fashion and beauty brands
The shortlist replaces the shelf
A search page shows dozens of options; an assistant names three to five. Being sixth is the same as being absent. Visibility becomes winner-takes-most at the level of each question.
Product data becomes the pitch
Assistants recommend what they can justify. For fashion that means stated materials, fit, measurements, care and occasions; for beauty, full ingredient lists, actives, skin-type suitability and evidence for claims. Evocative copy without facts gives the assistant nothing to quote.
Third-party consensus matters more
Assistants look for agreement between your claims and independent sources: editorial roundups, retailer reviews, communities. In beauty, retailers dominate — Similarweb found that more than half the sources cited in ChatGPT beauty conversations were retail and ecommerce domains.
Accuracy becomes a brand risk
If an assistant quotes the wrong price, a discontinued shade or an ingredient you removed, the shopper blames the brand. Keeping product pages, feeds and retailer listings consistent is now part of brand protection.
Agents need machine-readable everything
As agents start to complete purchases, they need structured product data, live stock and prices, and clear shipping and return policies they can read. Merchant feeds and schema.org markup move from “SEO nice-to-have” to infrastructure.
4. What to do now
- Measure. Track how ChatGPT, Gemini, Claude and Perplexity answer 20–40 of your shoppers' real questions, weekly.
- Open access. Make sure AI search crawlers can reach your pages and that content is served without JavaScript.
- Fix hero products first. Rewrite the pages of the products that should win your most valuable questions around facts. See the AI-ready product page checklist.
- Feed the machines. Keep Product markup, merchant feeds and retailer listings complete and consistent.
- Earn the sources. Target the roundups, retailers and communities each assistant cites for your questions.
Get your products named when shoppers ask AI
Sayrank tracks how ChatGPT, Claude, Gemini and Perplexity answer your shoppers' questions, fixes what stops AI crawlers on your site, and rewrites your product pages so assistants can quote them. Built for fashion and beauty brands.
Frequently asked questions
How big is AI shopping traffic for retailers?
Still small in absolute terms but growing fast. Adobe measured a 693% year-over-year increase in traffic from generative AI tools to US retail sites over the 2025 holiday season and forecast a further 130% rise for the 2026 season. Similarweb reported AI referrals growing 278% in fashion and 312.5% in beauty between June 2025 and May 2026.
Can shoppers buy directly inside ChatGPT?
Yes, for participating merchants. OpenAI introduced Instant Checkout in ChatGPT in September 2025, built on the open Agentic Commerce Protocol developed with Stripe. Availability depends on merchant participation and region.
What is agentic commerce?
Shopping where an AI agent does more of the work on the shopper's behalf: researching, comparing, and in some cases completing the purchase. It makes accurate, machine-readable product data and clear policies even more important.
What should a brand do first?
Measure how the major assistants answer your shoppers' questions today, make sure AI crawlers can read your site, and fix the product facts on your highest-value products. Then work on third-party sources and product feeds.