AI Shopping Is Here: How E-commerce Brands Get Found Inside ChatGPT

Shoppers increasingly ask an AI assistant what to buy, and act on the products it names, sometimes buying without leaving the chat. For e-commerce and D2C brands, being found is no longer about ranking a product page. It’s about being the product ChatGPT or Perplexity recommends when someone describes their need. The brands that win are the ones an AI can confidently identify, verify and quote.

Product discovery is moving from “browse and compare” to “ask and receive.” When a customer says “best noise-cancelling earbuds for a long flight under HK$1,500,” the assistant does the comparing and hands back a short, named list. Your job is to be on it.

What changed in how people shop?

The behaviour shift that hit search has now hit shopping. Across the board, 68% of searches end without a click as AI answers the question in place, and product research is no exception. Buyers describe a need in natural language and the assistant returns specific recommendations. ChatGPT, now with around 900 million weekly users, has rolled in-chat shopping into that flow, and major commerce platforms have integrated so their merchants are discoverable inside the assistant by default. The category page is being replaced by a conversation.

The traffic this sends is small but unusually high-intent. A shopper arriving from an AI recommendation has already been pre-qualified by a source they trust, rather than landing cold from an ad. They’re closer to buying before they ever reach you.

What gets a product named by an AI?

The same levers as the rest of GEO, applied to commerce. First, a clear brand and product identity the engine can resolve, so it knows exactly what you sell and doesn’t confuse your line with a competitor’s. Second, structured product data (materials, dimensions, use-cases, price) that an assistant can read and match precisely to “rain jacket, hiking, under HK$1,500.” Third, corroboration from reviews and credible third-party mentions, which tell the model the product is genuine and well-regarded. And finally, answer-first content written the way customers describe their problem, not the way your catalogue is organised. (For the deeper mechanics of product selection, see why AI recommends one product over another.)

Accuracy matters as much here as in any regulated field. If your specs are wrong or inconsistent, the model learns not to trust your data, and you drop out of recommendations. Get it consistently right and you become the reliable match.

This is the same system behind every GEO win; the surface is just a shopping cart instead of a citation. The foundations are in what GEO is and what AEO is.

Why most stores aren’t ready, and why that’s the opening

Most e-commerce brands still optimise product pages for Google rankings and paid ads, not for being synthesised into an AI recommendation. That gap is the opportunity. The brands that structure their data, earn genuine reviews and write for how people actually ask are already being named, while competitors wait for the trend to “settle.” It won’t settle in your favour by waiting.

Building that, with clean product entities, structured data, review authority and answer-first content held together as platforms evolve, is ongoing, specialist work. If you’d rather be the product the AI recommends, that’s the work I do.

Frequently asked questions

How do customers actually shop with AI now? They describe what they want and the assistant returns specific products and brands, increasingly with in-chat purchase. The AI does the shortlisting. If your product isn’t in that answer, it isn’t considered.

What gets a product recommended by AI? A clear, machine-readable product and brand identity, structured product data, corroboration from reviews and mentions, and answer-first content matching how people describe their need.

Is this just SEO for product pages? It builds on it, but you’re optimising to be the product the assistant names, not just to rank. The target is synthesis, not only position.