A buyer who once opened a stack of browser tabs to compare running shoes can now describe their stride, budget and weekly distance to an AI assistant and get a shortlist in one reply. The research, the comparison and much of the decision happen inside the conversation, often before the buyer visits a single store.
That shift raises a practical question for any brand that sells products: what does the assistant know about yours, and where did it learn it?
What is AI shopping?
AI shopping is the use of an AI assistant to research, compare and choose products within a conversation. The buyer explains what they need, and the assistant suggests options, explains the trade-offs and points to where the products can be bought.
It differs from a search results page in one important way. Instead of a list of links to evaluate, the buyer gets a synthesized answer, usually with a small number of named products. The assistant has done the comparing, and the buyer’s shortlist is whatever it chose to include.
What is ChatGPT Shopping?
ChatGPT Shopping refers to ChatGPT showing product results when someone asks a shopping question, typically with images, prices and links to where the item is sold. It is one of the most visible examples of AI shopping.
How ChatGPT selects and orders those products is decided by OpenAI and can change, so this post does not describe it. What can be said at a general level is that the assistant needs clear information about products from merchants and other sources to show them at all. Noma does not track ChatGPT Shopping, so the guidance below is general rather than drawn from Noma data.
The shortlist moves into the chat
In AI shopping, the comparison happens before the click. A product that is not in the assistant’s answer may never be considered, however good its product page is.
This is the same dynamic that drives answer engine optimization for brands generally, applied to individual products. Being findable is no longer enough. The product has to be describable, with facts an assistant can state confidently.
How is AI shopping different from ecommerce search?
Ecommerce search matches keywords to listings and leaves the comparison to the buyer. AI shopping takes a description of a need and returns a reasoned shortlist, so the product that wins is the one whose facts best match the need, not the one with the best keyword match.
Buyers also give assistants far more context than they would type into a search box: the activity, the constraint, what they tried before and did not like. A product can only be matched to that kind of request if the details that decide it are written down somewhere. Vague copy that works as a listing headline gives an assistant little to reason with.
AI shopping answers are hard to observe from outside
There is no ranking report for AI shopping. The answer a buyer sees depends on how they phrase the question, the details they include and the conversation around it, so two buyers asking about the same product can see different results.
The practical way to get a picture is to ask the questions yourself. Describe the needs your products serve, vary the constraints a real buyer would add, and note which products are suggested and what is said about yours. Repeat it periodically, since answers change as sources change.
What information do AI assistants rely on?
AI assistants rely on clear product data, prices and availability, reviews, and pages from both merchants and third parties. The more consistent those sources are, the easier a product is to recommend.
Product data
Names, descriptions, specifications, variants and identifiers. An assistant matching a product to a request such as “waterproof, under a certain weight, fits a wide foot” needs those facts stated somewhere it can read them.
Prices and availability
Buyers often ask for options within a budget or in stock now. Prices and availability that are current and clearly stated make a product usable for those questions.
Reviews
Reviews describe how a product performs in practice, which is what many buying questions are really about. Genuine reviews that mention specific uses give an assistant something concrete to work with.
Merchant and third-party pages
Your own product pages, retailer listings, review sites, comparison articles and community discussions all describe the same product. An assistant can draw on any of them, which means the description a buyer sees is not always the one you wrote.
Inconsistency is the quiet problem
When the same product has different specifications or prices on different sites, an assistant has to choose between them or hedge. Conflicting information makes a product harder to recommend with confidence.
This happens easily. A retailer copies an old description, a marketplace listing keeps a discontinued variant, or a review article describes the previous model. None of it is visible from your own site, and all of it can end up in an answer.
How can brands prepare for AI shopping?
Make product information accurate, structured and consistent everywhere it appears, and answer the questions buyers ask when choosing. The work is mostly good product data hygiene done thoroughly.
- Structured product data. Use Product and Offer markup on your pages, and keep any product feeds you provide to platforms and retailers accurate.
- Consistent specs and pricing. Audit the major retailers and marketplaces that list your products, and correct descriptions, variants and prices that have drifted.
- Genuine reviews. Ask real customers to review products, and never buy or fabricate reviews.
- Comparison answers. State on the product page how it differs from similar models, who it suits and who it does not.
- Plain-text facts. Keep specifications in readable text, not only in images or PDFs.
Much of this overlaps with preparing pages for search engines’ AI features. Our guide on how to win Google AI Overviews for product pages covers the page-level detail.
What to do first
In rough order of return on effort:
- Check that each key product page states what the product is, its specifications and its price in plain text.
- Add accurate Product and Offer structured data, and keep it in step with the visible page.
- Audit how your top products are described by the retailers and marketplaces that sell them, and fix the drift.
- Answer the comparison and suitability questions buyers ask, directly on the product page.
- Build a steady flow of genuine reviews.
AI shopping rewards the brands whose products are easiest to describe accurately. That is less a new discipline than an old one, product information done properly, with a new reader.