How to Appear in ChatGPT Shopping Results
Product discovery is moving into the chat window. Buyers now type “best noise-cancelling headphones under $200” into ChatGPT and get a shortlist of specific products — with prices, pros and cons, and links to buy. If your products aren’t in that shortlist, you don’t get the click, the cart, or the customer. Here’s how ChatGPT shopping results actually get assembled, and what to change on your store this month.
How ChatGPT builds a shopping shortlist
A shopping answer is assembled from three inputs, not one:
- Structured product data — machine-readable price, availability, brand, ratings and identifiers (GTIN/MPN) pulled from your product pages and merchant feeds.
- Live retrieval — a search pass across the open web at answer time, which favours pages that load fast, are crawlable, and answer the buyer’s exact qualifier (“under $200”, “for wide feet”, “for sensitive skin”).
- Third-party consensus — review sites, roundups, forums and comparison articles that name your product alongside competitors. This is the layer most stores never work on, and it’s often the deciding one.
The model isn’t trying to rank pages. It is trying to justify a recommendation. It names products it can describe confidently and attribute to a source. Anything ambiguous gets dropped rather than risked.
What quietly gets you excluded
| Problem | Why it kills the mention |
|---|---|
| Price and stock only rendered by JavaScript | The retrieval pass sees an empty product page and can’t verify the “under $200” claim |
No Product + Offer schema | No confirmed price, currency, availability or rating to quote |
| Missing GTIN/MPN | Your item can’t be matched to the same product elsewhere, so external reviews don’t reinforce it |
| Vague copy (“premium comfort, iconic design”) | Nothing matches the buyer’s qualifier, so a competitor with specs wins |
| Zero third-party coverage | No consensus signal; the model has only your word for it |
Blocked AI crawlers in robots.txt | You opted out of the shortlist entirely |
The fix list, in priority order
- Server-render the commercial facts. Price, currency, availability, shipping and returns must be present in the raw HTML — not injected after load.
- Ship complete
Productschema.name,brand,gtin13/mpn,image,offers(price, priceCurrency, availability) andaggregateRatingwhere genuine. See schema markup for AI for the exact patterns. - Write to the qualifier. Buyers shop by constraint: budget, size, skin type, use case, compatibility. Put those constraints in plain sentences on the page (“fits wrists 140–210 mm”, “under $200”, “dishwasher safe”). Models quote specifics; they skip adjectives.
- Add a real comparison block. A short honest “how this compares” table (including who it’s not for) is the single most quotable asset on a product page.
- Answer buying questions on-page. Sizing, materials, warranty, delivery time, returns window — as Q&A with
FAQPagemarkup. - Keep your feed clean. Feed errors — mismatched price, out-of-date stock, missing identifiers — propagate straight into shopping surfaces.
- Earn third-party mentions. Get into roundups and comparison posts in your category. One credible “best X for Y” list naming your product moves shopping answers more than a month of on-page tweaks.
- Allow the crawlers. Confirm GPTBot, OAI-SearchBot, PerplexityBot and Google-Extended are permitted, and publish an llms.txt pointing at your category and product hubs.
How to test whether it worked
Don’t ask “is my brand in ChatGPT?” Ask the ten questions your buyers actually ask, the way they ask them — with the budget, the use case and the location baked in. Run the same ten every week across ChatGPT, Perplexity, Gemini and AI Overviews, and log three things: whether you were named, which competitors were named, and which sources the answer cited. The cited sources are your roadmap — those are the pages you need to appear on next.
Movement here is slower than a product-page A/B test and faster than classic SEO. Most stores that fix structured data and add one honest comparison asset per category see mentions appear within four to eight weeks.
Where to start
Pick your five highest-margin products, not your whole catalogue. Fix the structured data, write to the qualifiers, add the comparison block, then go earn one third-party mention per product. Work the same loop described in how to do AEO, and use the AEO checklist to confirm nothing basic is missing. If the whole concept is new, start with what is AEO.
And before you change anything, get a baseline. A free AI Visibility Scan shows you exactly which products and competitors the answer engines name today — so you can prove the difference eight weeks from now.