The short answer.
To be recommended in ChatGPT shopping and Google AI Mode, a product needs complete, accurate and current data in the places those systems read: a product feed (Shopify’s catalogue, Google Merchant Center, or OpenAI’s own feed), genuine product reviews, structured data that matches the visible page, and product pages that answer the questions shoppers ask. There is no paid placement inside these organic results to buy your way in.
Most of the work is product data quality, which many DTC brands have neglected because ads platforms tolerated thin feeds. AI shopping answers do not: they compare products on attributes, and a product with missing attributes cannot match a specific question. This is the product side of our AI search visibility work.
- 01 · DATAYour catalogue, feed and pages describe the product
- 02 · INGESTOpenAI, Google and Shopify’s catalogue take in the data
- 03 · QUESTIONA shopper asks for a product with conditions
- 04 · MATCHThe system runs searches and compares attributes, price, stock and reviews
- 05 · ANSWERA few products are shown, with links to buy
You control step 1 completely, and most of step 2.
How AI shopping answers work now.
Both OpenAI and Google now show product results inside AI answers, built from merchant data, and both send the shopper to the merchant to buy. The traffic is still small for most brands, but it is growing fast and converts well.
OpenAI’s merchant page says shopping “is currently live for ChatGPT users in the U.S.”, so UK brands selling into the US are affected now, and UK-only brands should prepare. Google said in May 2025 that its Shopping Graph has “more than 50 billion product listings”, of which “more than 2 billion” are refreshed every hour, and that AI Mode “will start a query fan-out, which means it runs several simultaneous searches” for shopping questions.
Shopify’s own data, published in May 2026, gives a sense of the growth. AI-referred traffic to Shopify storefronts grew more than 8x year on year as of Q1 2026, and AI-referred orders grew nearly 13x. Shopify also reports that AI-referred sessions convert at nearly 50% higher rates than organic search, with average order values 14% higher.
Shopify Enterprise, AI search insights, published 11 May 2026. Shopify describes traffic as ‘more than 8x’ and orders as ‘nearly 13x’.
Holiday data from Adobe, covering more than 1 trillion visits to US retail sites, found generative-AI traffic to US retail sites rose 693.4% in the 2025 holiday season compared with the year before, and 670% on Cyber Monday. Salesforce reported that AI and agents influenced 20% of global retail sales over the 2025 holidays, and that shoppers arriving from AI search converted nine times more often than those from social media referrals. Large percentages on a small base, but the direction is clear.
What shoppers ask.
Shoppers ask AI the way they would ask a knowledgeable shop assistant: with a use, a budget, a constraint and often a comparison. Every condition in the question is an attribute the system has to find in someone’s product data.
Search data shows the interest. In Google search data via DataForSEO for November 2026, “chatgpt shopping” draws about 1,000 searches a month in the US and 260 in the UK. Those are people, including merchants, trying to understand a channel that did not exist two years ago.
ChatGPT shopping: what OpenAI says it reads.
OpenAI documents two routes into ChatGPT shopping: a product feed sent to OpenAI, and its search crawler reading your site. If you sell through Shopify or Etsy, the feed route is already handled for you.
OpenAI’s merchant page states: “If you sell through Shopify or Etsy, your catalog is already integrated, and no additional setup or application is required.” Other merchants apply through a waitlist and send a feed. The feed specification lists nine required fields: item_id, title, description, url, brand, seller_name, image_url, availability and price. It also carries product-level review fields, star_rating on a 0–5 scale and review_count, with the instruction to “exclude seller and store reviews.” OpenAI recommends “uploading a full snapshot at least once a day”, delivered by SFTP.
Two lines in OpenAI’s documentation are worth pinning to the wall. First, being eligible for search does not mean being shown: OpenAI says so in the feed specification itself. Second, recommended attributes “like rich media, reviews, and performance signals” improve “ranking, relevance, and user trust”. The required fields get you in; the optional ones decide whether you are shown.
Do not block the crawler
OpenAI says OAI-SearchBot “is used to surface websites in search results in ChatGPT’s search features”, and that sites opted out of it “will not be shown in ChatGPT search answers”. Robots.txt changes take about 24 hours to take effect. Check your robots.txt and any bot-blocking rules at your CDN or firewall, which sometimes block AI crawlers by default.
For how ChatGPT decides what to recommend more generally, see our explainer on how ChatGPT chooses companies to recommend.
Google AI Mode and Merchant Center.
Google’s AI shopping results draw on the Shopping Graph, which is fed by Merchant Center and by structured data on your pages. Google recommends providing both, and keeping them in step with what the page says.
Google’s product structured data documentation says: “Providing both structured data on web pages and a Merchant Center feed maximizes your eligibility to experiences … and helps Google correctly understand and verify your data.” Its guidance on AI features adds two practical checks: “Making sure your structured data matches the visible text on the page” and “Checking that your Merchant Center and Business Profile information is up-to-date”. Google is also explicit that there are “no additional requirements to appear in AI Overviews or AI Mode” beyond normal search eligibility.
Free listings matter here. Merchant Center says products “can appear at no cost on Google Search, Google Maps, Gemini, YouTube” and other surfaces. A brand that only uses Merchant Center for paid Shopping ads, with free listings switched off or neglected, is giving up organic visibility in AI answers.
New attributes for conversational shopping
In January 2026 Google announced “dozens of new data attributes in Merchant Center designed for easy discovery in the conversational commerce era”. In September 2026 it reported that merchants adopting core Merchant Center feed best practices “see a 5% increase in conversions the following month” on average, and that conversational attributes submitted by one brand were incorporated 50% of the time. Read the current attribute list in Merchant Center and fill in what applies to your products.
Product data that answers questions.
The products that get recommended are the ones whose data answers the shopper’s conditions. Good product data is specific, complete, consistent across feed and page, and current on price and stock.
| Field | What good looks like | Common mistake |
|---|---|---|
| Title | Brand, product type, key attribute, variant: “Trail running shoe, wide fit, waterproof” | Marketing names only: “The Voyager” |
| Description | Use, materials, sizing, care, who it suits, in plain sentences | Brand story with no product facts |
| GTIN and brand | Correct GTIN where one exists; brand always set | Invented or reused GTINs |
| Price and availability | Matches the page and checkout, updated at least daily | Feed shows stock or a price the site does not |
| Variants | Size, colour and material as separate attributes | All options crammed into the title |
| Shipping and returns | Countries, times, costs and return window stated | Policy only in a footer PDF |
| Reviews | Product-level rating and count, from real buyers | Store reviews passed off as product reviews |
| Images | Clean main image plus use and detail shots | Text overlays and promotional badges |
On GTINs, Merchant Center’s specification is blunt: “Only provide a GTIN if you’re sure it is correct. When in doubt don’t provide this attribute.” Brand is required, and MPN is needed only where there is no GTIN.
Accuracy is a policy issue, not just a ranking issue
Merchant Center’s misrepresentation policy says “Scamming people by hiding or misrepresenting info about your business, products or services isn’t allowed”, and egregious violations can be suspended “upon detection and without prior warning”. Mismatched prices, missing return policies and unclear business contact details are common triggers. Clean data protects the account as well as the ranking.
Reviews.
Reviews do two jobs: they are a signal AI shopping systems read, and they are what persuades the shopper once they click. A product with a handful of genuine reviews is far more likely to be bought than one with none.
The Spiegel Research Center at Northwestern University found that “the purchase likelihood for a product with five reviews is 270% greater than the purchase likelihood of a product with no reviews.” The effect was larger for higher-priced products, where conversion rose 380%, than for lower-priced ones, where it rose 190%. Purchase likelihood typically peaks at ratings between 4.0 and 4.7, not at a perfect 5.0, and a verified-buyer badge improves the odds of purchase by 15%.
Spiegel Research Center, Northwestern University, How online reviews influence sales (research first published 2017; page accessed 10 November 2026).
- Ask every buyer, at the right time, after delivery and use, not at dispatch.
- Collect at product level, because OpenAI’s feed and Google’s product results use product ratings, not store ratings.
- Never write, buy or filter reviews. A slightly imperfect average is more persuasive, and fake reviews break platform policies.
- Answer negative reviews with a fix; shoppers and engines both read them.
Structured data that matches the page.
Structured data helps Google read your product pages, but only when it agrees with what the shopper sees. Mismatched markup is worse than none, because it undermines trust in the rest of your data.
Google’s merchant listing documentation notes that merchant listing experiences “require a price greater than zero”, and that where shipping or return policies are defined in more than one place, Google applies an order of precedence in which Merchant Center settings outrank on-page markup. In practice, set shipping and returns in Merchant Center, then make sure the page and markup say the same thing.
A structured data checklist for product pages
- Product markup with name, image, description, brand and GTIN where correct.
- Offer with price, currency and availability that match the visible page.
- Aggregate rating and review markup only for reviews shown on that page.
- Shipping details and return policy consistent with Merchant Center.
- One product per page, or variants grouped correctly.
- Re-tested with Google’s structured data testing tools after every theme or app change.
Our short play on checking schema against the page shows how to audit it, and our guide to page patterns AI engines quote covers the written content around it.
What Shopify does for you.
Shopify now syndicates products to AI assistants by default, so most Shopify brands are already eligible. Eligible is not the same as competitive: the data Shopify sends is only as good as the data in your admin.
Shopify’s Help Centre lists ChatGPT, Google AI Mode and Gemini, Microsoft Copilot and Meta as agentic storefront channels, and says “Agentic storefronts is active by default for eligible stores.” Its Winter ’26 Edition explained that Shopify Catalog “will syndicate and surface your products to AI chats everywhere”. In March 2026 Shopify said ChatGPT users “can now shop from Shopify merchants without ever leaving their chat”, with “no transaction fees beyond standard processing rates”. Separately, the Google & YouTube channel “automatically syncs your products and relevant information about your Shopify store with the Google Merchant Center”.
| Where | Check |
|---|---|
| Agentic storefronts | Active, with the channels you want; products you do not want sold excluded |
| Product admin | Titles, descriptions, product type, vendor, variants, barcodes (GTIN) filled in |
| Category and attributes | Product category set; the attributes shoppers filter on filled in, not left in free text |
| Google & YouTube channel | Sync healthy, no disapprovals, free listings on |
| Policies | Shipping, returns and contact pages complete and linked |
| Reviews app | Product-level reviews, with markup on the page and data in the feed |
Checkout inside the chat.
Both OpenAI and Google have tested ways to buy without leaving the AI answer, and the details keep changing. What has stayed constant is that the merchant remains the seller, so your checkout, policies and fulfilment still decide the experience.
OpenAI and Stripe launched Instant Checkout on 29 September 2025 with US Etsy sellers, using the Agentic Commerce Protocol, which Stripe describes as “an open standard co-developed with OpenAI”. OpenAI’s merchant page now says: “We’re moving away from a standalone Instant Checkout experience in ChatGPT”, with purchases completing on the merchant’s own site or app and “no fees on purchases that start in ChatGPT”. Shopify’s Help Centre says ChatGPT users complete purchases “on your online store checkout in a ChatGPT in-app browser.”
Google rolled out agentic checkout in AI Mode in November 2025 with US merchants including “select Shopify merchants”, and in January 2026 announced the Universal Commerce Protocol, “co-developed with industry leaders including Shopify, Etsy, Wayfair, Target and Walmart”, adding that “Retailers remain the seller of record.” By September 2026 Google said UCP “already enables direct checkout for hundreds of thousands of brands and retailers across Google.”
The practical point: a shopper may arrive at your checkout inside an in-app browser, already sure what they want. A slow, cluttered or login-first checkout loses that sale. If you are also weighing paid placements in ChatGPT, our ChatGPT ads readiness checklist covers that separately.
The product data scorer.
Pick one of your best-selling products and tick each statement that is true for it today. You get a score out of 100 and the gaps to fix first. Repeat for a few products; the pattern is usually the same across the catalogue. Nothing you tick is sent anywhere.
Product data score: 0 / 100
Tick what is true today for one product.
Title, description, price consistency and reviews carry the most weight because they decide whether a product can match a question at all. The rest makes the match trustworthy. We have written similar guides for dental IT support providers and testing and calibration labs.
Measuring AI shopping traffic.
Measure AI shopping as its own channel: visits and orders from ChatGPT, Gemini, Perplexity and Copilot, separated from organic search and direct. Without that split, growth from AI answers is hidden inside other numbers.
In GA4, set up a channel for AI assistants and mark purchases as key events, as our short play on tracking AI referrals in GA4 describes. Some AI visits arrive with no referrer and look like direct traffic, so add a post-purchase “how did you hear about us?” question with an AI option. Our guide to SEO for continuing education providers takes a similar approach to measuring course sales.
Shopify Enterprise, AI search insights, published 11 May 2026. Shopify describes the conversion difference as ‘nearly 50% higher’.
Report three numbers monthly: AI-referred sessions, AI-referred orders and revenue, and the share of priority shopper questions where your product is shown. Re-test the same questions each month, because AI answers change from run to run. Our guide to checking your AI search visibility sets out the method. Our GEO and AEO services start from a defined list of buyer questions.
Questions brands ask.
How do I get my products into ChatGPT shopping?
If you sell on Shopify or Etsy, OpenAI says your catalogue is already integrated. Otherwise, apply through OpenAI’s merchant page and send a product feed with the nine required fields. Then improve titles, descriptions, reviews and other optional attributes, because OpenAI says being eligible does not mean being shown.
Is ChatGPT shopping available in the UK?
OpenAI’s merchant page says shopping is currently live for ChatGPT users in the US. UK brands that ship to the US are affected now; UK-only brands should get their product data ready, because the same data feeds Google’s AI Mode and Gemini.
Can I pay to appear in ChatGPT or AI Mode shopping answers?
Not in the organic product results described here. They are built from merchant data and what the system finds. Paid formats are separate and change often, so treat them as a different decision from getting your product data right.
Do I need structured data if I already have a Merchant Center feed?
Google recommends both. Its documentation says providing structured data and a Merchant Center feed maximises eligibility and helps Google verify your data. Make sure the two agree with each other and with the visible page.
How many reviews does a product need?
There is no published threshold for AI answers. Spiegel Research Center found five reviews raised purchase likelihood by 270% compared with none. Aim for a steady flow of genuine product-level reviews rather than a target number.
Sources: OpenAI, ChatGPT merchants; OpenAI product feed spec; OpenAI commerce get-started guide; OpenAI feed upload; OpenAI commerce docs; OpenAI crawlers; Stripe, Instant Checkout; Google, AI Mode shopping; Google, agentic checkout; Google, Universal Commerce Protocol; Google, September 2026 shopping update; Google AI features; Google product structured data; Google merchant listing; Merchant Center product data spec; free listings; misrepresentation policy; Shopify Winter ’26 Edition; Shopify Help, agentic storefronts; Shopify, March 2026; Shopify Google & YouTube channel; Shopify AI search insights; Adobe holiday 2025; Salesforce holiday 2025; Spiegel Research Center; Google search data via DataForSEO. Accessed 10 November 2026.