
Nov 3, 2026
How to check if your products show up in ChatGPT Shopping and Google AI Mode, and fix wrong prices and details
A customer emails to say ChatGPT quoted your bestseller at last year's price and sent her to a marketplace seller you've never heard of. You open ChatGPT, ask the same question, and get a different product entirely. Welcome to shopping in a chat window, where your catalog is on display and nobody sent you the login.
Disclosure: GetMint publishes this guide and makes one of the tools mentioned. Prices and features were checked on vendor pages in September 2026.
Both big surfaces changed in 2026. ChatGPT dropped in-chat checkout and added sponsored product cards. Google added AI reporting to Merchant Center. Below is how to check whether your products appear at all, why they don't when they don't, how to fix wrong prices and details, and how ChatGPT Shopping and Google AI Mode differ in what they read and how fast they update.
TL;DR
Spot-check with prompts first: read the answers to ten buyer questions on ChatGPT and on Google AI Mode, noting product, seller, price quoted and cited source, to catch a wrong price, a wrong seller or a missing SKU. Run them on two days, since answers vary. It is a check for errors, not a measure of visibility.
Two eligibility questions get confused: can the assistant discover the product (retrieval, feed, schema), and can it sell it in the chat (Google's UCP checkout, Copilot Checkout). ChatGPT dropped in-chat checkout in March 2026, so on ChatGPT discovery is the whole game.
Wrong prices come from three places: a stale feed, schema out of sync with the page, or a third-party page the assistant trusts more than yours. The cited source tells you which one to fix.
Google AI Mode reads Merchant Center: approved-item updates usually show within a day, and automatic item updates correct price and availability from your page markup. ChatGPT Shopping reads merchant feeds (Shopify and Etsy catalogs are already connected), your pages and third-party pages; page changes can take days or longer to show, so push price changes through the feed.
For Google alone, Merchant Center's AI performance insights (US pilot) report your share of voice in AI Mode, AI Overviews and Gemini. For every engine at once, a tracker runs the prompts daily and records which product appeared, where, and on which retailer.
What is ChatGPT Shopping and why should brands care?
ChatGPT Shopping is what happens when a prompt has buying intent: the answer includes product cards or a carousel with image, price, seller and a short reason, and a click sends the shopper to the merchant to buy. OpenAI retired in-chat Instant Checkout in March 2026, so ChatGPT today is a discovery surface: the sale happens on your site or inside a retailer's own app in ChatGPT. A general answer about a product category gives text. A shopping answer gives products. The switch happens on prompts with a purchase signal: a budget, a "best", a "buy", a comparison of two named items.
Where the product data comes from: merchant feeds (Shopify and Etsy catalogs are connected by default; other merchants apply at chatgpt.com/merchants), product metadata from third-party data providers, the product page ChatGPT retrieves live (title, price, availability, schema markup), review platforms and marketplaces, and third-party pages naming the product (roundups, forum threads, press). OpenAI's help page on shopping with ChatGPT search describes these sources but publishes no trust order between them. When they disagree on a price, the assistant picks one, and it is not always yours.
Why it matters as a channel: shoppers who arrive from chatgpt.com have usually compared options already, so the click comes late in the decision. Since May 2026, GA4's default channel group files recognized assistants, ChatGPT included, under a native AI Assistant channel, and ChatGPT tags the links it cites with utm_source=chatgpt.com. Compare the channel's conversion rate with organic search on your own store before you size it.
Google AI Mode and the Gemini app ground their product answers in the same Merchant Center data, and Google's AI performance insights report covers AI Mode, AI Overviews and the Gemini app (Google). A fix to your Merchant Center feed reaches all three.
What about Claude? Claude has no shopping surface yet: no product cards, no merchant program, no checkout. It can name products in an answer, but there is no shopping result to track, so this guide leaves it out.
How do I check if my products show up in ChatGPT Shopping and Google AI Mode?
Read the answers to a handful of buyer prompts on both assistants to catch what is wrong, then confirm both assistants read your product pages. The spot check costs nothing and shows errors no feed checker sees.
Spot checks. Write ten prompts a shopper would type for your category, half with your brand name, half without. Run each on ChatGPT, once logged out and once logged in (memory and custom instructions change the carousel), and on Google AI Mode. Note per answer: products shown in order, the seller each card links to, the price quoted, the availability shown, every source pill, and whether the card carries a "sponsored" label. Repeat on a second day; answers vary, so one run can miss a problem. The goal is to find errors to fix, not to score how often you appear.
Access tests. Confirm the assistants read the pages. Fetch a product URL with a text-only client (curl, or a "view as text" browser extension) and check the title, price and availability sit in the HTML instead of being injected by JavaScript. Run the page in Google's Rich Results Test and confirm the Product markup validates with an Offer carrying price and priceCurrency. Check robots.txt for rules blocking OAI-SearchBot (the crawler behind ChatGPT search and shopping; changes take about 24 hours to apply) and Googlebot. GPTBot and Google-Extended only govern model training: blocking them does not remove you from ChatGPT Shopping or AI Mode.
Accuracy tests. For every product that appeared, compare the quoted price and details with your live page and your feed. The wrong-price section below turns each mismatch into a fix.
Free checks. The spot checks cost nothing. Google's Rich Results Test and Merchant Center diagnostics are free. For Google, Merchant Center is adding AI performance insights: share of voice against similar brands, a discovery-evaluation-purchase funnel, the product terms shoppers use and missing product attributes, across AI Mode, AI Overviews and the Gemini app. Google announced it on May 27, 2026 as a US-first pilot, with Canada, Australia, India and New Zealand to follow. If it is in your account, start there for Google. It stops at Google: it says nothing about ChatGPT, Perplexity, Copilot or Alexa for Shopping (formerly Rufus), and it shows no prompt-level answers or rival sellers.
A spot check tells you what is wrong in the answers you read. It can't tell you how often your products appear, or whether that is changing.
Why a hand count misleads. AI answers are sampled, not fixed. In a SparkToro and Gumshoe study, 600 volunteers ran 12 identical prompts nearly 3,000 times on ChatGPT, Claude and Google's AI, and the odds of getting the same list of brands twice were under 1 in 100 (SparkToro, January 2026). What holds steady is how often a brand appears across many runs, and measuring that takes volume. On 75 answers (25 prompts run three times), a 50% mention rate carries a margin of about ±11 points at 95% confidence; on 15 answers, about ±25 points. A hand count can't tell a real move from chance. Read answers by hand to understand them; count them with a tool that runs every prompt many times.
Tracking which SKUs appear, how often and in what position is a job for a tracker. GetMint's AI Shopping module runs the same prompts daily per client and country across ChatGPT Shopping, Google AI Mode and AI Overviews, Copilot, Perplexity and Alexa for Shopping, and records which product appeared, at what position and on which retailer. There is no free plan or trial: plans are sized on a call. Advanced tracks 3 AI models, Enterprise & Agency a custom set (pricing).
What tool shows if my products appear in ChatGPT Shopping?
No free tool shows your actual ChatGPT Shopping appearances across a catalog; the free options either check one engine (Google's) or check your feed without looking at answers. To spot-check real answers for errors, read them yourself; to track how often products appear, use a tracker that samples every prompt repeatedly.
Prices shown in euros and US dollars, converted at €1 = $1.14 (24 September 2026) from the currency each vendor bills in; the vendor's own currency applies.
Tool | Free or paid | Engines | What it checks | Best for |
|---|---|---|---|---|
Manual spot check (reading answers by hand) | Free | Any engine you open | Errors in the answers you read on the day: wrong price, wrong seller, missing SKU, and the source behind each | Diagnosis before you fix or buy; not tracking |
Google Merchant Center AI performance insights | Free, US pilot | Google AI Mode, AI Overviews, Gemini app | Share of voice, shopping funnel, product terms, missing attributes | Google-only reporting |
Free | ChatGPT (feed readiness only) | Whether your feed meets OpenAI's spec, not whether products appear | Feed QA before you apply | |
Paid, from €25 ($29) a month | ChatGPT, AI Overviews, Perplexity, Copilot; Gemini and AI Mode as add-ons | ChatGPT Shopping cards and ads: products, prices, retailers | Brand-level tracking that includes shopping cards | |
GetMint AI Shopping | Advanced sized on a call (12-month commitment); Enterprise & Agency on quote | ChatGPT Shopping, AI Mode, AI Overviews, Copilot, Perplexity, Alexa for Shopping | Product, position and retailer per answer, below SKU level, per country, with the raw answer behind each number | A catalog tracked across engines and countries |
One limit to know on GetMint: it monitors AI answers. It does not edit your feed or audit it. For more tools and prices by vertical, see our guide to AI shopping visibility tools.
Why aren't my products showing up in ChatGPT Shopping?
Most often the crawler is blocked, the price is invisible to a text-only reader, or the feed and the page tell different stories. The table maps each symptom to its check.
Blocker | Symptom | Check | Fix |
|---|---|---|---|
robots.txt blocks search crawlers | Products never appear, competitors do | robots.txt for OAI-SearchBot and Googlebot | Allow OAI-SearchBot and Googlebot; GPTBot and Google-Extended (training only) can stay blocked |
noindex or wrong canonical on product pages | Variants appear, the main product doesn't | Page source, canonical tag | One canonical per product, no noindex on sellable pages |
Price and availability rendered by JavaScript | Product appears with no price or an old one | Text-only fetch of the page | Server-render the price block, or put it in the markup |
Product markup missing or invalid | Wrong price, wrong currency, "unavailable" | Rich Results Test | Product markup with an Offer (price, priceCurrency, availability), plus sku, gtin, brand, image |
Feed missing GTIN or MPN | Offers not merged, wrong variant shown | Merchant Center item errors, OpenAI feed validation | Fill identifiers for every SKU and variant (required by Google where a GTIN exists, optional but recommended by OpenAI) |
Feed and page disagree | Different price in the answer and on the page | Diff feed against page | Same source of truth, same refresh |
Out-of-stock items in the feed | Assistant recommends, page says sold out | Availability field | Sync inventory to the feed daily or faster |
Landing page URL in the feed differs from canonical | Click lands on a variant or a redirect | Feed link column | Canonical URL in the feed |
Before you rewrite a single description, confirm every SKU carries a correct price, availability, identifier and image, and a landing URL matching the canonical. Our feed optimization guide covers each field in both OpenAI's and Google's specs.
How do I fix wrong prices and details in ChatGPT answers?
Find the page the assistant took the price from, then fix it at the source. The source pill under the answer names the page.
Start from your accuracy test. Each wrong price falls into one row:
The quoted price matches | Where the error comes from | Fix |
|---|---|---|
Your live page | Nothing, unless the page itself is stale | Update the page and its markup together |
Your feed, not your page | The feed or the page is out of date | One source of truth for both, pushed on the same refresh |
Neither | A third-party page: marketplace, reseller, roundup | Ask that site to correct it, and make your own feed carry the current price |
When the pill points to your page. Your page or its markup is stale. Old markup does more harm than none: a Product block with last quarter's price gets quoted as the current price, and availability "InStock" on a sold-out variant sends a shopper to a dead end. Generate the markup from the same data as the feed, validate it after every template edit, and add priceValidUntil to promotions so a sale price expires in the assistant's view when it expires on the page.
When the pill points to someone else's page. Ask that site to correct the page, and make sure your own feed carries the current price: on ChatGPT through Shopify, Etsy or a feed shared via chatgpt.com/merchants, on Google through Merchant Center. A feed gives the assistant your price instead of a guess.
On Google. A price cut on the page with a stale feed produces a mismatch warning in Merchant Center, and a disapproval if it persists. Automatic item updates often correct it first by reading your page markup.
After each fix, reread the answers to the same prompts to confirm the error is gone. On Google the change usually shows within a day. On ChatGPT it depends on which source you fixed, as the next sections explain.
What product data helps ChatGPT recommend my products?
Descriptions the assistant parses and matches. "Cordless stick vacuum, 45-minute runtime, 2.5 kg, tangle-free brush for pet hair" gives it attributes to match to a shopper's qualifier; "the ultimate cleaning experience" gives it nothing. Write your own copy instead of the manufacturer text every retailer shares.
Product markup. Google requires name, image and an Offer with price and priceCurrency. Treat the recommended properties as required for AI shopping anyway: availability, url, sku, gtin (or mpn) and brand, because they let an assistant match your page to the right product and quote the right stock status.
Reviews. Add AggregateRating and Review markup, and keep reviews current on the platforms your source pills name. Volume and recency count together: on a prompt asking what's good "right now", a product with recent reviews often beats one with a larger but older review base.
The field-by-field detail lives in our feed optimization guide.
How often do price and inventory updates appear in ChatGPT and Google AI Mode?
Google AI Mode reads the Shopping Graph, fed by Merchant Center. Updates to already approved items usually process within hours; new items go through a review of several business days. Automatic item updates, on by default, periodically cross-check price and availability against your page markup and correct the feed when they differ. Google gives no official latency for AI Mode itself.
ChatGPT Shopping is less uniform. Merchant feed data refreshes on the feed's schedule: OpenAI's integration expects at least a daily snapshot, and Shopify and Etsy sync their catalogs for you. Page data refreshes when ChatGPT re-retrieves the page, and OpenAI publishes no figure for how often it does. Third-party pages update when their owners update them, which for a two-year-old roundup is never. The practical consequence: on ChatGPT, the feed is the fastest route for a price change.
To reduce the lag, keep one feed as the source of truth with price and availability pushed at least daily, keep product URLs stable so retrieval doesn't chase redirects, generate the markup from the same data, and sync inventory so the feed never lists what the warehouse doesn't have.
ChatGPT Shopping vs Google AI Mode: what merchants should know
ChatGPT Shopping | Google AI Mode | |
|---|---|---|
Ranking inputs (as observed) | Conversational relevance to the prompt, plus reviews and third-party corroboration | Shopping Graph relevance, Merchant Center data quality, classic shopping signals |
Product data | Merchant feeds (Shopify and Etsy connected, others apply), third-party data providers, live pages | Merchant Center, required for shopping surfaces |
On-page reliance | High: retrieves product pages and third-party pages | Medium: feed first, page markup for automatic item updates |
Citations shown | Yes, source pills | Yes, retailer and review links |
Paid placement | Sponsored shopping cards from ChatGPT Ads, labeled, generated from your product feed | Shopping ads in AI Mode for Google Ads advertisers |
Checkout inside the assistant | None since March 2026: Instant Checkout was retired, purchase happens on the merchant site or in retailer apps inside ChatGPT | Agentic checkout via the Universal Commerce Protocol (UCP) for eligible US merchants, onboarding in Merchant Center |
Update speed (no official SLA) | Feed: daily snapshot or faster; page data: days or longer | Approved items: usually within a day; automatic item updates from page markup |
Microsoft Copilot reads Microsoft Merchant Center feeds, shows Microsoft Advertising ads, and added Copilot Checkout in January 2026, with Shopify merchants auto-enrolled and others applying. Our Perplexity and Copilot shopping guide covers it in detail.
Discovery eligibility versus transactional eligibility. Discovery means the assistant finds, understands and recommends the product: retrieval access, feed, markup, reviews. Transactional means the purchase completes in the chat: Google's UCP checkout in AI Mode and Gemini, or Copilot Checkout. ChatGPT tested this with Instant Checkout and retired it in March 2026 after fewer than 15 Shopify merchants went live; on ChatGPT the click goes to your site. Most brands chase the buy button and skip discovery. A product nobody discovers never reaches a buy button. Fix discovery for the whole catalog, then join the checkout programs for the SKUs with margin.
Organic or paid? Organic product results are not for sale: OpenAI says they are selected for relevance and are not ads. Paid placement now exists next to them. Since 2026, ChatGPT Ads can run as shopping ads, product cards labeled "sponsored" and generated from the same product file retailers send Google. In your spot checks, note which cards carry the label before you read a card as an organic result. What programs affect is accuracy and paid reach: a merchant feed gives the assistant your price instead of a guess, and the same feed makes the SKU eligible for sponsored shopping cards if you choose to buy them. You influence discovery with data quality, content and reviews.
For the rest of what each engine says about your brand outside shopping cards, GetMint's ChatGPT tracker and AI Overviews tracker pages describe the engine-level monitoring.
Quick start for a founder with no team
Write ten buyer prompts for your category, half with your brand name, half without.
Run them on ChatGPT (logged out and logged in, since memory changes the carousel) and on Google AI Mode; read product, seller, price and source, and flag any wrong price, wrong seller or missing SKU.
Check robots.txt allows OAI-SearchBot and Googlebot.
Validate one product page in the Rich Results Test and fix price and availability in the markup.
Connect your catalog (Shopify and Etsy are already on ChatGPT; use Merchant Center for Google) and reread the answers in two weeks to confirm the fixes show.
Step 2 catches errors; it does not tell you how often your products appear. Measuring that takes a tool that samples every prompt repeatedly, and GetMint runs the prompts daily for you.
Frequently asked questions about products in ChatGPT Shopping
Which product categories appear in ChatGPT Shopping? Consumer categories with clear attributes and prices lead: electronics, home, beauty, fashion, sports, some grocery. B2B and configurable products appear as text recommendations rather than cards. The carousel triggers on buying intent, not on category, so a niche product with a good feed appears when the prompt asks for it.
What is ChatGPT shopping research? Shopping research is OpenAI's longer shopping mode, launched in November 2025. ChatGPT first asks clarifying questions (budget, brand, use), then reads retailer pages and reviews and returns a buyer's guide with a shortlist and reasons. It reads more pages than a quick answer, so page quality and review recency matter more, not less.
Which assistant is best for which shopping use? ChatGPT for conversational shortlists and comparisons with a reason attached. Google AI Mode for price and availability accuracy and local stock. Perplexity for considered purchases where the shopper wants many sources. Alexa for Shopping (Amazon renamed Rufus in May 2026) for anything bought on Amazon, since it recommends from Amazon's catalog.
How does an agency track it for ten clients? One prompt set per client category, run daily on each assistant, with product, position, retailer, price and cited page per answer, reviewed weekly for lost placements and wrong prices. GetMint's AI Shopping module runs it per client and per country, across ChatGPT Shopping, Google AI Mode and AI Overviews, Copilot, Perplexity and Alexa for Shopping, reports your gains and losses, and every number links back to the raw AI answer. Multi-client agency workspaces sit in the Enterprise & Agency plan. A hand-run spreadsheet can't carry this: at a few runs per prompt, the gains and losses it shows are noise.
How do I know if a wrong price comes from my site or from someone else's? The source pill under the answer. If it points to your page, your page or its markup is stale. If it points to a marketplace or a roundup, ask that site to correct the page and push the current price through your own feed. The wrong-price section above walks through each case.
Check, fix at the source, check again
Your catalog is already on display in AI answers, whether or not you look. Ten prompts read by hand show you what is wrong in what shoppers see; the source pill tells you where a wrong price comes from; the feed is the fastest way to correct it. To know how often your products appear and whether that is moving, you need a tool that runs every prompt many times: book a GetMint demo of the AI Shopping module.
BUILT FOR ENTERPRISE BRANDS
Your competitors are already visible in AI. Are you?
See what ChatGPT, Gemini and Perplexity say about your brand. Set up in 5 minutes.

