Generative engine optimization: get your product facts right
Generative engine optimization, usually shortened to GEO, is the work of showing up well in answers written by AI: Google's AI Overviews (the AI summary shown on some Google searches) and AI Mode (its chat-style search), Gemini, ChatGPT and the rest.
It has two halves. The first is getting into those answers at all. The second is making sure that when an answer mentions you, the facts are right. This page covers both, and spends most of its time on the second.
What generative engine optimization is
Google describes it in its guide to generative AI features in Search. Of the two names you will meet, AEO (answer engine optimization) and GEO:
"These are both terms you may see used to describe work specifically focused on improving visibility in AI search experiences. From Google Search's perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO."
So GEO, AEO and "AI search optimization" are three names for one job. This page uses products as its example, because a wrong price or stock level costs a sale. The same check applies to any fact a customer acts on, such as a plan price, a service area or your opening hours.
How to get into AI answers at all
Google's guide is blunt about what matters most. Creating content that people find unique, compelling and useful "will likely influence your website's presence in generative AI search in the long run more than any of the other suggestions in this guide." Its example of commodity content is a generic tips list.
The technical floor is the one you already know from ordinary search. It is set out on Google's page about AI features. "To be eligible to be shown as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet". Indexed means Google has stored the page; a snippet is the short description under a search result.
Then check one setting. In Search Console, Google's free tool for site owners, go to Settings > Search generative AI. Google's help page on the control lists three choices: include your site, exclude it, or inherit the choice of a parent property. A property is Search Console's word for a site you have added; a parent is a broader one, such as your whole domain, that contains yours. Including is "the default control for all properties", and inheriting is "the default if your property has a parent property". So Search Console's Search generative AI control includes a site by default, unless the property inherits its parent property's choice. Confirm it says include, or inherits from a parent that does.
If you sell products, keep your product data current in Google's own systems too. "Using products like Merchant Center (such as Merchant Center feeds) and Google Business Profiles can help your products and services to be visible in both AI responses and other Google Search results." Merchant Center is where shops send Google their product list.
ChatGPT has its own way in. OpenAI's help page says: "To make a website eligible for inclusion, allow OAI-Searchbot to crawl the site and confirm that the website host or content delivery network allows traffic from OpenAI’s published searchbot IP addresses." In practice, make sure your robots.txt file, which tells crawlers what they may read, does not block OAI-SearchBot, and ask whoever runs your hosting whether its traffic gets through.
To see whether it is working in Google, open the Generative AI performance report. "Impressions are how many times links to your site were shown to a user in a generative AI feature on Google Search." It covers AI Overviews and AI Mode. It will not tell you whether the facts in those answers were right.
What you can skip for Google
Google's guide lists popular GEO tactics that Google Search does not need.
Special AI files. "You don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search". That includes LLMS.txt.
Chopping content into pieces. "There's no requirement to break your content into tiny pieces for AI to better understand it."
Rewriting for machines. "You don't need to write in a specific way just for generative AI search."
Inauthentic mentions. Seeking inauthentic "mentions" across the web "isn't as helpful as it might seem."
Structured data as a GEO requirement. "Structured data isn't required for generative AI search". The guide still recommends it for rich results in ordinary search; to see what your product pages already publish, try Check your product schema in five minutes.
Being mentioned is not the same as being right
GEO guides mostly measure mentions: how often your brand is named, how often your pages are cited, how you compare with competitors. Some also name accuracy as a goal. None of the ones we read gives a method for checking the facts.
That matters because AI answers can be wrong, and OpenAI says so. OpenAI's help page on ChatGPT search says: "Search results and citations can be incomplete, outdated, or incorrect." A wrong price or stock level in an answer is a customer deciding on bad information.
How to check what AI says about your products
Pick one product to start with, ideally a best seller.
Write down the facts, from your source of truth. Use your store admin or product database. Start with what a customer decides on: the price, whether it is in stock, the sizes or variants on offer, and policies such as shipping or returns.
Then check your own product page against that list. Google's AI answers draw on pages from its search index, and your product page is one of them. Make sure it shows the same price and stock before you judge anyone else.
Write the questions a shopper would actually type. Short and specific works best: "how much is the [product name]", "is the [product name] in stock in size 10", "does [your brand] offer free returns". Use the name a customer would use.
Ask from your customers' country. Prices and stock can differ by market, and AI tools use location. OpenAI's page says: "ChatGPT may use an approximate location based on your IP address to provide relevant local results." An answer seen from abroad may quote a different price that is right for that country. If you sell abroad, note OpenAI's caveat: "A VPN or network location may affect the approximate location inferred from your IP address."
Ask where your customers ask. In Google, an AI Overview may not appear; Google says they "are only shown when our systems determine that it is additive to classic Search, and as such, often don't trigger." If none appears, rephrase the question, and note that no answer was given. You can also try AI Mode, but it is a different answer: "AI Mode and AI Overviews may use different models and techniques, so the set of responses and links they show will vary." In ChatGPT, turn on web search. OpenAI says: "Responses that use web search may include citations. Select a citation to open its source."
Record each answer, its sources and the date, and ask twice. A screenshot is enough. Where the tool shows which link supports which statement, note that too. One answer is one sample: when we asked Google how much the Allbirds Tree Runner costs, twice on 2026-09-13, the two AI Overviews pointed to different places for the lower prices.
Mark each fact right, wrong or missing. Those Allbirds answers show the kind of thing you will meet. The standard price was right in both; we did not check the rest. Each also gave a different model's price and third-party discount prices, and the second linked a shopping-deals article from March 2026. None of that is automatically wrong. A retailer's own sale price is not a wrong fact about you, and a different model's price is fine when it is labelled as that model. The same numbers presented as this product's current price would be.
How to fix wrong information about your product in an AI answer
Open the sources shown with the wrong statement, one by one, and look for the wrong value. OpenAI's advice fits here: "Open a cited source to check that it supports the answer, review when it was published or updated, and use an authoritative source when accuracy matters."
If the wrong value is on your own site, correct it. Then look for pages you forgot: an old sale page or a discontinued product that still loads can still be found. Update it, or redirect it to the current product. When the page is right, ask Google to re-read it with the URL Inspection tool in Search Console, which lets you "Request that a URL be crawled by Google." Crawling is Google fetching the page again. Google's page on the tool says "Indexing typically takes only a day or so, but can take much longer in some cases."
If it is on a page you control but do not host, such as your listing on a marketplace or a product feed you send to retailers, fix it there as well.
If it is on someone else's page, such as a retailer, a review or a deals article, ask the owner to correct it. Make sure your own page states the right fact plainly, so there is a clear, current source to cite.
If no source shows the wrong value, or no sources are shown, the answer may have combined several pages. Fix any source that is out of date and make your own page unambiguous. If your own page is among the sources and already right, request a crawl of it anyway: the tool reports Google's "most recently indexed version of a page, not the live version on the web", and that copy can be older than what you see.
Then check again later, and expect to wait. Google's AI features page says "crawling can take anywhere from several days to several months, depending on how often our systems determine a page needs to be refreshed." We have not tested whether correcting a page changes what an AI answer says about your product, or how long that takes. None of the Google pages we cite says how long an answer takes to catch up, so a re-check is the only proof.
If you would rather not run this by hand, we make Ground Truth, and you can run a free scan. Paste a product or store URL and it reads the price and availability your store publishes. For up to a handful of prices and stock levels per scan, prices first, it asks Google Gemini a shopper's question, with Google Search switched on. That is Gemini's answer, not the AI Overview a shopper sees in Google Search. It marks each answer right or wrong. When the answer also gives your price, a retailer selling below it counts as a retailer discount, not a mistake. For a wrong answer, it marks the cited page showing the wrong value as the confirmed culprit, or failing that, the oldest dated cited page as the likely one. It checks Gemini only, and only price and availability.
What we checked, and what we did not
We ran the check twice on one product. The standard price was right both times, so the tracing step has not been tried on a real error.
The pages ranking for these searches, which we read on 2026-09-13, explain how to be visible in AI answers, and some name accuracy as a goal. None of the ones we could read gave a method for checking an answer's facts against your own data and tracing a wrong one to its source. That method is what this page is for.