How to Show Up in AI Search: A Practical Guide to Generative Engine Optimization

What Is Generative Engine Optimization (GEO)?

Generative engine optimization (GEO) is the practice of making your content easy for AI search tools to find, understand and cite, so your brand shows up in the answers they write. When someone asks ChatGPT, Perplexity, Gemini or Google's AI Overviews a question in your field, GEO is the work that gives your site a chance to be one of the sources behind the answer.

You will also see the names answer engine optimization (AEO), AI SEO and LLM SEO. They overlap so much that you do not need to memorize the differences. All of them aim at the same thing: being present inside AI-generated answers.

The difference from classic SEO is the target. Traditional SEO tries to get a page to a high position on a results page so a person clicks through. GEO tries to get your content used inside the answer itself, so your brand is a source the assistant names or links. A page can rank first on Google and still never be mentioned by ChatGPT, because different systems decide each one.

People already ask these tools questions they used to type into a search box, and they get one written answer instead of ten links. If your brand is missing from that answer, a reader who stops there never meets you. An AI mention puts your name in front of the reader. An AI citation also links to your page and can send a visitor. Both are worth earning, and both come from the work in this guide.

This guide explains how AI search picks its sources, how to check whether you appear today, and the practical steps in order, with free tools for the parts you can act on right now. One honest limit up front: nobody outside the AI companies can see exactly how these systems choose. Where a company has published guidance, we quote it. Where it has not, we say so and treat the advice as a working assumption you can test.

How AI Search Chooses Its Sources

In simplified terms, most AI search tools work in two steps. First they retrieve candidate pages that look relevant to the question. Then a language model writes an answer from what it retrieved and cites some of those pages. Retrieval is where your page either makes the shortlist or does not.

Google describes the first step for its own AI features. It says AI Overviews and AI Mode may use a "query fan-out" technique, which it describes as "issuing multiple related searches across subtopics and data sources." In practice, one long question becomes several smaller searches. A page that answers one of those smaller questions clearly can be picked even if it does not rank for the full question.

What makes a page a likely source? For its own features, Google's answer is the same foundation as the rest of Search: the page is indexed, it can be shown with a snippet, it follows Search policies, and its content is helpful, reliable and written for people. The other AI companies publish less about how they choose, so treat the following as working assumptions rather than rules:

• A visible person or organization stands behind the page, so the source looks credible.
• The page answers the question as it was asked instead of circling it.
• The content covers the real intent behind a query, including the follow-up questions.
• Pages on topics that change are kept current, because a stale answer is a weak thing to cite.

Independent mentions matter too. An assistant that answers from the open web builds its reply from whatever pages it retrieves, so what other sites say about you can shape the answer, not only what you say about yourself. Google says its generative AI features can show what is being said about products and services across the web, including in blogs, videos and forum discussions. If several independent pages describe a brand as a strong option in its category, the retrieved set can contain that description. We cannot see inside any assistant, so test this assumption with the audit below. It is the reason the authority section of this guide exists.

Not every question triggers an AI answer. Google says AI Overviews "are only shown when our systems determine that it is additive to classic Search, and as such, often don't trigger." So do not assume your target question produces one. Run it and look.

Finally, answers vary. Ask the same question twice and you may see different brands in a different order. Think in frequencies, not fixed positions: the goal is to appear more often across repeated runs, not to win one screenshot on one day.

How to Check If You Show Up in AI Search (Free)

Before you change anything, find out where you stand. Monitoring dashboards exist and have their place, but a manual audit costs nothing and is enough to start.

1. Open ChatGPT, Perplexity, Gemini and Google (an AI Overview or AI Mode answer).
2. Write 10 to 15 prompts your audience would type before they know your brand. Examples: "best [your category] tools", "what is the best way to [the problem you solve]", "which [product type] should a small business choose".
3. For each answer, note four things: whether your brand appears, in what position if it is listed, which competitors appear, and which sources the answer cites.
4. Run each prompt three to five times, because answers vary. One result tells you almost nothing. How often you appear across several runs is the real signal.
5. Ask each tool "what do you know about [your brand]". This shows what the assistant currently believes about you and whether it is accurate. A wrong, outdated or missing description is itself a GEO task.

Perplexity deserves special attention because it links the sources it used in its answers. That transparency is a gift. You can see which pages feed the answers in your category, and if a competitor's article or a forum thread is the source, that is your map of where to earn a mention.

For Google there is also a free measurement route. Search Console has a Generative AI performance report that shows impressions from AI Overviews and AI Mode over time, by page, country and device. Google says it rolled the report out to all websites on August 31, 2026, and you may not see it until your site has enough impressions in those features. It tells you about Google only. It does not tell you whether ChatGPT or Perplexity mention you, so the manual audit still matters.

Our own report covers the part you control. The AI Search Readiness section of the free SEO audit reads your robots.txt for six AI crawlers (GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended and CCBot), checks whether an llms.txt file exists, and looks at structured data, the H1, meta description, canonical tag, indexability and content depth. It skips Claude-SearchBot and the user-initiated fetchers in the table below, so read those rules in your file yourself. It also does not measure whether any assistant mentions you, because that cannot be measured without asking the assistants, which is what the manual audit does.

Make Sure AI Crawlers Can Reach Your Content

This is the technical foundation and the quickest thing on the list to fix. If a crawler cannot read your page, nothing else in this guide matters. Google lists it first among the SEO fundamentals for its AI features: "Ensuring that crawling is allowed in robots.txt, and by any CDN or hosting infrastructure."

Search crawlers and training crawlers are different

The main AI companies run more than one crawler, with different jobs and separate robots.txt settings. OpenAI says so directly: it uses OAI-SearchBot and GPTBot tags "to enable webmasters to manage how their sites and content work with AI," and a site can allow OAI-SearchBot to appear in search results while disallowing GPTBot so its content is not used for training. This table summarizes what each company documents.

The practical reading is simple. For being cited in answers, the search and user crawlers are the ones that matter. Blocking a training crawler is a separate business decision, and the companies document it as a separate setting. Some sites block training crawlers on purpose. Just do not block everything by accident, because a blanket rule written for one purpose can remove you from the other. A file like this blocks two training crawlers and leaves the search crawlers alone, as long as your * group does not block them:

User-agent: GPTBot
Disallow: /

User-agent: ClaudeBot
Disallow: /

You can build a clean, valid file with our robots.txt generator, and our guide to robots.txt files explains how groups and precedence work.

Google works differently

Google's AI features run inside Search, so the control is Googlebot, not a separate AI token. In Google's words, "robots.txt directives for Googlebot is the control for site owners to manage access to how their sites are crawled for Search." To limit what Search shows from a page, you use nosnippet, data-nosnippet, max-snippet or noindex. Do not treat Google-Extended as the AI Overviews switch. Google says it governs Gemini training and grounding in some other systems and "does not impact a site's inclusion in Google Search."

Search Console also has a Search generative AI control under Settings. The default includes your site. If someone switched it to exclude, your links and content will not appear in AI Overviews, AI Mode or the generative AI features in Discover, and you will get no impressions or traffic from them. Google says this control does not affect AI training, which Google-Extended handles. Check it once, especially on a property someone else set up.

A stray noindex or nosnippet is an easy way to disappear from Google's AI answers without noticing. Our meta tag analyzer shows the robots meta tag a page carries. It does not read an X-Robots-Tag header, a googlebot-specific meta tag or data-nosnippet attributes, so check those too, for example with the URL Inspection tool in Search Console.

Your CDN and the content itself

Robots.txt is not the only gatekeeper. A CDN or firewall can refuse AI crawler requests even when robots.txt allows them. Cloudflare, for example, has AI bot settings that sort crawlers by purpose (search, agent or training) and can block each group. Its documentation also describes updated defaults for new domains, from September 15, 2026, that block training and agent bots on pages that display ads. If you use a CDN, check its bot rules as well as your file, because a default you never looked at, or a setting you enabled once and forgot, can undo everything else.

Then look at the content itself. Google lists "making sure that important content is available in textual form." Text behind a login cannot be read at all. Googlebot can process JavaScript, and Google says it can "as long as it isn't blocked," but we found nothing in the crawler pages from OpenAI, Anthropic or Perplexity about whether their crawlers run scripts, so keep key text in the HTML. A quick test: open the page source, or fetch the URL with curl, and search for a sentence from the page. If it is missing there, Google can usually still render it, but other AI crawlers may never see it.

You may also read advice to add an llms.txt file. Google's guide to optimizing for generative AI says you "don't need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search," and that creating llms.txt files will "neither harm nor help" your visibility there, "as Google Search ignores them." We found nothing in the crawler documentation from OpenAI, Anthropic or Perplexity that says the file affects which sites they cite. Create one only if a tool you use reads it.

Structure Your Content So AI Can Extract It

Once crawlers can reach your pages, write them so the answer is easy to find, for readers and for any system that quotes you.

The most useful habit is to put the main point first. Lead each section with a direct answer to the question it addresses, then explain underneath. If your conclusion is buried at the end of a long build-up, it is harder to lift and easier to skip.

Structure the rest for extraction as well. Use descriptive headings, keep paragraphs short, use bullet points where they fit, and format real question-and-answer content as questions and answers. Dense blocks of prose are also harder for readers to scan.

Think about fan-out too. Because AI features may split a long question into several smaller searches, cover the sub-questions around your topic and use the natural words a person would search with. Write the way people phrase real questions. Repeating a keyword adds nothing a reader or a model can use.

Keep to one useful page per topic. Google warns that creating separate pages for every possible variation of a query, or for fan-out queries, mainly to manipulate rankings or generative AI responses violates its scaled content abuse spam policy. It also says there is no requirement to break content into tiny pieces or to write in a special way for generative AI search, so treat these habits as clear writing for people first.

Two more habits carry weight. Keep pages on changing topics current and date them honestly. And add specifics: real numbers, first-hand examples and observations from your own work. That is the experience part of E-E-A-T, the signal that you actually know the subject. Vague text that could have been written by anyone gives an assistant no reason to pick your page over another.

Structured Data: What Google Says It Does and Does Not Do

Schema markup comes up in almost every GEO guide, and many overstate it, so here is the version Google's own documentation supports.

Valid structured data helps search engines understand what a page is. Google's AI features guide lists "making sure your structured data matches the visible text on the page" among the fundamentals, so the habit worth building is accuracy: the markup says what the page says, no more.

Now the caveat most guides skip. The same guide says: "There's also no special schema.org structured data that you need to add." No tag unlocks inclusion in AI Overviews or AI Mode. The content and the authority behind it do the work.

The practical takeaway: add the types that honestly describe each page, such as Organization on your home page, Article on posts and Product where you sell one, keep them identical to the visible content, and validate them with our schema markup validator. Structured data supports your visibility. It does not manufacture it.

Build Authority Beyond Your Own Site

This is the half of GEO that beginners miss. An assistant does not learn about you only from your own website. It also sees what the rest of the web says about you, so a large part of the work happens off your own pages.

Start with mentions on sources people and assistants already read: reputable publications in your field, respected directories, and comparison articles in your niche. When someone asks an assistant for the best tools in a category, the sources it lists can include best-of and comparison pages. Search your category in Perplexity and look at which pages it cites.

Be present, genuinely, in the communities your audience uses. In some niches the sources Perplexity lists include forum and community threads. If yours do, a helpful, honest answer in those threads can be worth more than another page on your own site. Spam is easy to spot and does more harm than good. For products, review platforms are another place an assistant can learn about you.

None of this is a separate universe from classic SEO. The pages assistants retrieve are ordinary web pages, so a strong comparison article, a good piece of digital PR or a useful answer in a community can pay off in search rankings and in AI answers at once. Authority like this is earned over time through usefulness. Google also says that seeking inauthentic "mentions" is not as helpful as it might seem, so earn them rather than manufacture them.

GEO and Traditional SEO Work Together

If all this sounds like a reason to abandon SEO, it is not. GEO is a layer on top of good SEO. Google says so for its own features: "The best practices for SEO remain relevant for AI features in Google Search." It goes further and says that optimizing for generative AI search is optimizing for the search experience, "and thus still SEO."

The foundation is shared. Crawlability, clean page structure, helpful content and topical authority serve both your rankings and your AI citations. Where they differ is the goal and the measurement. SEO targets a position on a results page. GEO targets a mention or citation inside an answer. Track both, because a page can rank well and still never be cited, and the only way to know is to check each channel separately.

For a small team the message is reassuring. You do not need to throw out your SEO playbook and run a separate AI operation. Get the fundamentals right first, and a quick audit of your site is the fastest way to see where you stand, then add the GEO steps in this guide on top. If you are new to reading an audit, our guide on how to read your SEO audit report explains what the findings mean, and our technical SEO checklist covers the groundwork that feeds both channels.

Your GEO Starting Checklist

If you take nothing else from this guide, start here. These steps are in order, and most are free.

• Check your visibility. Run the manual audit across ChatGPT, Perplexity, Gemini and Google's AI answers, and record where you appear. Use the Generative AI performance report in Search Console for your Google impressions.
• Open the doors. Confirm the search and user crawlers you want are allowed in robots.txt and at your CDN, that important pages are indexable and eligible for a snippet, that the Search generative AI setting in Search Console includes your site, and that key text is in the page's HTML, not behind a login.
• Structure for extraction. Lead sections with direct answers, use clear headings and short paragraphs, and include real specifics.
• Keep structured data accurate. Add the types that describe the page, match the visible text, and expect support, not magic.
• Build authority. Earn genuine mentions in trusted publications, communities and comparison content.
• Keep the fundamentals strong. Google says SEO best practices remain relevant for its AI features, so keep the basics strong.

AI search is new and moving fast. Nobody has it fully figured out, and anyone promising a guaranteed formula is overselling. The honest approach is to start with the free steps above, measure where you appear and repeat as things change. The simplest first move is to run a free SEO audit to get your foundation right, then build your AI visibility from there. Google's AI features and your website guide and OpenAI's crawler documentation are primary sources behind the claims above, alongside the Anthropic and Perplexity crawler pages, and they are worth rereading as the tools change.