SEO & GEO

What is GEO? How to get cited in AI search answers

The short answer

Generative engine optimisation (GEO) is the work of making your business easy for AI answer engines to find, understand and cite. Google AI Overviews and AI Mode, ChatGPT search, Microsoft Copilot and Perplexity mostly draw on pages they can crawl and index, so GEO rests on SEO foundations: crawlable pages, helpful original content, consistent facts about who you are, structured data and a genuine reputation. Nobody can guarantee a citation.

Key takeaways

  1. GEO is mostly SEO done properly: crawlable, indexed, genuinely helpful pages with consistent facts about who you are and what you do.
  2. Google says AI Overviews and AI Mode need no special markup, llms.txt or AI text files. Pages must be indexed and eligible to show with a snippet.
  3. Check crawler access engine by engine (Googlebot, Bingbot, OAI-SearchBot, PerplexityBot), including CDN and firewall rules, not just robots.txt.
  4. Make pages easy to quote: plain definitions, direct answers, original data or frameworks, and clear authorship backed by real expertise.
  5. Measure with a fixed monthly prompt set plus Search Console, Bing Webmaster Tools and ChatGPT referral data, and read the results as trends.

What GEO means

Generative engine optimisation (GEO) is the practice of making your business easy for AI answer engines to find, understand, trust and cite. It covers the content, structure, technical access and reputation signals that systems such as Google AI Overviews, ChatGPT search, Microsoft Copilot and Perplexity draw on when they assemble an answer and choose which sources to link.

The term comes from a 2023 research paper by Aggarwal and colleagues, which tested ways of rewriting content to improve its visibility in AI-generated responses. The researchers reported gains of up to 40% in their experiments, and found that what worked varied by subject area. That is a useful research result. It is not a recipe that transfers directly to your website.

The commercial point is simpler. A buyer who asks an AI tool "who builds custom CRMs in Victoria" or "what should a dental implant consultation include" gets one written answer with a handful of sources. If your business is not among them, you may never know the question was asked.

How AI answer engines choose what to show

Each platform publishes some guidance for site owners. None publishes a full selection formula. Here is what their own documentation says, as at October 2026.

Google AI Overviews and AI Mode

Google says both features rely on its core Search ranking systems to retrieve relevant, up-to-date pages from its index. They may use "query fan-out": the model runs several related searches across subtopics, so an answer can link a wider set of pages than a classic results page.

To be eligible, a page must be indexed and eligible to show in Search with a snippet. Google states there are no additional requirements, and that you don't need new machine-readable files, AI text files or special schema markup. Its spam policies also apply to generative AI responses.

Sites are included by default. Search Console has a Search generative AI control that lets an owner exclude a site from AI Overviews and AI Mode, and Google says that choice is not used as a ranking signal elsewhere in Search.

ChatGPT search

OpenAI says any public website can appear in ChatGPT search. For your content to be used in its summaries and snippets, don't block its search crawler, OAI-SearchBot, in robots.txt. GPTBot is a separate crawler used for model training, so blocking it is a different decision. OpenAI's publisher FAQ does not describe ranking factors.

Microsoft Copilot and Bing

Microsoft's Bing team has been the most explicit. Its October 2025 guidance says AI assistants break pages into smaller pieces, judge which pieces earn a place in the answer, and assemble a response from several sources. It recommends clear headings, direct questions with clear answers, lists, tables and schema, and warns against hiding key answers in tabs or expandable menus. It also says there is no secret formula for selection.

Perplexity

Perplexity says its crawler, PerplexityBot, is designed to surface and link websites in its search results and is not used to train foundation models. It recommends allowing PerplexityBot in robots.txt. A separate fetcher, Perplexity-User, retrieves pages when a user's question calls for them and generally ignores robots.txt, because a person requested the fetch.

What is documented and what is inference

Much of the GEO advice online is confident and unsourced. Before you spend money on it, separate what the platforms say from what practitioners infer.

ClaimStatusWhat the evidence says
Pages must be crawlable and indexed to be citedDocumentedGoogle requires indexing and snippet eligibility. OpenAI and Perplexity ask you to allow their search crawlers.
You need an llms.txt file or special AI markupContradicted for GoogleGoogle says Google Search doesn't use them.
Structured data alone earns citationsNot documentedGoogle says it isn't required for generative AI search, remains good SEO practice, and must match the visible text.
Clear headings, direct answers and tables helpDocumented by MicrosoftGoogle says you don't need to break content into tiny pieces. Both positions fit a well-structured page written for people.
Third-party mentions and reviews raise your oddsReasonable inferenceAnswers draw on many sources, so consistent independent references plausibly help. Google warns that chasing inauthentic mentions isn't as helpful as it might seem.
FAQ markup wins extra space in GoogleNo longer trueGoogle stopped showing FAQ rich results from 7 May 2026. FAQs still help readers.
A tool can show your "AI ranking"Treat with careGoogle says no third-party tool has access to its internal ranking or AI systems.

Our working rule: act first on what is documented, then on inferences that would improve the page for a human reader anyway. Skip tactics that only make sense if a machine is the audience.

How GEO relates to SEO

Google describes optimising for its generative AI features as still SEO. In practice, five foundations do most of the work for both.

  1. Crawlability. Search and AI crawlers can reach your pages, nothing important is blocked in robots.txt or at your CDN, and key content sits in the page as text.
  2. Helpful, original content. Google asks whether content provides original information, reporting, research or analysis, and warns against content a generative AI model could easily produce.
  3. Structured data. Markup that describes your organisation, articles, authors and locations, matching what the page visibly says. Google says organisation markup helps it tell your business apart from others with similar names.
  4. Entity consistency. The same business name, service names, address and people across your website, Google Business Profile, directories and social profiles, so systems can connect them.
  5. Reputation. Reviews, industry mentions and links that exist because people found you useful. For health and finance topics, Google says its systems give even more weight to content showing strong experience, expertise, authoritativeness and trust.

What changes is the unit of competition. In classic search you compete for a position on a page of links. In an AI answer you compete for a sentence or a citation inside someone else's summary.

Pages that state facts plainly, define terms and answer specific questions are easier to quote. That is inference, but it is consistent with Microsoft's guidance and with how the answers read.

A practical GEO checklist

Work through this on your ten most commercially important pages before touching the rest of the site. You can get a quick read on where you stand with our GEO visibility check.

Clarity

  • Definitions and direct answers. Open each service page with a 40 to 70 word answer to "what is it and who is it for", then expand.
  • Consistent service naming. Pick one name per service and use it everywhere: navigation, headings, schema, proposals and your Google Business Profile.
  • Real questions answered. Cost drivers, timeframes, risks and who it is not for. Buyers ask AI tools exactly these questions.
  • FAQs written for people. Keep them on the page they relate to. They no longer earn a Google rich result, so write them because they answer real objections.

Credibility

  • Author and organisation information. Bylines that link to an author page with real background. Google encourages accurate authorship information where readers would expect it.
  • Original data and frameworks. Your own method, survey results, benchmarks or worked calculations give an answer engine something no one else has. Our content & creative work starts here.
  • Third-party mentions and reviews. Earned coverage, association listings, speaking, partner pages and genuine reviews. Don't buy them.

Technical

  • Crawler access. Check robots.txt for Googlebot, Bingbot, OAI-SearchBot and PerplexityBot. Google specifically tells site owners to check CDN and hosting rules as well, because bot protection there can block crawlers you meant to allow.
  • Schema that matches the page. Organisation, article, author profile and local business markup, consistent with the visible text.
  • Answers in the page, not hidden. Microsoft warns that AI systems may not render content inside tabs or expandable menus. If your current site makes this hard, it is a website problem, not a content one.

Local and fresh

  • Local signals. A complete Google Business Profile for each location. Google says Business Profiles can help your services appear in AI responses. Location pages need substance, not a swapped suburb name.
  • Freshness. Show when a page was last reviewed, and update facts, prices and team details when they change. Microsoft says regular updates help AI systems reference the current version of a page.

How to measure AI visibility with a fixed prompt set

AI answers change with the user, the location, the day and the model, so there is no stable "rank" to report. The practical answer is a fixed prompt set, run on a schedule, alongside the platform data that now exists.

Build the prompt set

  1. Write 20 to 40 prompts your buyers would actually type. Mix definitions ("what is a client portal"), problems ("how do we cut quoting time"), comparisons ("custom CRM or HubSpot"), local intent ("dental marketing agency Geelong") and brand checks ("what does [your business] do").
  2. Fix the conditions: same engines, logged out where possible, same location and the same wording each run.
  3. Run it monthly. For each prompt and engine, record whether you are mentioned, whether you are linked, which URL, whether the description is accurate, and which competitors appear.

Read it as a trend

An illustration, using assumptions rather than real data:

ItemValue
Prompts in the set (assumption)30
Engines checked (assumption)4
Checks per monthly run30 × 4 = 120
Answers citing you (assumption)18
Citation rate18 ÷ 120 = 15%
Citations that describe you inaccurately (assumption)4
Error rate among citations4 ÷ 18 = 22%

The error rate is often the more useful number. An answer engine repeating an old service, a wrong suburb or a discontinued price points you straight to the page or profile that needs fixing.

Add the platform data

  • Google Search Console. The generative AI performance report shows impressions from AI Overviews and AI Mode by page, device and country. Google says it was available to all websites by 31 August 2026, though it only appears once a site has enough impressions. It reports impressions, not clicks.
  • Bing Webmaster Tools. AI Performance, in public preview since February 2026, shows citations across Microsoft Copilot and AI-generated summaries in Bing, the grounding queries behind them, and page-level citation counts. Microsoft notes the data is a sample.
  • Analytics. OpenAI says ChatGPT adds utm_source=chatgpt.com to referral links, so you can see those visits and what they do next.

Then connect it to revenue. Visibility that never turns into enquiries is a vanity metric, which is why we report booked revenue, not leads.

Where to start, and what to avoid

For most established businesses, the first 90 days look like this:

  1. Confirm crawler access and indexing for your key pages.
  2. Run a baseline prompt set so you can see change later.
  3. Rewrite your top service pages with a plain definition, direct answers and real detail on cost drivers, process and fit.
  4. Tidy your entity: one business name, consistent service names, author pages, organisation markup and complete Business Profiles.
  5. Publish one piece of original material, such as a framework, a benchmark or a worked calculation that others will want to reference.

Avoid mass-producing thin pages for every possible question. Google says a high quantity of pages doesn't make a site higher quality, and its spam policies apply to AI answers too. Avoid paid "mention" schemes for the same reason.

Be realistic about outcomes. No one controls which sources an AI answer cites, and no one can guarantee you will appear. What you can control is whether your business is clear, accessible, consistent and worth citing.

In regulated sectors, accuracy matters more than reach. Our dental group worked example shows how search, AI visibility and enquiry handling fit together, and our guide to dental advertising rules covers what health businesses can say. If you want help with the work above, see our SEO & GEO service.

Altitude Creative is a Geelong-based AI, business systems and growth marketing company working with established organisations across Australia. Our insights are written to be useful on their own: every third-party fact links to the source listed below. About Altitude Creative

Questions, answered.

Is GEO different from SEO?

Mostly not. Google says optimising for AI Overviews and AI Mode is still SEO, built on the same crawling, indexing and ranking systems. The differences are in emphasis: answers quote passages rather than list pages, independent mentions matter alongside links, and you measure with prompt sets and new platform reports rather than rank tracking.

Do we need an llms.txt file?

Not for Google, which says Google Search doesn't use llms.txt or other AI text files and needs no special markup. The OpenAI and Perplexity publisher guidance we reviewed asks for crawler access in robots.txt instead. Spend the effort on clear pages and open access.

Should we block AI crawlers?

Decide crawler by crawler. Blocking OAI-SearchBot or PerplexityBot keeps your content out of those search answers, while GPTBot relates to OpenAI model training and Google-Extended controls use in other Google AI systems, not Search. Google's Search generative AI control in Search Console excludes a site from AI Overviews and AI Mode without affecting ranking elsewhere.

Will GEO get us cited by ChatGPT or Google?

Not reliably, and anyone promising placements is overselling. The platforms don't publish selection formulas, and answers vary between users and over time. GEO improves the odds that you are found, understood and described accurately, which is why we report trends across a fixed prompt set.

Want this applied to your business?

Book a 30-minute call. You'll leave with a clearer view of where to start.

Sources

Third-party facts on this page, checked 14 October 2026.

  1. AI features and your website, Google Search Central. developers.google.com/search/docs/appearance/ai-features
  2. Google's guide to optimizing for generative AI features on Google Search, Google Search Central. developers.google.com/search/docs/fundamentals/ai-optimization-guide
  3. Search generative AI control, Google Search Console Help. support.google.com/webmasters/answer/16908024?hl=en
  4. Generative AI performance report (Search), Google Search Console Help. support.google.com/webmasters/answer/16984139?hl=en
  5. Latest Google Search documentation updates (FAQ rich result deprecation; spam policies and generative AI responses), Google Search Central. developers.google.com/search/updates
  6. Creating helpful, reliable, people-first content, Google Search Central. developers.google.com/search/docs/fundamentals/creating-helpful-content
  7. Organization structured data, Google Search Central. developers.google.com/search/docs/appearance/structured-data/organization
  8. Article structured data (author markup), Google Search Central. developers.google.com/search/docs/appearance/structured-data/article
  9. Publishers and developers FAQ, OpenAI. help.openai.com/en/articles/12627856-publishers-and-developers-faq
  10. Optimizing your content for inclusion in AI search answers, Microsoft Advertising (Microsoft Bing). about.ads.microsoft.com/en/blog/post/october-2025/optimizing-your-content-for-inclusion-in-ai-search-answers
  11. Introducing AI Performance in Bing Webmaster Tools public preview, Microsoft Bing Webmaster Blog. blogs.bing.com/webmaster/February-2026/Introducing-AI-Performance-in-Bing-Webmaster-Tools-Public-Preview
  12. Perplexity crawlers, Perplexity. docs.perplexity.ai/docs/resources/perplexity-crawlers.md
  13. GEO: Generative Engine Optimization (Aggarwal et al.), arXiv. arxiv.org/abs/2311.09735v3