What is AI SEO?

Generative engine optimisation, AEO and LLM SEO: one job, several names

Generative engine optimisation is the name most often used for the work this guide calls AI SEO: making a business easy for AI answer tools to find, understand and name. The labels multiplied because researchers, agencies and software vendors started describing the same shift at about the same time, and each picked a term that suited them. Google's own guide, Optimizing your website for generative AI features on Google Search, spells out that AEO stands for answer engine optimisation and GEO for generative engine optimisation. It then adds that, from Google Search's point of view, optimising for generative AI search is optimising for the search experience, and so still SEO.

That's a useful anchor. The labels describe where the result shows up, not a different craft. When a proposal uses one of them, ask which AI products it covers, what will actually change on your site and listings, and how the result will be checked. The table decodes the common names. If you're comparing services rather than terms, the generative engine optimisation, answer engine optimisation and LLM SEO pages describe the work done under each name.

One clash needs its own warning. GEO in capitals means generative engine optimisation. Geo SEO, as this site uses it in the Geo SEO section, means geographic SEO: showing up in Google Maps and local results across the suburbs you serve. The two support each other but they are different jobs, and our local SEO vs Geo SEO guide separates them in detail. If a quote says GEO and you can't tell which meaning it uses, ask before you sign.

Where the term came from, and what the research actually tested

The phrase generative engine optimisation comes from a research paper titled GEO: Generative Engine Optimization, by Pranjal Aggarwal and colleagues, first posted to arXiv in November 2023 and accepted to the KDD 2024 conference. The authors described generative engines as search tools that gather information from several sources and summarise it with a large language model. They framed GEO as a way for content creators to improve their visibility in those responses, because creators otherwise have little control over when and how their content is shown.

The paper built a test set called GEO-bench and reported that its methods could lift visibility by up to 40 per cent in generative engine responses. Read that figure carefully. It came from a research benchmark, measured with the authors' own visibility metrics, and the abstract itself notes that results varied across subject areas. It isn't a measurement of ChatGPT, Gemini or Google AI Overviews, and it says nothing about how an electrician in Nambour or a physio in Noosa will fare. Anyone quoting it as an expected result for your business is misusing it.

So the term has a real origin, and the underlying idea, that content can be written so AI tools are more likely to use it, holds up. What doesn't hold up is the claim that a set of tricks will reliably place a business in AI answers. Google's guide is direct about its own products: the usual SEO practices remain relevant, because its generative AI features are rooted in its core Search ranking and quality systems. That's why this guide treats GEO as an extension of what SEO already covers, not a replacement for it.

How AI answers choose their sources: search first, then write

AI answer tools don't rank pages the way Google's classic results do, but most of them still start with a search. Google's guide describes retrieval-augmented generation, also called grounding, as a technique used to improve the quality, accuracy and freshness of AI responses. It describes query fan-out as a set of related searches the model runs at the same time to gather more information. In plain terms, the tool breaks a question into smaller searches, reads what comes back, and writes one answer from the passages it trusts.

Three things follow for a local business. A page has to be reachable and indexed before it can be retrieved: Google says a page must be indexed and eligible to show in Search with a snippet to appear as a supporting link in AI Overviews or AI Mode. The fan-out means your page competes on the sub-questions too, such as cost drivers, licences or call-out areas, not only the headline query. You don't need to reshape pages for this. Google's guide says there's no requirement to break content into small pieces for AI and no need to write in a specific way, because its systems can find the relevant part of a page. Clear, specific answers to common questions are still general good practice, for customers first.

The sources these tools weigh still include your website and whether it answers questions clearly, your Google Business Profile and reviews, mentions of your business on other sites and directories, and how consistent your name, address and phone details are everywhere. Google's guide says products such as Google Business Profile can help your services be visible in both AI responses and other Google Search results. Confidence is the key word: gaps or mismatched details give a tool less reason to name you. Our guide to how AI search works walks through each step in more depth. The table covers the products a local business is most likely to meet, using only what each owner documents.

Crawler access: the cheap check that sites skip

Before any wording matters, each tool's crawler has to be allowed in. Several AI companies run more than one crawler, and they do different jobs, so a single broad rule in robots.txt can switch off search visibility when the owner only meant to stop model training. Website builders, firewalls and security plugins can also turn away unfamiliar bots without anyone noticing. Nothing on the site looks broken, so the problem can sit there for months.

The decision rule is simple. If you want to appear in an AI product's answers, allow the crawler that product uses for search, whatever you decide about training. The table sets out what each company says its crawler does, and the numbered checks run in order.

One more check belongs here. Make sure the main text of each service page is in the page itself, not loaded only after a script runs, and not hidden behind a pop-up or a login. Open the page source, or run URL Inspection in Search Console and view the crawled page. If the paragraph you want quoted isn't there, a crawler may not see it either.

1. Open yourdomain.com.au/robots.txt and look for any rule naming OAI-SearchBot or PerplexityBot, or a catch-all rule that disallows every user agent from the pages you want cited.
2. Ask whoever manages your hosting, firewall or security plugin whether automatic bot blocking is switched on, and which AI search crawlers it lets through.
3. After a change, allow time before you test: OpenAI says it can take about 24 hours for its systems to adjust to a robots.txt update.
4. Check that the pages you want cited aren't set to noindex or nosnippet. Google lists both among the controls that limit how content appears in its AI features.
5. Run URL Inspection in Search Console on your home page and main service pages to confirm Google has indexed them.

What the work involves for a local business

Most of the work is making your business easy for machines to read and easy to trust, and very little of it is new. Google's guide says you don't need to write in a specific way just for generative AI search, and that creating content people find unique and useful will likely do more for your presence in its AI features than any other suggestion it makes. The steps run in the order that makes sense for a small service business, starting with the checks that cost least.

Step four is where most sites fall short, so here's what it looks like. Google doesn't require this style; the point is that vague copy tells nobody what you do. Say a plumbing business in Maroochydore has a service paragraph that reads: 'We provide quality plumbing with friendly, professional service.' A customer asking 'who can fix a burst pipe in Maroochydore tonight' learns nothing from that sentence, because it names no job, no area and no hours. A rewrite might read: 'We repair burst and leaking pipes across Maroochydore, Buderim and Mooloolaba, including after-hours call-outs. The plumber isolates the water, finds the failed section and replaces it, and explains what drives the cost before starting.' Every claim in the rewrite has to be true for that business, and each gives a customer, or a tool, something specific to repeat.

Schema markup belongs in the work, with the right expectation. Google says structured data isn't required for generative AI search and there's no special schema.org markup to add. LocalBusiness markup is still worth keeping accurate as part of ordinary SEO, as long as it matches what visitors can see on the page. For the full list of checks in one place, use our AI visibility checklist.

1. Confirm access: run the crawler checks above and remove any block on the search crawlers.
2. Make your Google Business Profile the master record. Correct the name, address or service area, phone, hours, categories and services, then match every other listing to it; our guide to business listing accuracy covers the directories.
3. Publish one facts page, usually About or Contact, that states the same details in plain text: who you are, what you do, where you work, licence numbers where your trade requires them, and how to book.
4. Rewrite each main service page answer-first, as a clarity choice for customers rather than a Google requirement: open with the direct answer to the question a customer would ask, name the suburbs you actually serve, and give each common question its own heading.
5. Earn genuine mentions through real reviews, trade association and supplier listings, and local media coverage. Google's guide warns that seeking inauthentic mentions across the web isn't as helpful as it might seem.
6. Keep it current: when hours, services or areas change, update the profile, the site and the listings together, and show a revised date on pages you update.

Worked example: a Nambour electrician runs a prompt test

This business is hypothetical, and every number in the example is illustrative, not a benchmark. Say an electrical contractor in Nambour wants to know whether AI tools name it. The owner writes ten questions a customer might ask, such as 'electrician in Nambour for a switchboard upgrade' and 'who installs EV chargers near Nambour', and runs each one in ChatGPT, Gemini, Perplexity and a Google search that shows an AI Overview. Each run uses a private browser window, on the first Monday of the month, and the results go into a spreadsheet: named or not, linked or not, details right or wrong.

In this illustration the first run finds the business named in 3 of the 40 answers, and one answer quotes an old phone number. The owner traces the number to a directory listing nobody had updated and corrects it. A check of robots.txt shows a security plugin blocking OAI-SearchBot, so that gets allowed. The EV charger page is rewritten so its first paragraph answers the question directly. Over the next three monthly runs the business appears in more answers, while the switchboard question still names only competitors whose reviews mention switchboard work.

Two lessons carry over to real businesses. The fixes were ordinary: an access setting, one wrong listing, one clearer page, and a review gap that only real jobs can close. And a prompt log is a sample, not a score. Answers change between runs, users and locations, so the trend across months matters more than any single result. The table lists the signals worth tracking alongside the log, and what each one can't tell you. Our guide to comparing AI visibility reports covers paid tracking tools, and measuring ChatGPT referrals covers the analytics setup.

Does every business need it, and which offers to avoid

If your customers are starting to ask AI tools for recommendations, AI SEO deserves attention, but the priority depends on how they choose. An urgent call-out is usually decided in minutes from a map listing or a phone number. A considered job such as a renovation or solar system involves research over days or weeks, which gives AI tools more chances to shape the shortlist. Use the table as a starting rule, then check it against where your own enquiries come from.

Whatever the priority, some offers sold under the GEO or AEO label are worth turning down. Question these hard before you sign anything.

AI SEO and regular SEO aren't two separate budgets competing for the same money. Good AI SEO is built on solid SEO foundations, and the same work often lifts both. If you're weighing a proposal, the AI SEO section sets out what the work covers.

• Promised placement in ChatGPT, Gemini or AI Overviews. No agency controls whether an AI product names a business, so no one can guarantee it.
• An llms.txt file sold as the core of the strategy. Google's guide says you don't need new machine-readable files or AI text files to appear in Google Search.
• Hundreds of AI-written suburb or service pages. Google's spam policies list using generative AI tools to generate many pages without adding value for users as scaled content abuse.
• Hidden text or prompts aimed at AI tools. Google's spam policies treat content placed to manipulate search engines rather than be seen by visitors as hidden text abuse.
• Bought or invented reviews. The ACCC says it's against the law for a business to create fake or misleading reviews or to arrange for others to do so.
• Reports built around an 'AI visibility score' that don't list the questions, products and dates behind it.