Answer engine optimisation (AEO) is the work of making your business easy for AI answer engines to find, understand and quote. SEO earns a page its place in the index and the rankings; AEO makes what is on the page worth quoting. For its own search, Google says AEO is still SEO, and that Search ignores llms.txt.
We run AEO for clients as part of our AI SEO agency work and practise it on this site, so this guide uses our own structured data, robots.txt and llms.txt as worked examples. It also sets out what Google, OpenAI, Anthropic, Perplexity and Microsoft each say their systems read, checked on 1 October 2026.
What does AEO stand for?
AEO stands for answer engine optimisation. An answer engine is a search tool that replies with an answer rather than a list of links. ChatGPT search, Perplexity, Gemini, Claude, Microsoft Copilot and Google's own AI Overviews and AI Mode all work this way. Google's guide uses the term the same way, next to GEO (generative engine optimisation), as two names for work aimed at visibility in AI search.
The two labels describe the same job. If you are weighing them up, our guide to generative engine optimisation vs SEO covers GEO and the research behind it.
What is AEO vs SEO?
AEO and SEO share a foundation and differ in what you polish. SEO is judged on whether a page ranks and earns the click. AEO is judged on whether an AI answer uses your facts and names your business, whether or not anyone clicks. Day to day, the difference shows up in seven jobs.
| Job | The SEO habit | What AEO adds |
|---|---|---|
| Research | Keyword volumes for each page | The questions buyers ask an assistant, and the follow-up questions it searches for |
| Page structure | A title and headings that match the search | A direct answer in the first one or two sentences of each section |
| Evidence | Content that covers the topic | Numbers with a source and a date, quotations and first-hand experience |
| Entity | A brand name and a Google Business Profile | The same facts about the business on the site, in the structured data and on every profile |
| Structured data | Markup for rich results | Markup that matches the visible page and links the business, its people and its services |
| Crawlers | Googlebot and Bingbot | The AI search crawlers too, decided separately from training crawlers |
| Measuring | Rankings, clicks and organic sessions | AI Overview and AI Mode impressions in Search Console, the AI Assistant channel in GA4 and Copilot citations in Bing Webmaster Tools |
Sources for the crawler and measurement rows: Google, OpenAI, Perplexity, Anthropic and Microsoft documentation, checked 1 October 2026.
What is AEO in AI?
In AI terms, AEO is preparing your content for the moment an assistant goes looking for sources. Most answer engines work in two stages. First they search an index for relevant pages, then a language model reads the best of them, writes the answer and links the pages it relied on. Google calls the first stage grounding. Its AI features also break one question into several searches at the same time, a step it calls query fan-out.
Each engine has its own crawler for that search stage. OpenAI's is OAI-SearchBot, and a site that blocks it is left out of ChatGPT's search answers, although it can still turn up as a navigational link. Perplexity uses PerplexityBot and recommends allowing it. Anthropic says blocking Claude-SearchBot may reduce your visibility in Claude's search results. That is why AEO starts with crawler access, not copywriting.
How to do answer engine optimisation
Work through five jobs in order. Each one below uses our own site as the worked example, so you can check what we did.
1. Let the search crawlers in, and decide training separately
Our robots.txt allows every major AI crawler, for training and for search, and keeps all of them out of private areas such as client dashboards. That is a business decision, not a requirement. A business that wants to stay out of AI training but stay in AI answers can block the training crawlers (GPTBot for OpenAI, ClaudeBot for Anthropic and Google-Extended for Gemini) and leave the search crawlers (OAI-SearchBot, Claude-SearchBot and PerplexityBot) alone. Google says Google-Extended has no effect on inclusion or ranking in Google Search.
Google added one more switch in 2026. Search Console now has a Search generative AI setting: on Include, the default, your content can appear in AI Overviews and AI Mode; on Exclude, it can't. Check it before you do anything else.
2. State who you are, the same way everywhere
Answer engines need to resolve your business as one entity: a name tied to consistent facts. Our structured data states those facts once and has every other piece of markup point back to them.
| Property | What our markup says | Why it matters |
|---|---|---|
| @id | One identifier for the business, which every service, article and author points back to | Joins every mention into one entity |
| legalName | AM Marketing Group Pty Ltd | Connects the trading name, Profit Geeks, to the registered company |
| address | Sydney, NSW, Australia | Matches the head office named on the site |
| areaServed | Australia, plus five capital cities | States where we work in a form machines read |
| foundingDate | 2016 | Matches the founding year stated across the site |
| Person | Andy McMaster, Founder, working for the same business identifier | Ties the author of each guide to the business |
Source: the JSON-LD on profitgeeks.com.au, checked 1 October 2026.
The rule that matters more than any property: the markup has to say what the visible page says. Google asks for structured data that matches the visible text, and says there is no special schema you need to add for its AI features. If your site and your markup disagree about your address or your services, fix the site first.
The same goes for profiles. The sameAs property lists other pages that describe the same business, such as social and directory profiles. Only list profiles that are live and say exactly the same thing as your site.
3. Answer the question in the first sentence or two
Every question heading in this guide is answered in its first one or two sentences. That is for readers first: Google's guide says there is no need to break content into tiny chunks for AI, and that both shorter and longer pages can work. But a plain answer at the top of a section is also the easiest passage for an assistant to lift, and the easiest for a reader to check.
4. Put evidence on the page
Give the engine something to cite. The GEO research found pages were quoted more often once sources, quotations and statistics were added, and Google's guide asks for first-hand, non-commodity content. For most businesses the cheapest evidence is their own: what jobs cost, how long they take, what was measured and when. The tables in this guide are examples.
5. Measure what AI search sends you
Three reports cover it. In Search Console, the Generative AI performance report counts your impressions in AI Overviews and AI Mode. GA4's AI Assistant channel shows visits from assistants such as ChatGPT, Gemini and Claude. Bing Webmaster Tools' AI Performance report shows citations in Copilot. Our ChatGPT SEO guide walks through each one.
What is llms.txt?
llms.txt is a Markdown text file at the root of a website that summarises what the site is and links to the pages worth reading. Jeremy Howard proposed it on 3 September 2024 at llmstxt.org and last revised the proposal on 10 August 2026. Its stated purpose is to help AI agents use a website, such as a coding assistant reading documentation or a chat assistant reading pages to answer a question.
The format is simple. A heading with the site's name is the only required part. A short summary in a quote block usually follows, then sections of links with a line on each, and by convention an "Optional" section for pages an agent can skip.
Does llms.txt actually work?
Not for Google Search, and none of the AI search companies we checked says it uses llms.txt to choose sources. This is what each one's documentation says, checked on 1 October 2026.
| Company | What its documentation says |
|---|---|
| Google Search (AI Overviews, AI Mode) | Search ignores llms.txt files, and creating one "will neither harm nor help" your visibility or rankings. Source: Search Central guide, updated 10 July 2026. |
| Google Chrome (Lighthouse) | Checks for an llms.txt among its agentic browsing audits. A missing file is marked not applicable, and the audit fails only on a server error. Source: Chrome for Developers, updated 5 May 2026. |
| OpenAI (ChatGPT search) | Its crawler page covers its search, training and user crawlers and robots.txt, and does not mention llms.txt for search. Source: OpenAI crawler documentation. |
| Anthropic (Claude) | Its crawler article covers its training, user and search crawlers, and does not mention llms.txt. Source: Anthropic help centre, updated 7 April 2026. |
| Perplexity | Its bot page covers its search crawler and user agent, and does not mention llms.txt for search. Source: Perplexity documentation. |
| Microsoft (Copilot, Bing) | Its AI Performance announcement points to IndexNow for keeping content fresh in search and AI answers, and does not mention llms.txt. Source: Bing Webmaster Blog, 10 February 2026. |
Two details are worth knowing. OpenAI and Perplexity both publish llms.txt files for their own developer documentation, so coding tools can read their docs; that is a different job from choosing sources for a search answer. And Google's guide adds that it is "completely fine" to keep one for other services that use it.
So the honest answer: as a ranking or citation lever, no vendor documentation supports llms.txt. As a clean map of your site for AI agents, it does what it says. Fittingly, when we searched "llms.txt" in Australia on 1 October 2026, Google's AI Overview cited the two pages ranking first and second: the proposal itself and Chrome's Lighthouse documentation.
Why we publish an llms.txt anyway
Because it costs almost nothing and does no harm. Agents that read it get a clear map of the site instead of crawling around it, and Chrome's Lighthouse audit checks for it. Writing it also made us state who we are in one paragraph, which is the same entity work that matters on every page. Ours is at /llms.txt: a summary with our legal name, ABN, offices, founding year and standing offer, then links to our services, tools and guides.
What we won't do is sell it as a ranking tactic. Google lists AI text files among the "AEO/GEO hacks" you can ignore for its search, and we agree.
What should be in an llms.txt file?
The business's name as a heading, a short summary, and links to the pages that matter, each with a line of context. For a business site, that means:
- A heading with the business name, the only required line.
- A quote-block summary: legal name, location, what you sell and who you sell it to.
- Any facts an assistant would otherwise guess, such as service areas, how you price and how to book.
- Sections of links: services, locations, guides, tools and contact details.
- An "Optional" section for secondary pages, if you have them.
Keep every fact identical to the site and the structured data, and update the file when your offer changes. An out-of-date llms.txt gives an agent a second version of the truth, which is worse than giving it none.
How do you check or submit an llms.txt file?
There is nowhere to submit one: none of the documentation above describes a submission process. To check yours, open yourdomain.com.au/llms.txt in a browser. It should return plain text with a 200 status, not an error page or a redirect to your home page. Chrome's Lighthouse will also flag it if your server errors when the file is requested.
Is AEO replacing SEO?
No. Google's guide says SEO best practice still applies, because its AI features are built on the same ranking and quality systems as ordinary search, and only pages it has indexed and could show with a snippet can be linked from an AI Overview. What has changed is where the visibility shows up: in our 2026 study of Australian search results, AI Overviews appeared on 433 of the 498 marketing searches we checked.
We can run this on your site as part of our AI SEO agency service, scored on enquiries rather than mentions. Start with a free 30-minute profit audit: a senior operator reviews your site and your search data, and you keep the written notes.
Next step
Want this kind of work in your business?
Engagement intake is capped, senior operators run every account. If your attribution is leaking and your reports have stopped making sense, the next step is a 30-minute call.
