What is AEO, and how is it different from SEO?
- Author:
- Max Rosberg Somerlock
- Published:
- 10 min read
Short answer
AEO (answer engine optimisation) is the work of becoming the source that AI answers cite, in Google's AI Overviews and AI Mode, ChatGPT, Copilot and Perplexity. It rests on the same foundation as SEO: pages must be crawlable, indexed and the best answer. The difference is the goal and the measure: citations in answers, not only rankings and clicks.
AEO, GEO and SEO: one foundation, different goals
AEO and GEO describe the same job: getting your content used in answers that AI writes, not only listed among links. AEO stands for answer engine optimisation, GEO for generative engine optimisation, and you will also see ‘AI search optimisation’. The term GEO was introduced in a 2023 research paper by Aggarwal et al..
Google does not treat it as a separate discipline. Its guide to generative AI features in Google Search says that optimising for generative AI search is optimising for the search experience, ‘and thus still SEO’. We agree on the foundation and see the difference in the goal. SEO is judged on rankings and clicks. AEO is judged on how often your content is used and cited when someone gets a finished answer, often without clicking at all.
How do answer engines choose which sources to cite?
They fetch pages through a search index or their own crawler, then write the answer from what they retrieved. A page they cannot fetch cannot be cited. Here is how each service describes its own approach:
- Google AI Overviews and AI Mode are rooted in Google's core ranking systems. They retrieve pages from the search index and run several related searches at once, which Google calls query fan-out. A page must be indexed and eligible to show with a snippet, and the site must not be excluded from generative AI features in Search Console.
- ChatGPT search relies on the OAI-SearchBot crawler. OpenAI's crawler documentation says sites that opt out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links. GPTBot is a separate crawler for model training.
- Microsoft Copilot: since February 2026, Bing Webmaster Tools has shown how often a site is cited in Copilot and in Bing's AI-generated summaries, and which phrases the AI used to retrieve the content.
- Perplexity uses PerplexityBot to surface and link sites in its answers. Perplexity says the crawler is not used to gather content for AI foundation models.
The common thread: the AI chooses from pages it can already reach and understand. Technical SEO is the entry ticket for AEO, not a separate track.
What happens to clicks when AI answers?
People click through less when an AI summary appears. Pew Research Center analysed the browsing data of 900 US adults in March 2025 and published the results in July 2025:
- On searches with an AI summary, users clicked a traditional result in 8% of visits, against 15% without one.
- Links inside the summary itself were clicked in 1% of visits.
- Users ended their browsing session on 26% of pages with a summary, against 16% without.
- 18% of all Google searches in the study produced a summary, rising to 60% for queries that began with a question word such as who, what, when or why.
This is US data on Google from spring 2025. Google's guide says visitors arriving from AI features may be more inclined to engage, spend more time or convert, but gives no figures there. For a business, the practical reading is to expect fewer clicks on simple factual questions and to measure both whether you are cited and what the remaining visits lead to.
What does the GEO research show?
The paper that named the field, GEO: Generative Engine Optimization, was presented at the KDD 2024 conference. The authors built a benchmark of 10,000 queries, had a language model answer each one from the top five Google results, and rewrote the source pages using nine different methods.
- Adding citations, quotations from credible sources and statistics worked best, lifting visibility in the answers by about 40%.
- Keyword stuffing brought little or no improvement, and in a test on Perplexity it did 10% worse than the unedited text.
- Lower-ranked pages gained most. Citing sources raised visibility by 115.1% for the page in fifth place, while the top result lost 30.3%.
Read it with its limits in mind. Visibility meant the share of the answer text drawn from a given source, not clicks or sales. The Perplexity test covered 200 queries, with the source text uploaded as files. The authors themselves note that the methods may need to adapt as generative engines evolve. Treat it as direction rather than a recipe: be specific, back up your claims and cite your sources.
What to ignore, and what to avoid
Several popular AEO tips do not survive a look at the primary sources:
- FAQ markup as a shortcut. From 2023 Google showed FAQ rich results only for well-known government and health sites, and since 7 May 2026 they no longer appear at all. Visible FAQ sections can still help: Bing says clear headings, tables and FAQ sections make content easier for AI systems to reference accurately.
- llms.txt. It is a proposal from 2024, not a standard. Google says Search ignores it. We have found no statement from OpenAI, Microsoft or Perplexity that their answer engines use it, so treat any benefit as unconfirmed.
- Blocking Google-Extended to stay out of AI Overviews. Google's crawler documentation says Google-Extended does not affect inclusion in Google Search. It governs training of Gemini models and grounding in the Gemini app and Vertex AI.
- Chunking or rewriting text for AI. Google says neither is needed.
- Bought mentions and a page for every query variant. Google says inauthentic mentions are less helpful than they seem, that mass-produced pages aimed at AI answers break its scaled content abuse policy, and that its spam policies apply to AI responses too.
Manipulation is already happening. In February 2026 Microsoft's security researchers described ‘Summarize with AI’ buttons that try to plant hidden instructions in a user's AI assistant, such as remembering a company as a trusted source. They found more than 50 distinct prompts from 31 companies across 14 industries. Shortcuts like these erode trust in both the answers and the sender.
How do you measure visibility in AI answers?
Combine the official reports with your own spot checks:
- Google Search Console has a generative AI performance report showing impressions in AI Overviews and AI Mode by page, country and device. It reached all sites on 31 August 2026. The same traffic also counts in the standard Performance report.
- Bing Webmaster Tools shows citations in Copilot and Bing's AI summaries, the pages cited and the grounding queries behind them.
- Analytics: track referrals from chatgpt.com, perplexity.ai and copilot.microsoft.com as a channel of their own.
- Spot checks: each month, put the questions your customers ask to ChatGPT, Perplexity, Copilot and AI Mode, and log which sources are cited. The same question can return a different answer later, so look for patterns.
Be wary of tools that claim insight into how AI ranks you. Google states that no third-party tool has access to its internal ranking or AI systems.
Where should you start?
With the work that makes a page better for readers and search engines alike. Our checklist:
- Answer first. Put a short, direct answer at the top and the detail below it.
- One main question per page, with follow-up questions as clear subheadings. Do not create a new page for every phrasing.
- A clear sender. Show who wrote the page and what the company is called and does, using the same name everywhere.
- Structured data that matches the visible text. It is no shortcut to AI answers, but Google still recommends it because it can earn rich results.
- Let the right crawlers in. Check robots.txt, your firewall and your CDN for Googlebot, Bingbot, OAI-SearchBot and PerplexityBot.
- Fast, server-rendered HTML, so the text is in the page source. This is how we build new sites on Next.js and Sanity.
- Dated, sourced facts. Show when the page was last updated, link to sources and keep figures current. Bing recommends IndexNow to report changes quickly.
Frequently asked questions
- No. Google's AI Overviews and AI Mode run on the same index and ranking systems as ordinary results, and Google calls optimising for AI search simply SEO. ChatGPT, Copilot and Perplexity also fetch pages from the web before they answer. What changes is how success looks: more often a cited source inside an answer, less often a click. A sound technical base, useful content and a clear sender still decide the outcome.
- Not to appear in AI search. llms.txt is a proposal first published in 2024 and last revised in August 2026, not a standard; its aim is to help AI agents use a website. Google says Google Search ignores the file and that it neither helps nor harms. We have found no statement from OpenAI, Microsoft or Perplexity that their answer engines read it. Spend the time on your content first.
- There is no evidence that it does. OpenAI's crawler documentation does not mention structured data; it is about letting OAI-SearchBot in. Google says no special schema.org markup is needed for AI Overviews or AI Mode. Markup that matches what is visible on the page is still worth having, because it can earn rich results in Google. Treat it as basic hygiene, not a shortcut into AI answers.
- Test it and measure it. Ask ChatGPT, with search switched on, the questions your customers ask, such as which system suits a particular industry. Note whether you are mentioned, how you are described and which sources are cited. Repeat monthly, because answers change over time. Track traffic from chatgpt.com in your analytics. For Copilot, Bing Webmaster Tools reports citations; for Google, Search Console has a generative AI report.
- Separate search from training. OpenAI says sites that block OAI-SearchBot are not shown in ChatGPT search answers, and Perplexity recommends allowing PerplexityBot. GPTBot and Google-Extended relate to training and can be blocked without disappearing from ChatGPT search or Google Search. If you want to keep your site out of Google's AI features, Search Console has offered a setting since August 2026, but you then get no impressions or traffic from them.
- No service promises a timeframe, and Google says indexing and serving are never guaranteed. Technical changes are quick: OpenAI and Perplexity both cite about 24 hours for robots.txt changes, and Google's Search Console exclusion usually takes effect within 1–2 days. Getting Google to recrawl a changed page can take anywhere from several days to several months. Becoming the cited source takes as long as becoming the best page on the subject.
- The principles are the same; the questions differ. Pew found that longer searches trigger AI summaries far more often: 53% of searches of ten words or more, against 8% of one- or two-word searches. In our view, B2B buyers often ask exactly these detailed questions, about pricing, integrations, security and lead times. That is where specific, well-sourced answers earn their place, far more than broad sales copy.
Sources
- [1]Optimizing your website for generative AI features on Google Search · Google Search Central, 10 Jul 2026
- [2]Latest documentation updates · Google Search Central, 26 Oct 2026
- [3]Google's common crawlers · Google, 15 Oct 2026
- [4]Overview of OpenAI Crawlers · OpenAI
- [5]Introducing AI Performance in Bing Webmaster Tools Public Preview · Microsoft Bing, 10 Feb 2026
- [6]Google users are less likely to click on links when an AI summary appears in the results · Pew Research Center, 22 Jul 2025
- [7]GEO: Generative Engine Optimization · arXiv (Aggarwal et al., KDD 2024), 28 Jun 2024
- [8]Manipulating AI memory for profit: The rise of AI Recommendation Poisoning · Microsoft Security, 10 Feb 2026

Got a project in mind?
Tell us briefly what you want to build, and we'll get back to you and take it from there.