AI answers and your site: how assistants pick what they cite
A growing share of the searches that once ended on a results page now end in an answer written by an assistant, with a handful of pages cited. These guides say where each assistant gets its pages, what it prefers, which crawlers to allow, and how to check if you are named.
By Théophile Louvart, founder of Porteur · Updated 14 September 2026 · Markdown
What changed, and what did not
Google's AI Overviews launched in the United States in May 2024 and reached more than a hundred countries in the year after; AI Mode, a conversational tab in Search, followed in 2025. ChatGPT gained search in October 2024 and opened it to everyone by early 2025. Perplexity has searched the web for every question since 2022. Claude, Gemini and Microsoft Copilot search and cite too. On a question, the answer now sits above the ten links, and the pages it cites take the clicks that remain.
What did not change is where the pages come from. Every one of these assistants retrieves from a web index (Google's, Bing's, or its own crawl) and cites the pages that answer the question directly. There is no submission form and no way to pay for a citation. A page that ranks on a query is the page most likely to be cited on it, and the practice of being named in answers, generative engine optimization, is most of it the same work as search, done on pages that state the answer early.
The term GEO comes from a 2023 research paper; the older name, answer engine optimization, dates from the featured snippet and voice assistant years. The guides use both, and say plainly what is documented by the vendors and what is inference.
Where each assistant gets its pages
| Assistant | Its index | Its crawlers | What it shows |
|---|---|---|---|
| Google AI Overviews and AI Mode | Google's own; AI Mode fans one question out into several sub-queries | Googlebot; Google-Extended is a training control that does not affect Search | Links in the answer and a sources panel; clicks count as web search in Search Console |
| ChatGPT search | Search partners (Bing among them) and OpenAI's own crawl | OAI-SearchBot for search, GPTBot for training, ChatGPT-User on demand | Inline citations and a Sources panel; product cards since 2025 |
| Perplexity | Its own index, refreshed for every question | PerplexityBot for the index, Perplexity-User on demand | Numbered citations and a sources list |
| Claude | Reported to be built on Brave Search | Claude-SearchBot for search, Claude-User on demand, ClaudeBot for training | Inline citations with links |
| Gemini | Google Search, when it grounds an answer | Googlebot; Google-Extended controls training and grounding | Links under the answer |
| Microsoft Copilot | Bing | Bingbot; IndexNow for freshness | Citations from Bing's index |
Two consequences follow. A site absent from Bing is absent from Copilot and from part of ChatGPT: Bing Webmaster Tools, with its one-click import from Search Console, is the way in. And the answering crawlers do not run JavaScript: a page whose text arrives by script after load is invisible to all of them, whatever Google renders.
What gets a page cited
The same things across the assistants: the answer stated in the first sentences under a heading that matches the question; lists, tables and figures an answer can quote; the product named the same way on every page and on the third-party lists the assistants already read (review sites, alternatives pages, roundups); a visible date and a named author; and access, which means the answering crawlers allowed in robots.txt. Being on the lists an assistant already cites for a category is the most reliable route into its answers, which is why the launch and listing guides belong to this theme as much as to growth.
Measuring it is manual as of 2026: a list of the questions a buyer asks, put to each assistant on a schedule, with the names and links logged; the referrer segment in analytics (chatgpt.com, perplexity.ai, copilot.microsoft.com, gemini.google.com); and tracking products that automate the asking. Google's AI clicks arrive as ordinary search clicks and are not separated in Search Console.
The order to read the guides
- First: generative engine optimization, then GEO against SEO, then what AI SEO means. They fix the words and what is documented.
- Then access: robots.txt for AI crawlers (in the technical theme) and llms.txt, the map a site can write for agents.
- Then one assistant at a time: AI Overviews, AI Mode, ChatGPT search, Perplexity, Claude, Gemini, Copilot. Each says where its pages come from and what it favours.
- Then the page: writing pages assistants can cite, SEO for ChatGPT, answer engine optimization.
- Then measurement and spend: AI visibility tracking, GEO tools, and whether you need a GEO agency.
What to do in an afternoon
Run the AI crawler access checker on your home page: it tests every vendor token against your robots.txt and says whether the page's text is readable without JavaScript. Write an llms.txt with the generator, an hour's work that costs nothing. Open your pricing page and your top guide and put the answer in the first two sentences under the heading. Then write down fifteen questions a buyer would ask an assistant about your category, ask each assistant, and note who is named. That list is your GEO plan: the pages that are named are the ones to be on, and the questions where a rival is named without you are the pages to write.
Where the free check fits
The free check reads what robots.txt does to each AI crawler and whether your pages are readable without scripts. The report goes further: it asks the assistants the questions of your category, reads who is named and cited on each, and writes the pages and the listings that change it, as briefs an agent can take. Being named by the AIs is one of the report's eleven chapters.
The guides, in the order to read them
28 guides on the theme, each on one question, each ending in your own figures.
- Generative engine optimization (GEO): how to be named by AI answersGuide
How to get your site named in AI answers: sources, crawlers, lists, structure, llms.txt, comparisons, and how to measure by asking assistants.
- GEO vs SEO: what changes, what does notGuide
GEO names you in AI answers, SEO ranks you in results. Here is what overlaps, what changes, and five steps to win citations without ditching SEO.
- What AI SEO means, and the three things it can refer toGuide
AI SEO means three things: using AI to do SEO work, being cited in AI answers, and Google’s AI in Search. Here is what to do on each.
- llms.txt: what it is, examples, and how to write yoursGuide
llms.txt is a Markdown file at /llms.txt that maps your site for language models. See the format, a worked example, and write yours in twenty minutes.
- How to appear in Google’s AI OverviewsGuide
What overviews cite, which queries trigger them, and how to structure pages that get linked. Steps to find your openings and measure them.
- Google AI Mode: what it is and how a small site gets citedGuide
What AI Mode is, how it picks citations, how clicks appear in Search Console, and the page changes that make a small site quotable.
- How ChatGPT search picks and cites sourcesGuide
See how ChatGPT search finds pages and cites them, what your page needs to be picked, and how to check if your site is named.
- How Perplexity picks its sources, and how to become oneGuide
How Perplexity finds, indexes and cites pages, what it favours, the crawler tokens to allow, how to see your mentions, and a monthly GEO routine.
- Claude’s web search: how it fetches, cites and what to allowGuide
How Claude searches and cites, which Anthropic tokens to allow in robots.txt, and how to check if Claude names your site in answers.
- Gemini and your site: grounding, Google-Extended, and what to allowGuide
What AI Mode and Gemini are, how grounding and Google-Extended work, what to allow in robots.txt, and the simplest way to be cited.
- Microsoft Copilot cites from Bing: the way in is Bing Webmaster ToolsGuide
If Copilot is not citing you, fix Bing first. Set up Bing Webmaster Tools, add IndexNow, allow Bingbot, and write pages Copilot can quote.
- Writing pages AI assistants can citeGuide
Want citations from ChatGPT, Perplexity, Google AI Mode and Copilot? Use this page pattern, allow the right crawlers, and test by asking the assistants.
- SEO for ChatGPT: how assistants choose what to citeGuide
How ChatGPT with search finds pages and decides what to cite, what to fix on your site, and how to test it each month in under 30 minutes.
- Answer engine optimization: the older name for being the answerGuide
AEO in 2026: what it is, how it overlaps GEO, the page format assistants cite, the crawlers to allow, and how to check if you are named.
- How to track your AI visibility without a subscriptionGuide
Track AI visibility without tools: list 20 buyer questions, ask each assistant, log names and links, segment referrers, and know when automation is worth it.
- GEO tools: what they measure and what you can check yourselfGuide
What GEO tools really measure, how they sample prompts, and a simple monthly method to track AI answers yourself and act when rivals win.
- Do you need a GEO agency? What they do, and what you can do yourselfGuide
What GEO agencies actually sell, what is just SEO in new clothes, the checks to run yourself today, and when hiring one makes sense.
- AI-written content and Google: the policy, in Google’s wordsGuide
Google does not penalise AI writing as such. It acts on pages made at scale with nothing of their own. Here is the policy and where the line sits.
- AI content detectors: what a score means, and why you should not rewrite for itGuide
Detectors report a probability, not proof. They flag human writing and miss edited AI text. What Google says, and what to check on a page instead.
- What AI Overviews do to your traffic, and how to tellGuide
Google reports overview clicks inside web search totals, so nobody can isolate the effect. Here is the pattern to look for, and what to do next.
- AI-generated images on a product siteGuide
Where a generated image is fine, where only a real screenshot will do, what Google reads about an image, and what provenance metadata records.
- A writing workflow with AI that does not produce fillerGuide
Five steps: the facts, the brief, the draft, the human pass, the read-back. The shapes to copy, and the failure each step prevents.
- The best AI visibility tools, and the manual method that costs nothingGuide
See how AI assistants mention you, choose a tracker that fits, and run a zero-cost sheet to log citations across markets. Clear steps and criteria.
- ChatGPT for SEO: what it does well, what it inventsGuide
ChatGPT is good for briefs, outlines and rewrites, and useless for volumes and positions. The jobs it does well, four prompts, and the checks.
- Editing an AI draft: the eight things to cutGuide
The eight tells of an unedited draft, with a before and after for each, and the order to work through them so a page ends up worth reading.
- Is SEO dead? What changes with AI answers, and what does notGuide
Assistants changed where the click happens, not how pages are found. What actually moved, what did not, and the five things to do differently.
- Prompts for SEO work that are worth keepingGuide
A prompt is a brief: audience, facts, constraints, shape. Six prompts written out in full, what each is for, and where each one lies to you.
- When not to use AI for a pageGuide
Six pages a model cannot write: the case study, the opinion, your method, security, pricing, your story. Why, and what it can still do for them.
The free tools
Each checks one thing on one page or one site, without an account, and says what it found in words.
- AI crawler access checkerFree tool
Paste a site or a page and see, crawler by crawler, whether it may read it: root and page, the rule that decides, the text without JavaScript, llms.txt.
- llms.txt generatorFree tool
Write an llms.txt from your site or from a few fields, in sections, and check the one a site already has against the proposal, link by link.
- robots.txt generatorFree tool
Build a robots.txt that says which AI crawlers may train on your pages and which may cite them, with the sitemap line and the paths closed to everyone.
- robots.txt testerFree tool
Paste a site, paths and a crawler: what its robots.txt allows by the rules Google reads with, the rule that decided each, why it won, and the file mended.
- SERP checkerFree tool
Type a search and see the results page as Google draws it now in six markets: the top ten sites, the People also ask questions, and who the AI Overview cites.
The words
What each term means for a small site, and what to do about it.
- Generative engine optimization (GEO)Glossary
GEO means getting named and cited in AI answers. Learn how it differs from SEO, what to measure, which crawlers to allow, and what work to ship.
- AI OverviewGlossary
AI Overviews are Google’s generated answers above results. See when they appear, what they cite, how to measure impact, and how to get cited.
- AI citationGlossary
Learn what AI citations are, where they come from, how to get them, and how to measure clicks from ChatGPT, Perplexity and Google AI Overviews.
- AI crawlerGlossary
An AI crawler is a bot that fetches your pages for an AI. See the main bots, how to spot them in logs, and how to allow or block them well.
- ChatGPT searchGlossary
What ChatGPT search is, where results come from, how to allow or block its crawler, and how a small site gets cited and measures impact.
- llms.txtGlossary
llms.txt is a Markdown file at /llms.txt that maps your site for AI agents. See what to put in it, how to publish it, and the traps to avoid.
- Query fan-outGlossary
Query fan-out expands one question into many sub-queries. See how it picks citations, how to spot it in your data, and how to write for it.
- Brand mentionGlossary
What brand mentions are, why they matter for a small site, how to find and measure them, and how to turn unlinked mentions into links.
- Entity SEOGlossary
Entity SEO makes search engines recognise your product or company as a real thing, not just words. Here is what to set up and how to check it.
- Zero-click searchGlossary
Zero-click searches give answers on the results page. See how they show in Search Console, what to change on your pages, and when to ignore them.
Questions
No. Every assistant retrieves pages from a web index and cites the ones that answer the question; there is no submission form and no paid placement for citations. Presence on the lists and comparisons the assistant already cites is the reliable route.
Not by itself. llms.txt is a map of the site for agents, a proposal from 2024 that no major vendor has committed to reading; it costs an hour and helps an agent that does read it. Citations come from the index and the page.
On the queries where one opens, yes: searches with an overview send fewer clicks to the organic results below it, and the clicks that remain go to the cited pages. Being one of those pages is the only way to keep the traffic on a question.
Ask it. There is no report of citations as of 2026; the method is a list of buyer questions asked to each assistant on a schedule, with the names logged, plus the assistant referrers in your analytics.
Check my site, free
Paste your URL and the check reads what your robots.txt does to every AI crawler and whether your pages can be read without scripts; the report asks the assistants your category's questions and says who they name.
- Free check, no card
- Read-only, your own accounts
- Readable by your agent