/ โ† Learning Hub / AEO Fixes
On this page
  1. How AI assistants choose their sources
  2. The crawlers AI assistants use
  3. The fix guides
  4. What the findings mean
  5. Writing pages AI assistants can quote
  6. A sensible order of work
  7. Questions and answers

How AI assistants choose their sources

An AI assistant can answer from two places. The first is what the model learned during training, which stops at a cut-off date and cannot be changed after the fact. The second is a live search: the assistant runs a web search, reads a handful of pages and writes an answer from them. Live search is where most citations with links come from, and it is the part a website can influence directly.

For a page to be used in a live answer, four things have to be true:

  1. It can be found. The assistant's search system has to know the page exists. Some assistants use their own index; others lean on Google or Bing.
  2. It can be read. The assistant's crawler must be allowed by robots.txt and the firewall, and the content must be in the HTML rather than loaded later by JavaScript.
  3. It answers the question. The page has to contain a clear, direct answer that matches what was asked, not just a page broadly about the topic.
  4. It is trusted. Assistants favour sources that other sites mention and link to, that show who wrote them, and whose facts agree with other good sources.

Answers also vary. The same question asked twice can produce different sources, and answers change as models and indexes are updated. That is why AI visibility has to be measured across many prompts and over time, not judged from a single screenshot.

๐Ÿ’ก There is no way to pay for a place in an AI answer and no tag that guarantees one. Everything that works is about being findable, readable, useful and trusted.

The crawlers AI assistants use

Each AI company runs more than one crawler, and they do different jobs. Blocking the wrong one can remove you from AI answers without you noticing.

CompanyCrawlerWhat it is for
OpenAIOAI-SearchBotFinding pages to show in ChatGPT search answers.
OpenAIChatGPT-UserFetching a page when a user's request needs it.
OpenAIGPTBotCollecting content for training future models.
PerplexityPerplexityBot, Perplexity-UserIndexing pages for Perplexity answers, and fetching pages for a user's request.
AnthropicClaude-SearchBot, Claude-User, ClaudeBotSearch results, user-requested fetches, and training.
GoogleGooglebot, Google-ExtendedGooglebot feeds Google Search, including AI Overviews. Google-Extended only controls use in Gemini models.
MicrosoftbingbotFeeds Bing, which Copilot uses for its answers.

A common choice is to allow the search and user-request crawlers, so the site can appear in answers, while deciding separately about the training crawlers. Blocking Googlebot or bingbot to keep content away from AI would also remove the site from ordinary search, so it is rarely what anyone intends.

# robots.txt: appear in AI answers, decide separately about training
User-agent: OAI-SearchBot
Allow: /

User-agent: ChatGPT-User
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

Remember that firewalls and bot protection can block these crawlers even when robots.txt allows them. Search the server access log for each name to see whether they are getting through.

The fix guides

Each guide covers one finding from start to finish: what it means, what causes it, how to confirm it, a step-by-step fix and the questions people ask most.

By finding

Fixing invisibility in ChatGPT covers the OpenAI crawlers, ChatGPT search, and why a site that ranks in Google can still be missing from ChatGPT answers. Fixing invisibility in Perplexity covers PerplexityBot, how Perplexity picks the sources it links, and why pages that answer the question early do well. Fixing invisibility in Gemini covers the link between Google Search and Gemini, and what Google-Extended does and does not control.

Fixing low citation frequency is for sites that appear sometimes but not often, and covers becoming the source that other sources cite. Fixing wrong attribution covers assistants that credit you with claims you never made, or state wrong facts about your business, and what you can realistically do about it.

Fixing schema for AI answers covers the structured data that helps assistants identify who you are and what a page contains. Fixing content structure covers writing so that a clear answer can be lifted from the page and still make sense on its own. Fixing weak brand mentions covers the mentions on other sites, with or without links, that tell assistants your business is relevant.

By assistant

The assistant guides go deeper into each one: ChatGPT, Perplexity, and Gemini, Claude and Copilot, including where the work overlaps and where each assistant needs something different.

What the findings mean

FindingWhat it usually meansStart with
Invisible in every assistantThe site is blocked, too new, or does not answer the questions being tested.Crawler access, then content structure.
Invisible in one assistant onlyThat assistant's crawler is blocked, or it relies on an index where the site is weak.The guide for that assistant.
Mentioned but not linkedThe assistant knows the brand but uses other pages as sources.Content structure and citation frequency.
Cited rarelyOther sources answer the question more directly or are trusted more.Citation frequency and brand mentions.
Competitors cited insteadTheir pages answer the exact question, or are mentioned more widely.Compare their page with yours for that prompt.
Wrong facts about youOld, inconsistent or third-party information is being used.Wrong attribution, then schema.

Writing pages AI assistants can quote

Assistants build answers from short passages. A page that buries its answer in the fifth paragraph, or spreads it across a table, an image and a pop-up, is hard to quote. A few habits make a large difference:

None of this is different from writing a good page for people. Assistants reward the same clarity that readers do.

A sensible order of work

  1. Run the AI Visibility Tracker with the questions your customers actually ask, and note where you are cited, mentioned or missing.
  2. Check crawler access: robots.txt, the firewall and the access log for each assistant's crawlers.
  3. Check that the pages answering those questions exist, are indexed in Google and Bing, and have their content in the HTML.
  4. Rewrite those pages so the answer comes first and each key passage stands on its own.
  5. Add Organization, Person and Article markup so assistants can identify who you are and who wrote each page.
  6. Build mentions on sites your customers read: reviews, industry directories, local press and forums.
  7. Measure again after a few weeks, across the same prompts, and compare the trend rather than single answers.

For the full reference, see the AEO Guide, the example report or the beginner tutorial.

Questions and answers

1. What is AEO?

AEO stands for answer engine optimisation. It is the work of making a website easy for AI assistants such as ChatGPT, Perplexity, Gemini, Claude and Copilot to find, read, trust and quote when they answer questions. It overlaps heavily with good SEO: pages still need to be crawlable, indexed and useful. The difference is a stronger focus on direct answers, passages that stand alone, clear authorship and mentions across the wider web.

2. Why is my site missing from ChatGPT answers?

Usually for one of three reasons. OpenAI's search crawler, OAI-SearchBot, may be blocked by robots.txt or a firewall; the site may have no page that directly answers the question being asked; or other sources may be more widely mentioned and trusted. Check your access log for OAI-SearchBot and ChatGPT-User, then compare your page with the sources ChatGPT does cite for that question and close the gap in clarity, detail and trust.

3. Should I block GPTBot and other AI crawlers?

That is a business decision, but separate the crawlers by job. Training crawlers such as GPTBot and ClaudeBot collect content for future models. Search and user-request crawlers such as OAI-SearchBot, ChatGPT-User, PerplexityBot and Claude-SearchBot fetch pages to answer questions now. Blocking the search crawlers removes you from those answers. Many sites allow the search crawlers and decide separately whether to allow training, and set each crawler in robots.txt by name so the choice is explicit.

4. Does Google-Extended stop my site appearing in AI Overviews?

No. Google-Extended only controls whether your content is used for Gemini models. AI Overviews in Google Search are built from the normal Google index, which is fed by Googlebot, so a page that appears in Google Search can appear in AI Overviews whatever Google-Extended says. Blocking Googlebot would remove the page from AI Overviews, but it would also remove it from ordinary Google results, which is rarely what anyone wants.

5. Why do AI answers change every time I ask?

Because AI assistants generate each answer fresh. The search they run can return different pages, the model can choose different passages, and both the model and the index are updated over time. A single answer therefore proves very little. Measure the same set of questions regularly and look at how often you are cited or mentioned across many runs, which is what the AI Visibility Tracker reports over weeks and months.

6. Does schema markup help with AI visibility?

It helps assistants identify things, but it is not a shortcut. Organization markup with sameAs links tells them which profiles belong to your business, Person markup identifies authors, and Article markup gives dates and headlines. That makes it easier to attribute information to you correctly. It does not replace a clear answer on the page: assistants quote the visible text, so the content itself has to answer the question clearly and directly.

7. How can I correct wrong facts an AI assistant says about my business?

Start with the sources. Assistants repeat what they find, so make sure your own site states the facts clearly on an easy-to-find page, update your Google Business Profile and directory listings, and ask sites with outdated information to correct it. Most assistants also have a feedback button on each answer. Changes take time to appear, because the assistant needs to find and trust the corrected sources first, so check again after a few weeks.

Measure your AI visibility now

The AI Visibility Tracker asks the AI assistants the questions your customers ask and records whether your brand is cited, mentioned, recommended or missing, so you can see where to start and track progress over time.

Run the AI Visibility Tracker โ†’
๐Ÿ’ฌ Got a problem?
Related Guides: AEO Guide  ·  Example Report  ·  Beginner Tutorial  ·  Agent Readiness Guide  ·  Learning Hub

About aiwebpageseo

aiwebpageseo.com is a data-driven SEO and AEO (Answer Engine Optimisation) platform providing a free suite of technical SEO tools. Rather than relying on AI-theorised assumptions, the platform analyses live URL performance, delivering objective diagnostics, page speed metrics, CLS debugging, and site crawl data alongside actionable technical tutorials.