Can ChatGPT, Perplexity and Google AI cite your site?
Our free AEO audit checks any page the way answer engines read it: who's allowed to crawl it, how cleanly it splits into answers, whether machines know who wrote it, and whether it says anything worth quoting. You get a score, a fix for every issue, and the citable links that fit your niche.
Check whether answer engines can cite your page
What is an AEO audit?
An AEO (answer engine optimization) audit measures how likely a page is to be retrieved, understood and cited by AI answer engines. A classic SEO audit asks whether a page can rank. An AEO audit asks a narrower question: when someone asks an assistant about your topic, can it reach your page, pull a self-contained passage out of it, trust where that passage came from, and attribute it to you by name?
This tool fetches the page once as InTheAnswerBot, along with your robots.txt, llms.txt and sitemap.xml, and runs up to 28 weighted checks against the raw HTML. That matters: most AI crawlers don't run JavaScript, so copy that only appears after client-side rendering is invisible to them, and to this audit.
How answer engines choose their sources
Implementations differ, but ChatGPT search, Perplexity, Claude, Gemini and Google AI Overviews follow the same broad retrieval-augmented pattern. A page can drop out at any of these four steps.
- 1
Crawl and index
Each engine keeps its own index, or borrows one. ChatGPT search relies on OAI-SearchBot, Perplexity on PerplexityBot, Claude on Claude-SearchBot, and Google AI Overviews on the regular Googlebot index. If robots.txt, a firewall or client-side rendering hides your copy, the rest of the pipeline never sees it.
- 2
Fan out and retrieve
A single prompt is rewritten into several sub-queries. For each one the engine pulls a handful of candidate pages, then splits them into passages. Pages that state one idea per section, under a descriptive heading, produce cleaner passages.
- 3
Rank for trust and fit
Passages are scored on how directly they answer the sub-query, how authoritative and consistent the source looks, and how fresh it is. Structured data, named authors and dates, and corroboration from other trusted sites all feed this step.
- 4
Synthesize and cite
The model writes the answer and attributes claims to a few sources, usually the ones with the most specific, quotable statement of the fact: a number, a definition, a comparison. That's the passage you want to own.
What each check category means
Every check is weighted. A pass earns full credit, a warning half, a fail none, and the weighted total becomes your 0–100 score. The percentages are each category's share of the total.
AI crawler access
Whether OAI-SearchBot, ChatGPT-User, Claude-SearchBot, PerplexityBot, Googlebot and Bingbot may fetch the exact URL under your robots.txt, how training crawlers (GPTBot, ClaudeBot, Google-Extended, Applebot-Extended, CCBot) are treated, plus noindex/nosnippet directives, HTTPS, redirects, llms.txt and your XML sitemap.
Content structure
Title and meta description length, a single H1, a real H2/H3 outline, question-style headings, FAQ blocks, lists and tables, enough main-content words in the raw HTML, and sentence length. Together these decide how cleanly a page splits into quotable passages.
Entity & trust
JSON-LD types (Organization, WebSite, Article, FAQPage, Product, HowTo, BreadcrumbList, Person), Organization sameAs links, author bylines, published and modified dates, canonical URL, html lang, Open Graph tags and links to About and Contact pages: the signals that tell a machine who stands behind the content.
Citability
Concrete numbers and statistics, outbound links to research, government and major-news sources, your brand named in the title or H1, and an answer-first opening paragraph that an engine can lift as-is.
How citable links help
A perfect on-page score makes you quotable, but answer engines rarely take one site's word for anything. They cross-check claims against sources they already retrieve often: established news outlets, industry publications and reference sites. When those sources mention your brand alongside the topic, the engine has independent corroboration, and you become a much safer name to put in the answer.
That's what a citable link is: a placement on a site that answer engines actually retrieve, in a vertical they cite for your topic. InTheAnswer scores every placement on authority, real traffic, topical citability and link hygiene, so you can spend on the mentions most likely to show up in AI answers rather than on Domain Rating alone.
AEO audit questions, answered
What is the difference between SEO and AEO?
SEO earns a position in a list of links; AEO (answer engine optimization) earns a mention or citation inside the answer that ChatGPT, Perplexity, Claude, Gemini or Google AI Overviews generate. They share foundations like crawlability, authority and good content, but AEO puts more weight on passage-level clarity, machine-readable entity data, and corroboration from other trusted sites that mention your brand.
Should I block GPTBot and other AI crawlers?
Separate the two kinds of bot. Search and answer crawlers such as OAI-SearchBot, Claude-SearchBot and PerplexityBot fetch pages so those engines can cite them; blocking them removes you from their answers. Training crawlers such as GPTBot, ClaudeBot, Google-Extended and CCBot collect data for future models. Blocking those is a legitimate business choice, and you can allow the search bots while opting out of training.
Do I need an llms.txt file?
It helps but it isn't required. llms.txt is a proposed convention: a Markdown file at your site root that summarizes the site and links its most important pages for language models. Support across engines is still uneven, so treat it as cheap insurance on top of what matters more: crawler access, server-rendered HTML and clear page structure.
How long until fixes show up in AI answers?
Engines that search the live web, like ChatGPT search, Perplexity and Google AI Overviews, can reflect on-page changes once they recrawl the page, which typically takes days to a few weeks. What a model knows without searching changes only when it is retrained, which takes months. Off-site mentions build up over a similar period of weeks to months as new placements get indexed.
Do backlinks still matter for AI search?
Yes, though for a broader reason than PageRank. Answer engines prefer sources they already trust and look for agreement across several of them. A brand that is mentioned and linked on publications those engines retrieve is more likely to be named in the answer. That is why InTheAnswer scores every placement for AEO impact, not just Domain Rating.