How to Audit Your Site for AI Search (a Free DIY Checklist)

You can audit your site for AI search yourself in about 30 minutes. Run real buyer questions through ChatGPT, Perplexity, Gemini and Google to see if you show up, then work down a six-point checklist: answer-shaped content, structured data, crawler access, entity signals, and freshness. Most small-business sites fail on two or three of those, and the fixes are structural rather than complicated. The checklist below is the same starting point we use when we run an AEO audit for a client.

Quick links

• Do AI search engines already mention you?
• Is your content answer-shaped?
• Do you have the right structured data?
• Can AI crawlers actually reach your content?
• Is your brand a recognised entity?
• Is your content fresh and well-sourced?
• FAQ

Do AI search engines already mention you?

Start by asking the engines directly. This is the fastest way to find out where you stand, and it costs nothing but ten minutes.

More and more buyers now ask an AI assistant before they ever open Google. Google’s own AI Overviews reach over 2 billion monthly users, according to figures CEO Sundar Pichai gave on the company’s July 2025 earnings call. If those answers do not mention you, you are missing a research step that used to be a search result.

So run the test. Open each engine and type the questions a customer would actually ask, not your brand name:

1. ChatGPT, Perplexity and Gemini. Try prompts like “best [your service] in [your town]”, “who should I use for [your service] near me”, and “is [your company] any good”. Note whether you appear, whether a competitor appears instead, and whether the details about you are correct.
2. Google AI Overviews. Run the same questions in a normal Google search and watch for the AI-generated summary box at the top. See if you are named or linked, and which sites are being cited as sources.
3. Write down three outcomes per query: are you mentioned, is the information accurate, and who is being cited if not you.

If you are absent or the engines get your details wrong, that is the gap an answer engine optimisation programme exists to close. Everything below is about why that gap exists and how to fix it.

Is your content answer-shaped?

AI engines pull short, direct answers out of pages. If your content buries the answer, it does not get pulled. Answer-shaped content puts the question and a clear answer up front.

Go through your key pages and ask whether a machine could lift a clean answer from them in one pass. The checklist:

• Real questions as headings. Use the phrasing buyers use. “How much does a new website cost?” beats “Our pricing philosophy.”
• The answer in the first one or two sentences. Lead with it, then explain. Do not make the engine, or the reader, dig.
• An FAQ section on service and product pages, covering the questions you actually get asked. These map almost directly onto how people prompt AI.
• Short paragraphs and plain language. Long, hedged, jargon-heavy writing is hard to extract and harder to trust.
• Specifics over vagueness. Prices, timescales, locations and named services give an engine something concrete to repeat.

This is the single biggest lever for most small sites, and it overlaps neatly with good AI website optimisation generally. If a human skim-reader cannot find the answer in five seconds, an AI engine will not either.

Do you have the right structured data?

Structured data is code that labels what your pages are, in a format machines read directly. It is not essential, but it makes you far easier to understand and quote.

Schema markup clearly defines the entities and relationships on your site, which makes it easier for AI-driven engines to extract your content for direct answers. Google has leaned into this: it added a generative AI section to its structured data documentation, confirming that the same markup that powers rich results also feeds its AI features. Here is what to check:

• Organization schema on your homepage, with your name, logo, and social profiles. This is foundational for being recognised as an entity.
• LocalBusiness schema if you serve a local area, including your address, phone number and opening hours.
• FAQPage schema on pages with a genuine FAQ section.
• Article schema on blog posts, with an author and a published date.

To test what you have, run a few key URLs through Google’s Rich Results Test. It tells you what structured data Google can read on the page and flags errors. If it comes back empty, that is a clear, fixable gap. Most decent platforms add the basics for you, but plenty of sites have none at all.

Can AI crawlers actually reach your content?

If the crawlers cannot read your page, nothing else on this list matters. This is the check most people skip, and the one that quietly sinks the most sites.

Here is the catch that surprises people. Most AI crawlers, including GPTBot, ClaudeBot and PerplexityBot, read the initial HTML and do not execute JavaScript, according to Vercel’s analysis of crawler behaviour. Google has spent years learning to render JavaScript; the AI engines mostly have not. So a site that looks fine to you in a browser can look blank to them.

Run these checks:

1. Look at your robots.txt. Visit yoursite.com/robots.txt and check you are not blocking GPTBot, ClaudeBot, PerplexityBot or Google-Extended. Blocking them, often by accident, makes you invisible to those engines.
2. Test what loads without JavaScript. In Chrome, open DevTools, disable JavaScript, and reload a key page. If the main content disappears, AI crawlers are probably seeing that same emptiness. Sites built as JavaScript-only single-page apps are the usual culprits.
3. Check your content is in the page source. Right-click, “View page source”, and search for a sentence from your page. If it is not in the raw HTML, it is being injected by JavaScript, and the crawlers may miss it.

The fix is server-side rendering, so the content exists in the HTML the moment the page is requested. It is one of the first things we check when we audit a site, because it can wipe out every other effort.

Is your brand a recognised entity?

AI engines trust and repeat brands they recognise. The more consistently your business appears across the web, the more confidently an engine will name you.

This is about signals beyond your own site. The checklist:

• Consistent NAP. Your Name, Address and Phone number should match exactly everywhere they appear: your site, Google Business Profile, directories, social profiles. Mismatches make engines uncertain which “you” is real.
• A claimed, complete Google Business Profile. This is a primary source for local and AI answers.
• Reviews. Volume and recency of genuine reviews feed both trust and the content engines summarise.
• Mentions on third-party sites. Industry directories, press, partner sites and credible blogs all reinforce that your brand exists and does what it says.
• Wikipedia or Wikidata, where you genuinely qualify. Most small businesses will not, and you should not fake it, but if you are notable enough, an entry is a strong entity signal.

You cannot manufacture all of this overnight, but you can fix the easy parts today: claim the profiles, align the details, and ask happy customers for reviews.

Is your content fresh and well-sourced?

AI engines favour content that is current and backed by something. Stale pages and unsupported claims get passed over for sources that look more reliable.

Last checks before you close the laptop:

• Dates on the page. Show a published or updated date on articles and key pages. Refresh the genuinely time-sensitive ones.
• Cite real sources. Link to primary, authoritative references where you make a claim. It signals credibility to engines and readers alike.
• First-hand expertise. Original data, real examples and a named author with a bio all support what Google calls experience and expertise, and AI engines lean on the same cues.
• Prune the dead weight. Thin, outdated or duplicate pages drag down how trustworthy your whole site looks.

That is the full DIY audit. Work through all six and you will have a clear, honest picture of where you stand and a short list of fixes, most of them structural.