You don't need an agency to find out whether AI answer engines can see your website. Most of what a professional Answer Engine Optimization (AEO) audit checks first is visible from your own browser and a terminal, and you can work through it in an afternoon. This checklist runs 15 checks in four passes: whether AI crawlers can reach your content, whether engines can understand it, whether the supporting signals are in place, and what ChatGPT, Perplexity, and friends actually say about you today. Mark each item pass or fail and you'll finish with a concrete punch list instead of a vague worry.
How does the DIY audit work?
Work through the four passes in order: access first (a blocked crawler makes every other fix irrelevant), then content, then technical signals, then verification. Each check takes five to twenty minutes and needs only a browser, a free validator, and a terminal for one command.
Pass 1: Can AI crawlers reach your site?
The most common AEO failure is also the simplest: the bots never get in. Four checks.
1. Read your robots.txt
Open yourdomain.com/robots.txt and look for the AI crawler names: GPTBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot, Google-Extended, and Bingbot. A Disallow: / under any of them means that engine can't fetch your pages. Old "block all bots" rules and security plugins are the usual culprits — our guide to AI crawlers and robots.txt covers who's who and what to allow.
2. Check your CDN and firewall settings
A clean robots.txt means nothing if your CDN blocks the request at the edge. Log in to Cloudflare (or your host's security panel) and look for toggles like "Block AI bots," "Bot Fight Mode," or firewall rules matching bot user agents. Some ship enabled by default and silently turn away AI crawlers while human visitors see nothing wrong.
3. Fetch a page as GPTBot
From a terminal, run curl with GPTBot's user agent: curl -I -A 'GPTBot' https://yourdomain.com/. A 200 status means the crawler gets through; a 403, a 503, or a challenge page means something upstream is blocking it. Repeat with 'ClaudeBot' and 'PerplexityBot'. This single command catches edge-level blocks no dashboard shows.
4. View source and find your content
Right-click a key page, choose View Page Source, and search for a sentence from your main content. If it's missing from the raw HTML, your site paints content with JavaScript after load — and most AI crawlers don't execute JavaScript, so they see an empty shell. Server-side rendering or pre-rendering is the fix.
Pass 2: Will engines understand and quote your content?
Access gets you crawled; structure and clarity get you cited. Five checks.
5. Validate your schema markup
Schema markup is structured data — usually JSON-LD in your page source — that tells machines exactly what a page is about. Paste your key URLs into the free validator at validator.schema.org and confirm each important page carries the right type: Organization, LocalBusiness, Product, FAQPage, or Article, with no errors and no fields that contradict the visible page.
6. Sweep for entity consistency
AI engines assemble a picture of your business from every mention they can find. Check that your business name, address, phone, and one-line description are word-for-word identical across your site, Google Business Profile, LinkedIn, and major directories. A name that varies — 'Acme Dental' here, 'Acme Dental Group LLC' there — splits your identity into weaker fragments. Full method in our entity consistency guide.
7. Audit your headings for questions
Answer engines match questions to answers. Scan your key pages: how many section headings read like something a customer would actually ask? Then check whether the first sentence under each heading answers it directly, or makes the reader wade through three paragraphs of wind-up before the point.
8. Check your FAQ coverage
List the ten questions customers actually ask you — by phone, by email, in sales calls — and search your site for each one. Every unanswered question is an answer an engine will source from a competitor instead. Each answer should stand alone: two to four sentences that make sense with no surrounding context, because that's how engines quote.
9. Check your dates
Engines favor content they can date and distrust staleness. Confirm your posts and key pages show a visible published or updated date, that any dateModified in your schema matches reality, and that no page still leads with last year in its title.
Pass 3: Are the supporting signals in place?
Three quick technical checks.
10. Add or verify llms.txt
llms.txt is a proposed plain-text file at your site root that gives AI systems a curated map of your most important pages — the spec lives at llmstxt.org. Adoption by the engines is still uneven, but it costs about twenty minutes. Check whether yourdomain.com/llms.txt exists and actually lists your money pages.
11. Verify your sitemap
Open yourdomain.com/sitemap.xml. It should exist, be referenced from robots.txt, include every page you care about, and carry lastmod dates that genuinely change when pages do. Crawlers with limited budgets prioritize what the sitemap says is fresh.
12. Sanity-check titles, descriptions, and canonicals
Confirm each key page has a unique title and meta description matching its actual content, and that your site resolves to one canonical version — not www and non-www serving duplicate copies. Contradictory metadata muddies what an engine believes your pages, and your brand, are about.
Pass 4: What do the engines actually say about you?
The final three checks measure outcomes, not inputs.
13. Ask the engines about your business
Ask ChatGPT, Claude, Perplexity, and Gemini: 'What does [your business] do?' Note anything wrong, outdated, or missing — invented services, an old address, a competitor described in your place. These errors usually trace back to weak or contradictory public information.
14. Ask your money question
Now ask the question a ready-to-buy customer would ask — 'best [your service] in [your city]' or 'how do I fix [the problem you solve]' — and see who gets cited. Open the cited pages and note what they do that yours don't: direct answers, question headings, visible dates, specific numbers.
15. Check your analytics for AI referrals
In GA4, look for referral traffic from chatgpt.com, perplexity.ai, copilot.microsoft.com, and gemini.google.com. Even small numbers matter, because these visitors arrive pre-sold by a recommendation. Our guide to tracking AI referral traffic in GA4 shows the setup. Zero referrals plus failures in the earlier passes is your baseline — and your motivation.
What does a professional audit add beyond DIY?
The DIY checklist finds the obvious blockers; a professional audit finds the subtle ones and prices the fixes. Beyond the fifteen checks above, our audit tests rendering with real crawler user agents from multiple networks, checks server logs (where available) for actual bot hits, validates schema across every page template rather than spot-checked URLs, benchmarks your citation share against named competitors across several engines, and turns all of it into a prioritized fix list with effort estimates. The audit itself is free; if you want help executing, remediation runs $99/hour and ongoing citation monitoring is $19/month — but the report is yours either way.
Frequently asked questions
How long does the DIY audit take?
Plan for three to four hours. Pass 1 takes about thirty minutes; the content pass is the longest because you're reading your own pages critically.
What's the most common failure this checklist finds?
Crawler blocking at the CDN or firewall level — check 2 and check 3. It's invisible in normal browsing, often enabled by a well-meaning security setting, and it single-handedly removes you from every engine whose bot it blocks.
Do I need technical skills to run it?
One curl command is as technical as it gets, and you can copy it verbatim. Everything else is browser work: reading files at known URLs, viewing page source, pasting URLs into a validator, and asking chatbots questions.
How often should I re-run the checklist?
Quarterly is a sensible rhythm, plus an immediate re-run after any redesign, replatform, CDN change, or new security plugin — those are the moments blocks and rendering regressions creep in.
If your afternoon turned up more failed checks than you expected — or you'd rather get the complete picture in one pass — request a free professional AEO audit and we'll run the full version against your site, including the checks that need crawler-level tooling.