These are the exact 22 checks we run on every audit — the signals that decide whether ChatGPT, Google AI Overviews, Perplexity and Claude can read your website, understand who you are, and choose to recommend you. Every one is verifiable yourself, today, for free.
Open yoursite.com/robots.txt and look for GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended under a Disallow rule. Blocking any of them makes you invisible to that engine. This is the single most common — and most fatal — failure we find.
robots.txt can say one thing while your firewall does another. Fetch your homepage with an AI user agent (curl -A "GPTBot" yoursite.com) — if you get a 403 or a challenge page while browsers get your site, a CDN/WAF setting (often Cloudflare's "Block AI bots" toggle) is turning engines away silently.
A noindex meta tag or X-Robots-Tag header removes you from search and AI answers alike. They're often left behind by developers after a site launch.
/sitemap.xml should exist and list your real pages — it's how crawlers discover new content quickly.
An emerging standard: a Markdown file at /llms.txt describing what your business does and which pages matter. Cheap to add and signals AI-readiness even where support is still uneven.
Engines treat non-HTTPS sites as low-trust sources. Table stakes.
View source and search for application/ld+json. Structured data is how engines read facts about you without guessing.
The strongest way to control how AI describes you: name, address, phone, geo, services, and sameAs links to your profiles. For professional firms, use the specific type (LegalService, Dentist, AccountingService…).
AI answers are question-shaped. FAQPage schema maps your content directly onto the format engines quote from — the highest-leverage schema type for citations.
BreadcrumbList markup helps engines understand your site's structure and topic hierarchy.
Most AI crawlers read raw HTML and don't execute JavaScript. If your content only appears after scripts run, it doesn't exist to them. Disable JS in your browser and reload — what's left is what AI sees.
Your H1 should say, in one line, what you do and where. Not a slogan.
Engines lift answers from well-labeled H2/H3 sections. Walls of text don't get quoted.
Pages that literally contain the questions people ask AI — with direct 2–3 sentence answers underneath — are the ones that get cited. An FAQ page is the minimum.
The first two sentences every engine reads about you. 30–60 characters for the title, 70–160 for the description.
Visible current-year dates on content. Engines strongly prefer sources that look maintained.
Thin pages give engines nothing to work with. Cited sources average 1,000+ readable words on key pages.
A blog or insights section that answers real client questions ("How much does X cost in [city]?") is how businesses get named in AI responses — and how they appear in the "Best X in City" roundups AI assistants lean on.
Real names, credentials and photos are what separate a citable authority from an anonymous site. Engines score expertise and trust (E-E-A-T).
Local AI answers pull contact details from consistent on-page NAP (name–address–phone) data.
An address anchors your business to a location — essential for every "near me" answer.
Canonical tags stop your authority splitting across duplicate URLs; OG tags give engines a second, machine-readable description of every page.
We'll run the full audit on your site — scored 0–100, with side-by-side scores against two competitors — free, within 24 hours.
Get your free AI visibility audit