Most websites have never been audited for AI visibility. They were built for Google — optimized for keywords, backlinks, and Core Web Vitals — and they're blind to whether ChatGPT, Claude, Perplexity, or Gemini can actually find and cite them.
This is the 14-point checklist we use to audit any website end to end. Each check has a weight (its contribution to the 0-100 score), a reason it matters, and the concrete fix. Use it manually, or use a tool like Teehoo Martech's free GEO Audit to run the whole thing in under 5 minutes.
Each check earns up to its weight in points. Miss a check entirely and you earn 0; partial credit for partial implementation. Final score = sum of all 14 weights, normalized to 100.
| Grade | Score range | What it means |
|---|---|---|
| A | 85-100 | Category leader. AI consistently cites you. |
| B | 70-84 | Well-positioned. Some engines cite you; optimization will close gaps. |
| C | 50-69 | Middle of the pack. Technical fundamentals in place but inconsistent. |
| D | 30-49 | Below baseline. Significant remediation required. |
| F | 0-29 | Effectively invisible. Most of the 14 checks fail. |
Why: AI retrieval pipelines use Schema to identify who/what a page is about. Without it, AI has to infer from unstructured HTML. The single highest-impact check.
Pass criteria: Valid JSON-LD for at least Organization (or LocalBusiness for local), passes Google's Rich Results Test, includes name, url, logo, sameAs.
Why: If GPTBot, ClaudeBot, or PerplexityBot is disallowed, the respective engine can't index your content — period.
Pass criteria: robots.txt explicitly allows (or doesn't disallow) GPTBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot.
User-agent: GPTBot\nAllow: / (and equivalents for other AI crawlers) to robots.txt. Remove any inherited blocks from old plugins or WordPress defaults.Why: AI loves direct Q&A format. FAQPage-schema pages become the literal cited source for category queries.
Pass criteria: A dedicated FAQ page (or FAQ section on key pages) with at least 6 Q&A pairs, wrapped in FAQPage JSON-LD.
/faq page. Pull questions from support tickets, sales calls, and Google's "People Also Ask." Add FAQPage schema. Answers should be 50-200 words each — not one-liners, not essays.Why: Brands mentioned on 4+ platforms (Reddit, Wikipedia, review sites, directories, major publishers) get cited at 2.8x the rate of single-platform brands.
Pass criteria: Brand appears on 4+ distinct external platforms beyond its own site and Google Business Profile.
Why: The emerging "readme for AI" standard. Tells crawlers who you are, what you do, and which URLs matter most.
Pass criteria: /llms.txt exists, is <500 lines, includes positioning + key URL list.
Why: First signals AI reads when browsing your page. Weak meta = weak first impression in the model's working memory.
Pass criteria: Homepage has unique <title> (50-60 chars), <meta description> (140-160 chars), and og:title, og:description, og:type, og:image.
Why: AI summarization uses titles as the primary label for a page's topic. Generic titles ("Home", "About Us") signal nothing.
Pass criteria: Top 10 pages (by traffic) have unique, descriptive, keyword-aligned titles.
Why: Thin content rarely gets cited. AI biases toward comprehensive pages (2,000+ words) over listicle-style posts.
Pass criteria: Top 5 category-intent pages average >1,500 words and use H2/H3 hierarchy.
Why: AI heavily weights review-aggregation sources (Amazon, Yelp, G2, Trustpilot). Fresh 4.5+ ratings outperform stale ratings at any volume.
Pass criteria: At least one major review platform with 20+ reviews, 4.5+ average, most recent within 60 days.
Why: Clear hierarchy helps AI understand what the site is about. A dozen orphan pages dilutes topic signal.
Pass criteria: Every important page is within 3 clicks of the homepage. Related pages link to each other. No orphan pages in the top 20 by traffic.
Why: AI biases toward recently-updated content for time-sensitive queries. A 2024 "best X" article loses to a 2026 one.
Pass criteria: Top 10 content pages have an "Updated [month year]" timestamp within the last 12 months.
Why: AI crawlers use mobile user agents for many retrievals. Slow or broken mobile = missed citations.
Pass criteria: Mobile LCP < 2.5s, no layout shifts, all content accessible (no JS-gated content that fails without interaction).
Why: Duplicate-URL situations confuse AI about which version to cite. Canonical tags disambiguate.
Pass criteria: Every page has a <link rel="canonical"> pointing to the authoritative version.
Why: Helps AI crawlers discover all indexable URLs. Not a huge signal on its own but a baseline sanity check.
Pass criteria: Sitemap exists at /sitemap.xml, includes all indexable pages, referenced in robots.txt.
Sitemap: https://yoursite.com/sitemap.xml. Check it includes your 20 most important pages.For each site, work top to bottom:
yoursite.com/robots.txt. Search for GPTBot. Then view page source and grep for "@type". Score checks 1, 2, 14./llms.txt and /sitemap.xml. Score checks 5, 14.Sum the points. That's your AEO score. Grade it against the table above.
If you score below 50, the order of operations matters. Technical fixes first (they're one-time and high-leverage), content fixes second (they compound), platform-presence third (it takes months to build).
| Phase | Weeks | Checks tackled | Expected lift |
|---|---|---|---|
| 1 — Technical foundation | Weeks 1-2 | #1, #2, #5, #13, #14 | +25-35 points |
| 2 — Content structure | Weeks 3-6 | #3, #6, #7, #8, #10 | +15-25 points |
| 3 — Authority & presence | Weeks 7-12 | #4, #9, #11 | +10-20 points |
| 4 — Ongoing | Monthly+ | #11, #9, re-audit | +5/quarter |
Most brands go from Grade F to Grade B in 6-8 weeks with this sequence. Grade A takes 6+ months because authority signals (Wikipedia, genuine Reddit presence, authoritative citations) can't be fast-tracked ethically.
Our own website went from 49/100 to 94/100 in 30 minutes with 5 technical fixes (read the case study). Technical debt on AEO is easier to pay off than SEO debt — the wins are concrete, the fixes are bounded.
Automated scan across all 14 checks. Scored 0-100 with remediation steps. Takes 3 minutes.
Run GEO Audit or Scan your AI visibilityRelated reading
· How to Rank in ChatGPT: 12 Ranking Factors
· GEO for Agencies: The Complete Playbook
· AI Share of Voice Tracking: Complete Guide
· Case Study: 49/100 to 94/100 in 30 Minutes
About Teehoo Martech
AI visibility platform for brands and agencies. Free 14-point AEO audit, multi-engine SOV tracking, white-label reporting. Learn more.