State of AI Visibility
The web wasn't built for AI search, and the data shows it. Here is what we found across the public AI readiness checks run through WhoCanFindMe: who's getting found, who's invisible, and why.
Key findings
Most websites are invisible to AI search.
The median AI Readiness score across audited sites is 68/100, in the “Partially visible” band. The average is 61/100. By these measures, the majority of the web has significant AI visibility gaps that are entirely fixable.
68
Median AI Readiness score
out of 100: “Partially visible”
5%
of sites block at least one AI crawler
The single most common critical issue
29%
have FAQ structured data
The strongest on-page schema signal for AI citation
86%
lead with a direct answer
44% of ChatGPT citations come from a page's first third
60%
carry a freshness signal
Date metadata Perplexity uses for recency filtering
36%
are heavily JavaScript-rendered
AI crawlers fetch raw HTML, so JS-heavy pages appear near-blank
Score distribution
Where audited sites cluster
The distribution tells the story clearly: most sites land in the 20–59 range: partially visible at best. Very few have done the work to reach the top tier.
Signal breakdown
Where sites struggle most
Average scores by readiness signal reveal consistent weak spots and where there's the most opportunity to improve.
Top issues
The most common fixable gaps
These are the issues that appear most frequently across audited sites, ranked by prevalence. Each percentage is the share of audited sites where the issue is present.
Percentage = share of audited sites where this issue was found. Data is aggregate and anonymous.
Analysis
What the data means
Crawler blocking is the most impactful single fix. With 5% of sites blocking at least one AI crawler, this is the most common critical issue. OpenAI's two bots do different jobs and are controlled separately in robots.txt: blocking OAI-SearchBot hides you from ChatGPT Search, while blocking GPTBot keeps you out of training data. Plenty of sites want a different answer for each. The fix takes minutes: open your robots.txt and decide bot by bot which Disallow rules to remove or narrow.
Structured data adoption is low. Only 29% of audited sites carry FAQPage schema, the strongest on-page structured data signal for AI citation. Research on 1,508 real estate websites found FAQPage schema present on 6.2% of ChatGPT-visible sites versus only 0.8% of non-visible sites. The gap is real.
Content structure is the underrated lever. Only 86% of audited sites lead with a direct answer, yet 44% of ChatGPT citations come from content in the first third of the page. Answer-first writing isn't just good UX; it's the most extractable signal for AI engines.
Freshness signals are often missing. Only 60% of sites carry a freshness signal (datePublished / dateModified in JSON-LD, sitemap lastmod, or HTML meta date). Perplexity's retrieval pipeline applies aggressive recency filtering: pages without freshness signals are treated as potentially stale.
Data computed from 42 public, completed audits. Last updated: September 14, 2026 at 8:13 PM. Figures update in real time as new public audits are completed. Minimum 20 public audits required before any statistics are shown (privacy floor).
Stay up to date
Get the monthly State of AI Visibility report
We publish updated aggregate findings every month as the dataset grows. Enter your email and we'll send each edition directly to you: one email per month, no spam.
See where your site stands
Run a Free AI Check and get your personalised page readiness score across ChatGPT, Gemini, Perplexity, and Claude in under 30 seconds. One comes with every free account.