74/100
The median B2B site scores 74/100 for AI readiness; entry to the top decile starts at 85.
Scored across crawl access, structured data, extractability, answerability, entity clarity, trust, freshness, and off-site presence.
n=769 sites ·
Benchmark report
The State of AI Visibility
We audited 769 B2B websites for how visible, structured, and citable they are to AI answer engines. The median scores 74/100. Here is where the average site wins, where it falls short, and what it means for being cited by ChatGPT, Perplexity, and Google AI Overviews.
Live data, refreshed weekly · · Methodology
769
B2B sites analyzed
74/100
Median AI-readiness score
44–91
Score range
85/100
Top-decile threshold (p90)
Quotable, anchored, datestamped
Key findings
Every figure below is computed from the live dataset. Each card carries its own anchor and sample size, and the copy button gives you a ready-to-paste citation.
49/100
Content Freshness & Authority is the weakest layer: the average site scores 49/100, and a quarter of sites score 25 or below.
This is the layer that most directly decides whether an AI engine can parse and cite you correctly.
n=769 sites ·
66–81
Half of all audited sites land between 66 and 81: the middle of the field is mediocre together.
Clearing the pack does not require excellence; it requires fixing what most sites leave broken.
n=769 sites ·
99/100
Bot Access & Control Plane is effectively solved, averaging 99/100. Being fetchable no longer differentiates; being citable does.
Differentiation has moved down the stack, from access to structure and evidence.
n=769 sites ·
+35
The top 10% pull away hardest on content freshness & authority: 35 points between the median site (50) and the 90th percentile (85).
If you want to know what the best sites do differently, start here.
n=769 sites ·
91/100
No audited site scores 100. The best in the sample reaches 91; every site still fails something.
AI readiness is a maintained property, not a finished project.
n=769 sites ·
Overall score distribution
The shape of the field
The middle half of the field sits in the 66–81 band around a median of 74. The tail above 85 is where AI engines find sites they can parse, verify, and cite without guesswork.
By audit dimension, weakest first
Where sites win and lose
Most sites have the basics covered: bot access & control plane averages 99/100. The gap is in the machine-readable layer: content freshness & authority averages just 49/100. Each bar shows the middle half of the field (band), the median (tick), and the 90th percentile (dot).
Content Freshness & Authority
avg 49
p25 25 · median 50 · p75 63 · p90 85
Datestamps, bylines, and article schema. Engines discount undated, unattributed content, and half the field shows almost no freshness signals at all. This is also where the top decile separates hardest.
Structured Data
avg 49
The machine-readable layer. JSON-LD tells an engine who you are, what you sell, and which page answers what. The bottom quartile ships essentially none of it, which makes correct citation a coin flip.
Entity Clarity
avg 66
p25 63 · median 63 · p75 75 · p90 76
Whether a machine can tell who you are: organization schema, the brand name in title and H1, linked profiles. Ambiguity here is how engines mix you up with a competitor.
Trust & Security
avg 73
About, contact, privacy and terms pages, security headers, no exposed secrets. Engines weigh accountability signals when deciding what is safe to recommend.
Off-site Presence & Mentions
avg 79
p25 83 · median 90 · p75 92 · p90 100
What the rest of the web says about you: third-party mentions, source diversity, authority, recency. The hardest dimension to fake, and a large separator at the top of the field.
Content Answerability
avg 80
p25 75 · median 83 · p75 88 · p90 88
Question-shaped headings, definitions, lists, and concrete data points an engine can lift verbatim. Decent on average; the gap between adequate and quotable is where citations are won.
HTML Extractability & Main Content Clarity
avg 84
Clean titles, a single H1, sane text-to-markup ratio, alt text. Mostly competent across the field; failures here are self-inflicted and cheap to fix.
Fetch, Render, and URL Integrity
avg 90
p25 70 · median 100 · p75 100 · p90 100
HTTPS, fast responses, no redirect chains, and content that exists without running JavaScript. The median site passes outright; the bottom quartile pays a steep tax, often for JS-only rendering.
Bot Access & Control Plane
avg 99
p25 100 · median 100 · p75 100 · p90 100
robots.txt, sitemaps, and AI-crawler policy. Effectively solved: nearly everyone lets the engines in. Letting them in is not the same as giving them something to cite.
Gap between the median site and the 90th percentile
What separates the top decile
Where the spread between the median and the 90th percentile is widest, the best sites are doing something the rest are not. Where it is narrow, the dimension is either solved or uniformly neglected.
| Dimension | Median | p90 | Gap |
|---|---|---|---|
| Content Freshness & Authority | 50 | 85 | +35 |
| Entity Clarity | 63 | 76 | +13 |
| Off-site Presence & Mentions | 90 | 100 | +10 |
| Content Answerability | 83 | 88 | +5 |
| Fetch, Render, and URL Integrity | 100 | 100 | +0 |
| Bot Access & Control Plane | 100 | 100 | +0 |
How the data is collected
Methodology
Figures aggregate automated AI-readiness audits of 769 public B2B websites, scored 0–100 across nine dimensions covering crawl access, structured data, extractability, answerability, entity clarity, trust, freshness, and off-site presence. Each domain contributes its most recent completed audit inside a 365-day window. The sample is self-selected: these are sites whose teams chose to run an audit, which likely skews it toward the AI-aware end of the market. Data as of .
The full scoring rubric, including every check, weight, and known limitation, is public: how the audit scores sites. The live, interactive view and the per-brand leaderboard live in the audit app: app.nyman.media/insights.