How the AI Visibility Score Is Calculated

The score is a single number from 0 to 100 with a letter grade. It exists to tell you whether something needs attention, and it is deliberately not the most important thing on the page. A score can be high while a specific crawler you care about is still blocked.

Read the tiers, not just the total

A total can hide the thing that matters. Blocking a handful of low-tier scrapers barely moves the number and barely matters. Blocking one Tier 1 retrieval crawler can remove you from an AI answer entirely. Always look at Tier 1 before you celebrate a grade.

Why bots are tiered

Not every crawler carries the same consequence, so the database sorts all 265 of them into 6 tiers. Tier 1 is the set whose absence you would actually notice: the crawlers that fetch a page so an assistant can cite it, and the major search engines. The lowest tiers are scrapers with no upside for you at all.

This is also why two sites with the same score can be in very different situations. The tier weighting means the score reflects consequence, not a raw count of how many bots you happen to allow.

Training crawlers versus retrieval crawlers

This is the distinction that matters most, and it is the one most often missed.

Training

Collects text to train a future model. Blocking it is a reasonable editorial or licensing decision. It does not remove you from search results and it does not stop you being cited today.

Retrieval

Fetches your page at the moment someone asks a question, so the assistant can quote and link you. Block this and you become invisible in AI answers even though your SEO is untouched.

The common mistake

Copying a "block all AI bots" snippet from a forum. Those lists usually mix both kinds together, so people who only meant to opt out of model training also silently remove themselves from being cited. The report separates them so you can make the choice on purpose.

The GEO side of the report

Access is necessary but not sufficient. A crawler can reach a page and still fail to understand it well enough to quote. The 42 GEO checks cover that second half, grouped into these areas:

AI Crawler Access
Technical SEO
Content Structure
Structured Data
Trust & Authority

A WAF can make results look worse than reality

If a firewall blocks our request for your robots.txt, that is a fact about your firewall, not proof that crawlers are blocked. The report says so explicitly instead of reporting a confident wrong answer. If you see a WAF notice, verify by opening your robots.txt in a normal browser.

The score, worked through

How the 100 points add up

Score components
Component	Points	What earns them
Crawler access	70	Tier 1 crawlers allowed (partial access earns most of the credit)
AI infrastructure	15	llms.txt (10) and llms-full.txt (5)
Technical readiness	15	Sitemap (6), structured data (6), HTTPS (3)

Three example sites

Example scores
Site	Situation	What to do
Open shop	All crawlers allowed, sitemap and HTTPS, no schema or llms.txt	Add structured data; the access points are already earned
Cautious publisher	Training crawlers blocked, retrieval allowed	Nothing urgent; the lower access score reflects a deliberate choice
Accidental block	User-agent: * with Disallow: /	Fix immediately; almost every crawler is refused

How common each pattern is

In our robots.txt study of the Tranco top 10,000 (5,583 valid files), 20.1% of sites block at least one training crawler and 12.3% block at least one AI search crawler. Far fewer, 4.3%, refuse everything with User-agent: * and Disallow: /.

Sharing your score

If you share a score with a client or a manager, share the date and the URL with it, and say what the score covers: whether crawlers can read the page and whether the basic GEO files exist. A score is a starting point for a conversation about what to fix, not a ranking, and it reads best next to the access table that explains it.

Comparing your score over time

Compare scores for the same URL only, checked the same way. A home page and a blog article on one site can score differently, because firewall rules and structured data differ by template. Save each result with its date, and when the score moves, compare the access table first: a change there explains most jumps.

What the score cannot tell you

How often a crawler visits. That is in your server logs.
Whether an AI engine will cite you. That depends on content and authority.
What a crawler does with the page after fetching it.

Grades and what they mean

Grades
Grade	What it usually means
A	Every important AI crawler can read the page and the GEO basics are in place
B	Access is fine; one or two GEO items are missing
C	At least one important crawler is refused, or several GEO items are missing
D or F	Several important crawlers are refused; fix access before anything else

Raising a low score, in order

Unblock any Tier 1 AI crawler that is refused by mistake.
Fix server errors that stop crawlers reading robots.txt or the page.
Publish a sitemap and reference it in robots.txt.
Add an llms.txt with the generator.
Add structured data that describes the page, starting with Organization and the main content type.

Where the points come from

Score components
Component	Points
Bot access (can each AI crawler reach the page?)	70
AI infrastructure (llms.txt and related files)	15
Technical readiness (sitemap, schema, HTTPS)	15
of which llms.txt	10
of which sitemap	6
of which structured data	6
of which HTTPS	3

Why access carries most of the weight

A crawler that cannot fetch a page cannot cite it, however good the page is. That is why access is 70 of the points and everything else shares the rest. A site with perfect structured data and a blocked retrieval crawler will score lower than a plain site that every crawler can read, and that ranking matches what happens in AI answers.

Scores that surprise people

A high score but no AI citations: The score measures whether you can be read, not whether you are the best answer. Content and authority decide citations; see the guide to how engines choose sources.
A low score on a site with no robots.txt: No file means everything is allowed, so access is usually fine. Look at the firewall and the GEO items instead.
A score that changed with no edits: A CDN setting, a plugin update or a new crawler in our database can each move it. Compare the two results line by line.

What one point is worth

Do not chase single points. Moving from 5 to 6 sitemap points changes nothing a crawler will notice, while unblocking one Tier 1 crawler can change whether you appear in an AI answer at all. Our database has 14 Tier 1 crawlers; fix those first, then work down.

Test yourself

Can a site with every crawler allowed still score below 100?

Yes. Points for llms.txt, structured data, a sitemap and HTTPS are separate from access.

Common questions

What is a good AI visibility score?

Anything in the 80s or above usually means nothing is structurally wrong. But a score is a summary: check the Tier 1 row directly, because a high score can still coexist with one important retrieval crawler being blocked.

Why is my score low when my SEO is fine?

Traditional SEO and AI visibility are different problems. A site can rank well in Google and still be unreadable to assistants, most often because a broad robots.txt rule blocks retrieval crawlers, or because the page carries no structured data and no clear answer-shaped content.

Will blocking AI training bots hurt my Google rankings?

No. Training crawlers such as GPTBot and Google-Extended are separate from Googlebot. Blocking them has no effect on search rankings. Blocking Googlebot or Bingbot is what removes you from search results.

Does the score change if I block scrapers?

Only slightly, by design. Low-tier scrapers offer you little benefit, so allowing or blocking them barely moves the number. The weighting deliberately concentrates on crawlers whose absence has real consequences.

My report shows a firewall block. Is my site misconfigured?

Not necessarily. It means a WAF such as Cloudflare answered our request with a challenge instead of your file. Crawlers with more established reputations may not be treated the same way. Open your robots.txt in a browser to see the real file, and if you want crawlers treated predictably, allowlist the ones you care about at the firewall.

Ready to check your site?

265 crawlers, 42 GEO checks, no sign-up.

Run a free check