How it works
How we check your site
The score shows how easy one page is for AI tools to reach and read. It does not predict rankings or whether any AI tool will mention you.
Step 1
We read your page
Like an AI bot: the HTML your server sends, without running JavaScript. Plus your robots.txt, sitemap and llms.txt.
Step 2
We weigh the evidence
Each check has an evidence label. Only checks with Strong or Moderate evidence earn points.
Step 3
You get fixes
Problems with the strongest evidence come first, each with its sources.
100
Points to earn
Your score is the points you earn out of 100. Every point comes from a check with Strong or Moderate evidence. Click a block to see the proof behind it.
AI bots are allowed by robots.txt
A bot that follows robots.txt cannot read a page you block. Search bots count toward the score, and so does Claude-User, which Anthropic says follows robots.txt. OpenAI and Perplexity say their bots that fetch a page when a person asks may ignore robots.txt, so those are shown without points. Blocking training bots is your choice and is shown without points too.
Sources, checked 26 Oct 2026
The page loads and is not marked noindex
A page that returns an error, or asks bots not to index it, gives AI tools nothing to read or show.
Sources, checked 26 Oct 2026
The main text is in the HTML, not built by JavaScript
Vercel and MERJ reported in December 2024 that none of the major AI crawlers they studied ran JavaScript, with Google's Gemini as the exception. Text that only appears after JavaScript runs may never be seen.
Sources, checked 26 Oct 2026
The page is a web page and not too large
Very large pages may be cut off before the important text is read. The 2 MB limit is our own choice; AthenaHQ checks page size too but does not publish a number.
Source, checked 26 Oct 2026
A sitemap exists and the page has a canonical link
A sitemap helps bots find your pages, and a canonical link tells them which address is the main copy of a page.
Sources, checked 26 Oct 2026
44%
A short answer appears near the top
A study of 18,012 ChatGPT citations found 44% came from the first 30% of a page. It covers ChatGPT only and shows a pattern, not proof of cause. The 60-word limit for the first paragraph is our own choice; the study measured position, not length.
Source, checked 26 Oct 2026
6.0 vs 3.6
The page shows when it was last updated
In SE Ranking's study of 216,524 pages, pages updated in the last three months averaged 6.0 ChatGPT citations against 3.6 for older pages. Ahrefs also found AI tools cite somewhat newer pages than Google search does. These are company studies of page age; a visible date is how readers and tools can tell when you updated.
Sources, checked 26 Oct 2026
2 studies
Some headings are written as questions
The studies disagree. Indig found questions more often in text ChatGPT cited (18% against 8.9%), but SE Ranking found pages with question headings averaged fewer citations (3.4 against 4.3). Shown without points.
Sources, checked 26 Oct 2026
No data
The page uses lists or tables
Lists and tables are widely recommended, but we found no study that publishes a measured result for them. SE Ranking measured them and did not report a finding. Shown without points.
Source, checked 26 Oct 2026
3.8 vs 4.1
The page has an FAQ section
Evidence is mixed. In SE Ranking's data, pages with an FAQ section averaged 3.8 ChatGPT citations against 4.1 without, though the same article's summary says FAQs help. Shown without points.
Source, checked 26 Oct 2026
3.6 vs 4.2
Organization, Article or FAQ structured data is present
Google says there is no special structured data you need for its AI features. In SE Ranking's data, pages with FAQ structured data averaged 3.6 ChatGPT citations against 4.2 without. It can still help search engines understand a page. Shown without points.
Sources, checked 26 Oct 2026
97%
An llms.txt file exists
Ahrefs found 97% of llms.txt files got no visits at all in May 2026, and SE Ranking found no link between llms.txt and AI citations. Google says you do not need special AI text files. Shown for information, without points.
Sources, checked 25 Oct 2026
No data
The page responds quickly
Google says Core Web Vitals are used by its Search ranking systems, but that page does not mention AI features, and we found no study linking speed to AI answers. Shown without points.
Source, checked 26 Oct 2026
Worth 0 points, and why
The details
Pick a topic. One thing on screen at a time.
Evidence labels
Every check carries one of four labels. Only Strong and Moderate earn points, and together they make up all 100.
The AI companies' own documentation says it, or it follows from how the technology works (a blocked bot cannot read your page).
5 checks
A study of many sites shows it, even if a company that sells related tools ran the study.
2 checks
Only advice or blog posts, with no study showing a measured result.
4 checks
Studies disagree, or found no effect.
2 checks
Checks and sources
Every check, largest first. Click one to see its key number and sources.
AI bots are allowed by robots.txtStrong30pts
Backed by documentation or how the technology works.
A bot that follows robots.txt cannot read a page you block. Search bots count toward the score, and so does Claude-User, which Anthropic says follows robots.txt. OpenAI and Perplexity say their bots that fetch a page when a person asks may ignore robots.txt, so those are shown without points. Blocking training bots is your choice and is shown without points too.
Sources, checked 26 Oct 2026
The main text is in the HTML, not built by JavaScriptStrong20pts
Backed by documentation or how the technology works.
Vercel and MERJ reported in December 2024 that none of the major AI crawlers they studied ran JavaScript, with Google's Gemini as the exception. Text that only appears after JavaScript runs may never be seen.
Sources, checked 26 Oct 2026
The page shows when it was last updatedModerate15pts
6.0 vs 3.6average ChatGPT citations for pages updated in the last 3 months, against older pages
In SE Ranking's study of 216,524 pages, pages updated in the last three months averaged 6.0 ChatGPT citations against 3.6 for older pages. Ahrefs also found AI tools cite somewhat newer pages than Google search does. These are company studies of page age; a visible date is how readers and tools can tell when you updated.
Sources, checked 26 Oct 2026
The page loads and is not marked noindexStrong10pts
Backed by documentation or how the technology works.
A page that returns an error, or asks bots not to index it, gives AI tools nothing to read or show.
Sources, checked 26 Oct 2026
A sitemap exists and the page has a canonical linkStrong10pts
Backed by documentation or how the technology works.
A sitemap helps bots find your pages, and a canonical link tells them which address is the main copy of a page.
Sources, checked 26 Oct 2026
A short answer appears near the topModerate10pts
44%of 18,012 ChatGPT citations came from the first 30% of a page
A study of 18,012 ChatGPT citations found 44% came from the first 30% of a page. It covers ChatGPT only and shows a pattern, not proof of cause. The 60-word limit for the first paragraph is our own choice; the study measured position, not length.
Source, checked 26 Oct 2026
The page is a web page and not too largeStrong5pts
Backed by documentation or how the technology works.
Very large pages may be cut off before the important text is read. The 2 MB limit is our own choice; AthenaHQ checks page size too but does not publish a number.
Source, checked 26 Oct 2026
Some headings are written as questionsDisputed0pts
2 studiesdisagree: one found more questions in cited text (18% against 8.9%), the other found question headings averaged fewer citations (3.4 against 4.3)
The studies disagree. Indig found questions more often in text ChatGPT cited (18% against 8.9%), but SE Ranking found pages with question headings averaged fewer citations (3.4 against 4.3). Shown without points.
Sources, checked 26 Oct 2026
The page uses lists or tablesWeak0pts
No datano study we found publishes a measured result
Lists and tables are widely recommended, but we found no study that publishes a measured result for them. SE Ranking measured them and did not report a finding. Shown without points.
Source, checked 26 Oct 2026
The page has an FAQ sectionWeak0pts
3.8 vs 4.1average ChatGPT citations for pages with an FAQ section, against pages without
Evidence is mixed. In SE Ranking's data, pages with an FAQ section averaged 3.8 ChatGPT citations against 4.1 without, though the same article's summary says FAQs help. Shown without points.
Source, checked 26 Oct 2026
Organization, Article or FAQ structured data is presentWeak0pts
3.6 vs 4.2average ChatGPT citations for pages with FAQ structured data, against pages without
Google says there is no special structured data you need for its AI features. In SE Ranking's data, pages with FAQ structured data averaged 3.6 ChatGPT citations against 4.2 without. It can still help search engines understand a page. Shown without points.
Sources, checked 26 Oct 2026
An llms.txt file existsDisputed0pts
97%of llms.txt files got no visits at all in May 2026, across 137,210 sites
Ahrefs found 97% of llms.txt files got no visits at all in May 2026, and SE Ranking found no link between llms.txt and AI citations. Google says you do not need special AI text files. Shown for information, without points.
Sources, checked 25 Oct 2026
The page responds quicklyWeak0pts
No datano study we found links page speed to AI answers
Google says Core Web Vitals are used by its Search ranking systems, but that page does not mention AI features, and we found no study linking speed to AI answers. Shown without points.
Source, checked 26 Oct 2026
AI bots
We read your robots.txt as each of these bots. Search bots count toward the score, and so does Claude-User, because Anthropic says it follows robots.txt.
Search bots
Build the index AI tools search when they answer.
OAI-SearchBot, PerplexityBot, Claude-SearchBot, Googlebot, Bingbot
Fetch when a person asks
Open your page when someone asks an AI tool about it. OpenAI and Perplexity say theirs may ignore robots.txt, so only Claude-User counts.
ChatGPT-User, Claude-User, Perplexity-User
Training bots
Collect text to train AI models. Blocking them is your choice.
GPTBot, ClaudeBot, Google-Extended
Accuracy
The rules we follow for every check.
- A check counts toward the score only if a study or the AI company's own documentation backs it, and we have read it at the source.
- If we cannot find a source, the check is shown without points, and we say so.
- We never say a change will get you mentioned by an AI tool or ranked higher.
- Every source shows the date we last checked it.
- If you think a result is wrong, tell us. We check it against the source and fix it.