METRIC VAULT
100% Free · No signup

Free AI Citation
Audit

AI answers name their sources. This audit shows whether ChatGPT, Perplexity, Gemini and Claude will pick your page as one of them, with a 0-100 score and the exact fixes to earn the citation.

No signup · No credit card · Tests the live page, robots.txt and llms.txt
AI CITATION AUDIT

AI reads your page. Does it cite you?

Paste your domain and get a 0-100 AI-readiness score across the six signals ChatGPT, Perplexity, Gemini and Claude use to pick a source, plus the exact fixes that earn the citation.

Reads your live page, robots.txt and llms.txt. No signup, no cached guesses.
ai-citation-audit · example
PARTLY READY
AI readiness
0
out of 100
READ-READY
Crawlers
Allowed
GPTBot, ClaudeBot, PerplexityBot
GAP
Citable facts
0
found, 5 or more needed
GAP
Author and date
Missing
the trust tie-breaker
The problem

AI can crawl you and quote a rival.

Being fetched is not being chosen. An answer engine can read a hundred pages and name just one as its source. This audit checks the signals that decide which page gets the citation, so you stop being the site that was read and skipped.

Crawlable but invisibleYour page loads fine for GPTBot, yet the answer names another site. Reachable is only the first hurdle.
Facts too vague to quoteEngines cite a concrete number or date they can attribute. A page of adjectives gets read and passed over.
No author, no trustWhen two pages say the same thing, the one with a named author and a visible date wins the citation.
How it works

One domain in. A score and a fix list out.

Enter your domain

No login and no crawl setup. Paste a URL and run the check.

We read the AI signals

We fetch your live page, robots.txt and llms.txt, then score the six signals answer engines use to read, understand and cite a page.

Get your score and fixes

A 0-100 readiness score, a pass or fail on every signal with its point weight, and a ranked list of what to fix first.

Glossary

What these words mean

AI citation

When an answer engine names or links your page as the source behind part of its answer. It is the AI-era version of a backlink.

Perplexity listing your link as a source

AI answer engine

A tool that answers a question directly instead of showing ten blue links. ChatGPT, Perplexity, Gemini and Claude are the big four.

The answer, not the search page

AI crawler access

Whether bots like GPTBot, ClaudeBot and PerplexityBot are allowed to fetch your page in robots.txt. No answer can cite a page it was never allowed to read.

One Disallow line can hide you

Structured data

Schema markup in JSON-LD that labels your articles, FAQs and author so an engine knows exactly what your facts are and can lift them with a credit.

Article, FAQPage, Author

Citable fact

A concrete number, date or named specific an answer can quote and attribute to you. Precise beats vague every time.

"cut load time 42%" beats "fast"

Trust signals

A named author, a visible publish or updated date, and evidence of expertise. They are the tie-breaker when two pages say the same thing.

A byline plus "Updated 2026"
The readiness score

One number for "will AI cite this?"

The six signals roll up, weighted, into a single 0-100 score with a color band, so you know at a glance whether engines can cite this page or will skip it.

  • Weighted 0-100 from the six signals engines actually use
  • A color band: green is cite-ready, amber is partly ready, red is at risk
  • The exact number a re-run has to beat after you ship a fix

Rule of thumb: under 55 you are likely being skipped, 55 to 79 is partly ready, and 80 and up is cite-ready. Clearing the heaviest failed signal moves the number most.

AI readiness score
0 READY
At risk 0-54Partly 55-79Ready 80-100
What we score

Six signals, weighted by what wins a citation.

Not every signal counts the same. Citable facts carry the most weight because a concrete number is what an answer quotes. Crawler access, schema and author trust follow, so the score reflects the real levers.

  • Citable facts and data carry the most weight, 20 of 100
  • Crawler access, structured data and author trust each add 18
  • Question and FAQ shape adds 14, llms.txt and depth 12

Why it works: engines cite the page that backs a claim with a specific, attributable fact. That is why citable facts is the single heaviest signal here.

Signal weights
Citable facts
20 pts
Crawler access
18 pts
Structured data
18 pts
Author trust
18 pts
Question/FAQ
14 pts
llms.txt and depth
12 pts
Your fix list

A ranked to-do list, heaviest gap first.

Every failed signal becomes a concrete instruction, sorted by how many points it wins back, so you spend effort where it moves you from read to cited.

  • Failed signals only, ranked by point value
  • A reason and the exact change to make
  • Re-run to watch the score climb as each fix lands

Pro tip: start at the top. Fixing the two heaviest gaps on our example page, citable facts and author trust, is worth 38 points on its own.

Priority fixes · summitgear.com

62 / 100 now2 gaps+38 to fix
1 Add five or more concrete, citable facts (20 pts)
2 Add a named author and a visible date (18 pts)
3 Re-run to confirm 100 of 100
cut pack weight 18%tested over 240 milesmaking gear since 201940 hour battery9.2 oz
And the rest of the picture

Everything the audit reads in one pass.

Crawler by crawler

Sees which AI bots are allowed or blocked, so one wrong robots.txt line cannot quietly hide you.

GPTBotClaudeBotPerplexityBotGoogle-Extended

Schema types found

Lists the JSON-LD types on the page, so you can see what engines can identify and attribute.

ArticleFAQPageOrganization

Reads robots.txt live

Fetches the real file at run time and catches the single Disallow line that removes you from AI answers.

llms.txt check

Flags whether you point AI crawlers to your most citable pages with an llms.txt file.

Content depth

Counts the real words on the page, because thin pages rarely get chosen as the source worth quoting.

Google-Extended note

Separates crawl access from Google's generative-AI use, so you set each one on purpose, not by accident.

See it in action

How to read your result.

A score is only useful if you know the move it implies. Here is the exact read on our example domain, summitgear.com, the way an experienced SEO would think it through.

62/100Readiness
Read-ready, not cite-ready. Crawlers are allowed and the schema is clean, but the score stalls at amber because the two citation-deciding signals are missing.
2 foundFacts · 5+ needed
The heaviest gap. Engines quote a specific number and attribute it to you. Turn three vague claims into concrete stats and you win back 20 points.
MissingAuthor and date
The tie-breaker. Add a named author and a visible updated date and you win back 18 points, the trust that decides between two similar pages.
100Verdict
Add five concrete facts and a bylined author with a date, and summitgear.com goes from 62 and skipped to a perfect 100 and cite-ready. That plan came from six signals read in seconds.
Where it shines

Built for teams that want to get cited.

Founders

Get named in the answer

Your brand is new and Google's first page belongs to incumbents. See which signals keep you out of the AI answers your buyers already read.

SEO and marketing leads

Know why a rival gets cited

Stop guessing why ChatGPT names a competitor and not you. Get a scored, signal-by-signal read in seconds, not a research project.

Content teams and agencies

Turn the fix list into a brief

Every failed signal becomes a concrete instruction for the next page: the facts to add, the author to credit, the schema to ship.

Publishers

Protect the pages you rank for

Confirm your best content is still crawlable, attributable and trusted as engines keep changing how they read a page.

Good to know

Questions, answered.

It fetches your live page, robots.txt and llms.txt the moment you run it, and scores what it actually finds. No cached guesses and no login.

Six signals, each weighted by how much it influences a citation: citable facts 20, crawler access, structured data and author trust 18 each, question and FAQ shape 14, and llms.txt with depth 12. Your score is the weighted share you pass.

Yes. It scores the signals all four rely on to read and cite a page. Watching whether each one actually cites you over time, and whether a competitor gets cited instead, is part of the paid dashboard.

No. Ranking is your spot in the list of blue links. A citation is an answer engine naming your page as its source inside the answer itself. This audit scores your citation readiness, not your rank.

Yes. Every signal comes with a plain reason and an exact fix, and the list is ranked so you always know what to do next. No SEO background needed.

The free audit scores how citable your page is right now. The Metric Vault dashboard tracks whether the engines actually cite you, how often, which pages they name, and how you compare to competitors over time, plus the rest of the platform's 60+ SEO and AI tools.

Ready when you are

See if AI will cite your page.

It takes one domain. You will get a 0-100 readiness score, a pass or fail on all six signals, and a ranked list of the fixes that move you from read to cited.

See if AI actually recommends you

The checker tells you if AI can read your site. Metric Vault tracks whether ChatGPT, Gemini, Perplexity and Claude actually mention you, how you stack up against competitors, and how it trends over time, plus the rest of the platform's 60+ SEO and AI tools.