See which AI bots crawl your site, how many visitors AI sends you, and what to fix.
AI crawlers don’t run JavaScript, so your analytics never sees them.
GPTBot, ClaudeBot, PerplexityBot and the rest, by page and by day. Verified against each vendor’s IP ranges, so spoofed bots don’t count.
A 0–100 score for every page from 14 checks, with blockers flagged first and fixes ranked by impact.
Visits from ChatGPT, Perplexity, Gemini, Claude and Copilot, and the pages they land on. Most analytics file them as direct.
Enter your domain. Sign in with Google or email.
A tag for AI traffic, a collector for crawlers. Or install the WordPress plugin.
AI crawls and visits show up live. Fix what is flagged.
Google links to your page. AI has to quote it. Three problems cause most misses.
AI quotes the first passage that answers the question. An intro that restates it gives the engine nothing to lift.
Answers are built from chunks, not pages. A paragraph that opens with a bare pronoun makes no sense on its own, so it never gets quoted.
A robots.txt rule, a stray noindex, or content that only renders in JavaScript. Fixable in minutes, fatal until then.
14 checks, every weight published: 25 points for access, 15 for trust, 60 for a passage worth quoting.
See every checkanswer-first alone is worth 13 points.
One click copies a fix prompt for Claude Code, Cursor or Codex. Paste it, and the agent fixes what blocks AI.
Logged when they visit, checked against your robots.txt.
OpenAI (training + search index)
ChatGPT Search
Anthropic / Claude
Perplexity index
Google AI Overviews / Gemini
SEO and analytics tools measure a search box people use less every month. None of them tell you whether GPTBot can reach your pages, or whether ChatGPT sent you anyone last week.
CrawlRadar measures what can be observed: the crawls, the AI visits, and whether a page can be read. Then it hands you the fix, not just the finding.
No tool can promise you citations. CrawlRadar shows you what it can prove, and says so when a signal is unproven.
Ready to be cited
This page scores 91 on our own engine. Every public page on this site must clear 85, or the build fails.
From $9 a month. Every feature on both plans, and AI crawls never count toward your bill.
Both plans include every feature. Growth adds websites, monitored pages and audit size.
| Feature | Starter | Growth |
|---|---|---|
| Usage | ||
| Websites | 1 | 10, then $2/mo each |
| Monthly pageviews | 10k to 10M | Same, shared across sites |
| AI crawler visits | Unlimited | Unlimited |
| AI crawlers | ||
| AI crawlers identified by name | 56+ | 56+ |
| Verified against vendor IP ranges | Included | Included |
| Pages each crawler fetches, by day | Included | Included |
| AI traffic | ||
| Visits from ChatGPT, Perplexity, Gemini, Claude, Copilot | Included | Included |
| Landing pages for AI visits | Included | Included |
| Fix | ||
| Readiness score, 14 checks per page | Included | Included |
| Fix prompts for Claude Code, Cursor, Codex | Included | Included |
| Sitemap audit | 100 URLs per run | 500 URLs per run |
| AI checks on your own API key | Included | Included |
| Monitor | ||
| Monitored pages with email alerts | 10 | 100 |
| Score history and trend lines | Included | Included |
| PDF, PNG and CSV export | Included | Included |
| Support | ||
| Support | Priority, from the founder | |
FAQ
Frequently asked questions
What it does first, then the questions that decide whether to pay for it.
Score the page you’re on, even staging or behind a login. No site access requested, nothing sent to us.
The same engine runs on Free, on Pro and in the extension.
| Check | Lever | Points |
|---|---|---|
| answer-first | Selected | 13 |
| extractability | Selected | 10 |
| headings | Selected | 9 |
| scannability | Selected | 9 |
| crawler-access | Retrieved | 8 |
| specifics | Selected | 8 |
| render-dependency | Retrieved | 7 |
| entity-clarity | Ranked | 7 |
| faq | Selected | 7 |
| indexability | Retrieved | 6 |
| llms-txt | Retrieved | 4 |
| freshness | Ranked | 4 |
| topical-focus | Ranked | 4 |
| schema | Selected | 4 |
| Total | 25 / 15 / 60 | 100 |
Grades run A ≥ 85, B 70–84, C 55–69, D 40–54, F under 40. A check that cannot be measured, such as a robots.txt that times out, is reported as not checked and removed from the denominator, never scored as a zero. A score you cannot trust is worse than no score.
The heaviest single check is answer-first structure at 13 points, followed by chunk extractability at 10 and heading hierarchy at 9. Retrieval checks are cheaper individually: crawler access is 8 points and indexability 6, because they are usually one-line fixes in robots.txt or a meta tag, while an answer buried under two paragraphs of preamble is a rewrite.
25 pts
Retrieval decides whether GPTBot, ClaudeBot, PerplexityBot and Google-Extended can fetch this URL and read it at all, without running any JavaScript.
crawler-access
Reads robots.txt group by group, the way each agent does, and reports which AI crawlers are allowed on the exact path you scanned.
render-dependency
Compares raw HTML against rendered text. Content that only appears after JavaScript runs is content most AI crawlers never receive.
indexability
The noindex tag, the canonical target and the status code: three ways a page quietly removes itself from an index.
llms-txt
Checks for a published llms.txt. It is an emerging convention with no proven citation impact yet, which is why it is worth four points and not twenty.
15 pts
Ranking measures whether the page names an author, carries a machine-readable date, and holds to one subject tightly enough to be trusted as a source.
entity-clarity
Named author, publisher and organisation markup, so a model can establish who is speaking and why they would know.
freshness
Published and modified dates, and whether they are machine-readable rather than only printed for humans.
topical-focus
How tightly the page holds one subject. A page about everything is retrieved for nothing.
60 pts
Selection is where almost every page that already ranks on Google loses the citation, because a passage has to survive being lifted out on its own.
answer-first
Whether each section answers its own heading in the first sentence. This is the heaviest check on the model because it is the one that decides whether there is anything to lift.
extractability
Whether a paragraph still means something once it is lifted out of the page, which is the only state an answer engine ever reads it in.
headings
Whether the outline is well-formed and whether headings read as the questions their sections answer.
scannability
Paragraph length, sentence length, lists and tables: the structure that survives being split into chunks.
specifics
Numbers, dates, names and units. Concrete claims get quoted; generalities get paraphrased into someone else’s answer.
faq
Detects genuine Q&A structure, the format answer engines lift most readily because the question is already matched for them.
schema
Validates JSON-LD for the four types that carry citation weight: Article, FAQPage, HowTo and Organization.
