See your page the way a crawler does
Your browser shows you the finished page. A crawler gets raw HTML, no JavaScript, and one shot at understanding it. This shows you that version — structure, signals, what is missing — and ranks the fixes by what they are actually costing you.
The gap between your page and their page
You look at your site in a browser with JavaScript running, fonts loaded, images lazy-loading in as you scroll. A crawler makes one request, gets whatever HTML the server hands back, and decides from that what the page is about.
Most of the expensive technical SEO problems live in that gap. Content that only exists after a framework hydrates. A noindex left on after a staging deploy. A canonical quietly pointing at the homepage. Six H1s from a page builder. None of it is visible while you are looking at the rendered page — all of it is obvious the moment you look at the response.
The AI dimension makes it sharper. Googlebot will come back and render your JavaScript eventually. Most AI crawlers will not. For OAI-SearchBot, PerplexityBot and ClaudeBot, the HTML in this tool is not the first draft of your page — it is the whole page, permanently.
Indexability first
Status, robots directives and canonical decide whether anything else matters. That verdict sits at the top for a reason.
Structure second
Headings, schema and links are how a machine works out what the page is about and how it relates to everything else.
Delivery third
Viewport, compression, blocking scripts and first-byte time — the things that decide whether people stay long enough to convert.
Frequently asked questions
Why does this show different content than my browser?
Because your browser runs JavaScript and this does not. What you see here is the raw HTML the server returned — which is exactly what a crawler receives on its first pass. If your page looks empty here but full in a browser, its content is being assembled client-side. Googlebot does render JavaScript on a later pass, but most AI crawlers never do, so anything missing here is missing for them permanently.
What does the health score mean?
It starts at 100 and subtracts for each finding, weighted by consequence — a noindex directive or a missing viewport costs far more than a missing og:image. It is a summary of the findings below it, not an independent measurement, and it is not a ranking prediction. Read the findings; the number is just a way to compare pages at a glance.
Why did the fetch fail or return 403?
Some sites verify Googlebot by reverse DNS and reject anything claiming to be Googlebot from an IP that is not Google's. Wikipedia and several large publishers do this. It is not a fault in your site — switch the user-agent to "Regular browser" and try again. The tool offers that as a one-click retry when a fetch fails.
Which user-agent should I test with?
Start with Googlebot (mobile), because Google indexes the mobile version of your site. Then compare against OAI-SearchBot or PerplexityBot if AI visibility matters to you — some sites serve different markup to different agents, deliberately or by accident, and that comparison is where you find it. If a bot user-agent is being blocked outright, that is itself the finding.
Does this replace Google Search Console's URL Inspection?
No, and it is not trying to. Search Console tells you what Google actually did with your URL — whether it is indexed, when it was crawled, what the rendered DOM looked like. That is authoritative and this is not. What this does that Search Console cannot is work on any URL, including a competitor's, and let you compare across AI crawler user-agents. Use both.
Why does the word count look low?
The tool reads the main or article region when a page marks one up, and otherwise the body with navigation, footers and sidebars removed. That is deliberate — counting your menu inflates the number without telling you anything about the page. It reports which region it used alongside the count.
Is the page I check stored anywhere?
No. We fetch the URL you enter, parse it at the edge, return the analysis and keep nothing. The raw HTML is never stored and never returned to you verbatim — only the structured extraction you see. We do not log which URLs are checked.
Related tools and reading
One page is a spot check.
A site is a pattern.
The same three mistakes usually repeat across every template on a site. Finding them once is useful; finding the template that generates them is what actually changes the numbers.
Get a Free Technical Audit