Meta Tag Analyser

Every other meta tag analyser takes a URL and fetches it. A browser cannot fetch another origin, and this site has no server to do it on your behalf, so the input here is the page source instead — one Ctrl+U away for anyone doing this work.

What comes back is an audit of the meta tags the page actually carries: the title, the description, the canonical, robots, the viewport, the Open Graph block and the Twitter card. Each is checked against what it is for rather than against a list of tags a page ought to have, which is why the report is short — most of the two thousand meta tags in existence do nothing, and a checker that names them all is padding.

That limitation turns out to be an advantage. A tool that fetches can only read a page that is already live. This one reads a page that exists nowhere yet: a draft, a staging build, a template you are still writing. That is the moment when a missing canonical costs nothing to fix, rather than a recrawl later.

The title and description are measured, not counted. Both are checked against a pixel budget, because that is what Google truncates on. The same measurement drives the SERP preview, and it is the reason a vowelled Arabic title that looks far too long usually is not.

Ordered worst first. Errors, then what is worth a look, then what passed. Source order buries the one broken thing under nine green rows, which is how a checker gets skimmed once and never opened again.

The sample is a page doing four things wrong at once, all of them common: a title stuffed with the same phrase three times and far over the budget, a two-word description, a relative canonical, a relative og:image, and two h1 elements. Load it and the report separates the ones that break something from the ones that merely waste an opportunity.

The relative og:image is the quiet one. It resolves in a browser, so the page looks fine. Facebook, LinkedIn, Slack and WhatsApp all require an absolute URL and show no image at all — which is discovered when a link is shared, in public, by someone else.

A short description is flagged too. Under about 40% of the budget the space is being wasted, and it is the line deciding whether anyone clicks. Google substitutes its own passage when a description is missing, and chooses badly more often than not.

If the page declares language alternates, they are checked as a cluster with the same rules as the hreflang generator — including the missing self-reference, which discards the whole set silently.