Website Meta / SEO Tag Checker

A meta description is a sentence in the head of the page proposing how it should be summarised in a result listing. It is not a ranking signal, and search engines replace it freely when some other passage matches the query better, which is why the words you wrote are often not the words people see. Truncation is decided by pixel width rather than character count, so the familiar rules of thumb — roughly sixty characters for a title, around a hundred and fifty-five for a description — are conversions from that and shift by device. Nothing here counts characters for you; the full value is printed, so judge it by eye.

Three causes, roughly in order. The platform cached a preview from an earlier version and will serve it until its own debugger is asked to scrape the URL again. The og:image value is a relative path, where scrapers need an absolute one starting https. Or the image fails that platform's rules on dimensions, aspect ratio or format. The preview drawn below the results is a real check on the last two, because it is your own browser loading that exact address — if it comes back broken there, no scraper will do better. A shared link carrying a query string can also count as a different URL entirely.

Mostly not. X falls back to Open Graph, so a page with og:title, og:description and og:image produces a card with no twitter: tags at all. The one worth setting is twitter:card, which chooses the shape: a small square thumbnail beside the text, or the wide image most people are after. The twitter: versions of title, description and image exist for when a card should deliberately differ from what other platforms show, which is rare and a maintenance cost you probably do not want. An empty Twitter block sitting beside a full Open Graph one is usually correct rather than a fault.

Because you are looking at two different documents. The Elements panel in developer tools shows the page after scripts have run and rewritten it; view-source shows what the server actually sent, and that is what a crawler receives. A tag present in one and absent from the other is being written by JavaScript after load — the default for frameworks that render in the browser unless server-side rendering or prerendering is turned on, and a habit of some SEO plugins too. Compare view-source with your template before blaming the checker. Failing that, suspect a CDN handing an older copy to anyone not signed in.

Not from here, and for two different reasons. A hostname resolving to a private, loopback or link-local address is rejected before any request goes out, which covers localhost, the 192.168 and 10.x ranges, and anything that only resolves inside a company network. A public URL sitting behind basic auth or a password page will be fetched, but what comes back describes the gate rather than the page you meant. For either case, read the built HTML on your own machine, or publish to a preview URL that is public but held out of search with a robots meta tag.

Two things leave your machine. The URL travels to our own route as a query parameter on a GET request, the shape ordinary web server logs record; that route writes it to no database. The page itself is then fetched by our server, so the site being checked sees a request from our host. The preview image is the exception — your browser has to request it directly from whoever hosts it, so your address reaches that third party even though the page fetch did not. Results are held in the page's state alone: reload and they are gone. No account, no history, nothing to clear.

It names the address you want treated as the real one when the same content is reachable at several. Tracking parameters, session ids, print versions, http against https, with and without a trailing slash — each is a separate URL to a crawler, and left alone they compete with one another. Most templates emit a canonical pointing at the page itself, which is the sensible default. The expensive failure is a template pointing every page at the homepage, or at a staging domain after a migration: the hint is usually honoured, so whole sections quietly drop out of search. It is the field worth reading on every page you check.

Check robots first, because it is the one value here that is an instruction rather than a suggestion, and a noindex left behind by a staging build is the commonest single cause of a page that will not appear. Read it precisely: noindex removes the page, nofollow only concerns the links on it, and the two get muddled constantly. Note too that noindex and a robots.txt block fight each other — a URL disallowed in robots.txt is never fetched, so the noindex on it is never seen, and the address can linger in results with no description under it. If robots is clean, look at internal links and patience.