Meta Tag Analyzer

The tags that matter and the ones that do not

Title, meta description, canonical, robots and the Open Graph set do real work. The meta keywords tag has been ignored by Google since 2009 and by every other major engine since; its presence is harmless and its absence costs nothing. Author, revisit-after and most of the tags found in old templates have no effect on search at all.

Canonical errors are the expensive ones

A canonical pointing at the wrong URL removes a page from the index as effectively as noindex. The failure modes are consistent: a template that hard-codes the homepage canonical on every page, a staging domain left in place after launch, a canonical to a URL that redirects, or a chain where A points at B which points at C. Canonicals must be absolute, self-referential unless deliberately consolidating, and point at a URL that returns 200.

Conflicting signals get resolved unpredictably

A page that is canonicalised to another URL while also being noindexed sends contradictory instructions, and Google may apply the noindex to the canonical target. A page blocked in robots.txt while carrying a noindex tag will never have the tag read. When indexing behaves inexplicably, the cause is usually two signals disagreeing rather than one being wrong.

Open Graph is not optional any more

Every social platform, messaging app and link preview reads og:title, og:description and og:image. Without them the preview falls back to whatever the platform can scrape, which is usually unflattering. og:image needs an absolute URL, should be around 1200 by 630 pixels, and must be publicly accessible — an image behind authentication produces a blank card that is very hard to diagnose because it looks fine to you when logged in.

One H1, matching intent

Multiple H1s are not a penalty but usually indicate the page has no clear single subject. More useful is checking that the H1 and the title tag agree: when they describe different things, Google frequently replaces the title with the H1, and you lose control of the result.

Check the rendered output, not the source

On a JavaScript-rendered site, meta tags injected client-side may or may not be seen depending on how and when they are set. Viewing source shows what the server sent; the rendered DOM shows what a crawler executing JavaScript sees. Auditing only the first misses an entire class of problem.