How to Audit Meta Tags on Any Page

Meta Tag Extractor does that reading. If you give it a URL, it fetches that URL so it can parse the HTML. That is required for a live page. Do not feed it hosts you are not allowed to pull. For drafts, paste the strings into SERP Snippet Preview and Open Graph Preview without touching production.

Title and description as they exist in the HTML

The title is <title>. It is what tabs show and what search engines start from. Audit questions:

• Is it unique on this URL, or did every tool inherit “Free Online Tools”?
• Does it name the job (“JSON Formatter”) rather than a slogan?
• Is it so long that SERP Snippet Preview cuts the verb?

The meta description is meta name="description". It does not rank by itself. It is the sentence under the title when it is used. Audit questions:

• Is it empty, duplicated site-wide, or still “Lorem ipsum”?
• Does it describe this URL’s form or article?
• Does it end mid-clause at mobile width?

Writing those two well is a separate skill: How to Write Title Tags and Meta Descriptions. The audit is whether the shipped values match that intent.

If the extractor shows an old title after you “changed it in the CMS,” you are looking at cache, a different locale, or a tag the CMS does not control (hard-coded in a layout).

Canonical: one URL, not a suggestion

link rel="canonical" says which URL is the preferred copy. On a healthy tool page it is the clean path: https://www.example.com/json-formatter, HTTPS, production host, no leftover ?utm_.

Failure modes worth a red flag:

• Canonical points at http:// or www while the page redirects the other way - pick one story.
• Canonical points at / from every article.
• Parameterized URLs (?q=, ?sort=) either lack a canonical or canonicalize to themselves when they should fold into the clean tool URL.
• Staging host in the href after a copy-paste of the preview deployment.

The extractor reports the href as written. You still have to decide whether that href is the URL you want in an index. Tool-page specifics: On-Page SEO for Tool Pages.

Robots and noindex: the switch people forget

meta name="robots" (and googlebot) can include noindex, nofollow, noarchive. Pre-launch branches use noindex on purpose. The incident is leaving it on after go-live, or inheriting it from a parent layout.

Also check HTTP X-Robots-Tag if you serve PDFs or other non-HTML - that will not appear in an HTML extractor. For HTML, the meta robots line is the usual culprit.

nofollow on the whole document is rare and usually a mistake on a public tool. noindex on a thank-you page can be correct.

If the page is blocked in robots.txt, meta tags will not save it; the crawler may never see the head. That is a different file, outside this tool.

Open Graph and Twitter as a second layer

og:title, og:description, og:image, og:url, twitter:card, twitter:image are for link previews, not for the blue search link. They often copy the SEO title by default. That is fine when the title is already a human sentence. It is bad when the SEO title is “Best Free X Tool | Brand | 2026” and the share looks like spam.

Audit:

• Image exists, is reachable, and is large enough (aim ~1200×630 for og:image).
• og:url matches the canonical host.
• You did not set twitter:card to summary_large_image with a 16×16 favicon.

Preview the card in Open Graph Preview. Extract first so you are previewing what shipped, not what is in a Figma comment.

How to extract and review before launch

1. Deploy to a URL you are allowed to fetch, or use a local HTML dump if the app is not public yet.
2. Open Meta Tag Extractor, enter the public URL, run it. The host is contacted.
3. Write down title, description, canonical, robots, og:image.
4. Drop title + description into SERP Snippet Preview. Edit copy until both widths read as complete thoughts.
5. Fix the template. Extract again. The second pass is where noindex and staging canonicals show up.

View Source in the browser is the same data if you distrust a proxy fetch. The extractor is faster when you are checking ten URLs in a row.

Common mismatches after a CMS publish

• SPA fallback. The fetched HTML is <div id="root"> and the title is still the shell default. You need prerender or server-rendered meta.
• i18n. /de/tool still has the English title because the default locale filled the tag.
• Pagination. Page 2 has the same title as page 1.
• Trailing slash. Canonical and og:url disagree with the redirect.

None of these require a persona or a sample invoice PDF. They require the live head.

Extract the live head before you call it done

Run Meta Tag Extractor on the public URL you are about to call done. Then check truncation in SERP Snippet Preview. If the next job is the share card, continue with How Open Graph Tags Control Link Previews.