Auditing a page you can actually see
Crawlers are the right instrument for a thousand pages. For one page, or for the twenty that carry most of a site's traffic, a crawler is a subscription and a scheduled report standing between you and something you could check in a minute.
The manual audit is worth learning anyway, because it is what you fall back on when a crawler tells you a tag is missing and you cannot see why. It comes down to reading the head of a page carefully and knowing what should be in it.
View source is not what the page says
Press Ctrl+U, or Cmd+Option+U, and you get the HTML the server sent. That is not necessarily the page a search engine indexes. Anything a framework builds after load — and single-page applications frequently build the entire head this way — is absent from the source and present in the rendered document.
The rendered version lives in the browser's inspector. Open it, find the <head> node, right-click and copy the outer HTML. That is the markup worth auditing, and comparing the two is itself diagnostic: a title that appears in the inspector but not in the source is being written by JavaScript, which Google will usually pick up on a second rendering pass and which other crawlers and most social platforms will not.
The four checks that catch most problems
Paste the head into the [msm_tool slug="meta-tag-analyzer"] and it reports on a title, a description, an og:title and a viewport, with character counts for the first two. Those four cover the common failures, and the significance of each absence is very different:
- No title. Serious. Google will invent one from the page and you have surrendered your headline.
- No description. Minor. Google frequently rewrites descriptions anyway, and a page with no description gets a snippet pulled from its text. Worth writing for pages that matter, not worth a project.
- No
og:title. Usually fine. Platforms fall back to the title tag. Add one when the sharing headline should differ from the search headline, which is a real editorial decision rather than an error.
- No viewport. Serious and often overlooked. The page renders at desktop width on a phone, which fails mobile usability outright and is a far larger problem than any tag length.
Character counts are a proxy
Both limits are pixel measurements. Google truncates a title at a width, not at a count, so sixty characters of narrow lowercase may survive where fifty of wide capitals do not. The counts are a useful signal when you are well over — a seventy-character title is going to be cut — and near-useless when you are within a few characters of the figure.
The other reason not to trust them: being within the limit does not mean your text will be displayed at all. Google rewrites titles when it finds something on the page that fits the query better, and rewrites descriptions routinely. Writing a good one improves your odds; it does not buy a guarantee.
What pattern matching cannot see
Any tool that matches text rather than parsing a document has blind spots, and it is better to know them than to trust a clean report. A tag whose attributes are split across several lines, an unusual attribute order, a tag assembled by script after load — all read as absent when they are present.
So a reported absence is a prompt to look, not a verdict. Search the pasted markup for the tag name yourself. If it is there and the report says otherwise, the formatting is unusual, which is worth knowing on its own because other consumers of your markup may be equally literal.
Why no free tool takes a URL
A browser will not let a script on one site read the HTML of another. That is the same-origin policy, it is deliberate, and no amount of engineering gets around it in the browser.
Tools that do accept a URL are fetching it on a server. That works, and it also means the address of every page you audit — staging sites, unlaunched pages, client work under embargo — is sent to somebody else's machine and quite possibly logged there. Pasting markup is a small inconvenience in exchange for the audit staying on your own computer.
A short routine for one page
Copy the rendered head. Check the four above. Then read the rest of the head with your own eyes for the things a four-check tool does not cover: exactly one title tag rather than two, a canonical that points where you expect, a robots tag that does not say noindex on a page you want indexed, and a charset declaration near the top.
That last group is where the genuinely expensive mistakes live. A duplicated title is cosmetic. A stray noindex left behind after a launch removes the page from search entirely, and nothing about the page looks wrong when you visit it.