Meta tags that actually affect anything
Which tags still influence search and sharing, which have been ignored for two decades, and the lengths that get truncated.
Last reviewed · 1,256 words
In short
- The title tag is the single most important on-page element. 50 to 60 characters, with the distinctive words first.
- The meta description does not affect ranking. It affects whether anyone clicks, which is why it still matters.
- The keywords meta tag has been ignored by Google since 2009. Nothing reads it.
- Open Graph tags control how a link looks when shared, and og:image is what most people actually notice.
- Google rewrites the title it displays in roughly 60% of results, usually when the tag is too long or not descriptive.
Meta tags sit in the head of an HTML document and describe the page. Most of them have no effect at all, and a small number matter a great deal.
The title tag
Not technically a meta tag and the most important element on the page.
<title>Page Title — Site Name</title>
50 to 60 characters before truncation, though Google truncates by pixel width rather than character count, so a title in wide capitals is cut sooner than one in lowercase.
Put the distinctive words first. A title reading "Home — Company Name" wastes the position that matters, and a user scanning results reads the left-hand side.
Every page needs its own. Duplicate titles across a site are one of the most common technical SEO problems, and they leave the search engine to choose which page to show for a query.
Google rewrites the displayed title in a large share of results — studies put it above 60% — usually when the tag is too long, keyword-stuffed, or less descriptive than the page's H1. Writing a clear, accurate title is the way to keep the one you wrote.
The meta description
<meta name="description" content="What this page is about, in one sentence.">
It is not a ranking factor. Google confirmed this years ago and has not changed the position.
It is a click-through factor, which is why it still deserves attention. The description usually appears as the snippet under the title, and it is the closest thing to ad copy the page has.
150 to 160 characters before truncation, again by pixel width. Mobile truncates sooner, around 120.
Google frequently ignores it and generates a snippet from the page instead, particularly when the query matches text elsewhere on the page better than the description does. That is not a failure — a query-relevant snippet usually serves the searcher better.
Write it as a sentence that would make someone click, not as a list of keywords.
Tags that do nothing
<meta name="keywords"> — Google stopped using it in 2009 and announced doing so. It was abandoned because it was trivially gamed. Nothing meaningful reads it today, and including it merely shows competitors what you are targeting.
<meta name="revisit-after"> — never supported by any major search engine.
<meta name="author"> — no search effect. Useful as documentation.
<meta name="generator"> — identifies the CMS. No effect.
<meta name="rating"> — no longer used for content classification.
Removing these changes nothing about the page's performance, and it makes the head shorter and easier to read.
Tags that do something
Robots controls crawling and indexing:
<meta name="robots" content="index, follow">
noindex keeps a page out of results. nofollow stops link equity passing. noarchive prevents a cached copy. max-snippet, max-image-preview and max-video-preview limit what appears in a result.
Note the common error: a noindex page cannot pass any signal, and a page blocked in robots.txt cannot be read at all — so a noindex tag on a blocked page is never seen, and the page can still appear in results as a bare URL.
Canonical names the preferred version of a page with duplicate or near-duplicate content:
<link rel="canonical" href="https://example.com/page/">
Essential where the same content is reachable through multiple URLs — with and without a trailing slash, with tracking parameters, in a paginated series.
Viewport is required for mobile rendering:
<meta name="viewport" content="width=device-width, initial-scale=1">
Without it, mobile browsers render at desktop width and scale down, which fails mobile usability checks and is one of the few meta tags with a direct ranking consequence.
Charset should be the first thing in the head:
<meta charset="utf-8">
Hreflang signals language and regional variants, and it must be reciprocal — each version pointing at all the others, including itself.
Open Graph and social cards
Open Graph controls how a link appears when shared on Facebook, LinkedIn, WhatsApp, Slack and most other platforms.
<meta property="og:title" content="Page Title">
<meta property="og:description" content="One-sentence summary.">
<meta property="og:image" content="https://example.com/image.jpg">
<meta property="og:url" content="https://example.com/page/">
<meta property="og:type" content="article">
og:image is the one that matters most, because it is what a person actually sees in a feed. 1200×630 pixels is the standard, under 5 MB, and it must be an absolute URL — a relative path silently produces no image.
Twitter has its own tags and falls back to Open Graph for anything missing, so twitter:card set to summary_large_image plus the OG tags is usually sufficient.
A missing or broken og:image is one of the most visible avoidable errors on a website, because every share of that page looks broken in a way the page owner rarely sees.
Structured data
Schema.org markup in JSON-LD is separate from meta tags and is what produces rich results — star ratings, FAQ accordions, breadcrumbs, recipe cards.
It does not directly affect ranking. It affects how a result is displayed, and a result occupying more space with more information attracts more clicks.
Common types worth implementing: Article, Product, FAQPage, BreadcrumbList, Organization, LocalBusiness. Google's Rich Results Test validates the markup and shows what it would produce.
The rule that matters: the structured data must match what is visible on the page. Marking up an FAQ that is not on the page, or a rating that does not exist, is a policy violation and results in a manual penalty.
A workable head
For most pages, this is the complete useful set:
charset, firstviewporttitle, unique and descriptivedescription, written for a humancanonical- Open Graph: title, description, image, url, type
twitter:card- Structured data, where the page type has an appropriate schema
Everything else is optional, and the keywords tag is not optional so much as pointless.
Writing a title that survives
Google rewrites titles when it judges its own version more useful, and the patterns that trigger a rewrite are fairly consistent.
Too long. Anything beyond about 60 characters is likely to be shortened or replaced.
Keyword stuffing. "Buy Shoes | Cheap Shoes | Best Shoes Online | Shoe Store" is replaced, reliably.
Boilerplate repeated across a site. A brand name at the front of every title wastes the position and invites a rewrite.
A title that contradicts the H1. Where the two disagree, Google frequently prefers the H1, which is a reasonable signal that the H1 describes the page better.
The reliable pattern is a specific description of this page, then a separator,
then the site name: EMI Calculator with Amortisation Schedule — Calci. Distinctive
first, brand last, under sixty characters.
Checking what you shipped
Meta tags fail silently, which is why they are worth verifying rather than assuming.
View source, not the inspector. The rendered DOM shows tags injected by JavaScript, which a crawler may or may not see. The raw HTML is what matters for tags that must be present at crawl time.
Use a social card debugger. Facebook, LinkedIn and Twitter each provide one, and each caches aggressively — after changing an og:image you generally have to force a re-scrape or the old image persists for days.
Use Google Search Console. The URL Inspection tool shows what Google actually fetched and rendered, which is the only authoritative answer to whether a tag was seen.
Check on mobile. Titles and descriptions truncate sooner, and a description written to 160 characters loses its ending on a phone.
The single most common shipped error is an og:image with a relative URL, which renders nothing and is invisible to the page owner because they never see their own link shared.
What this tool assumes
- Character-count limits are approximate, since search engines truncate by pixel width and font metrics differ.
- Open Graph image URLs must be absolute; a relative path will not render.
- The output is a starting point that should be checked in Google's Rich Results Test and a social card debugger before it ships.
- Recommendations reflect current search engine behaviour, which changes.
- Everything runs in your browser. Nothing you enter is sent anywhere.