Every web page carries a hidden header — the
section — filled with meta tags that visitors never see but browsers, crawlers and social networks read carefully. They declare the page's title, its description, how it should preview when shared, and whether search engines may index it at all.Not all meta tags are equal, and their fortunes have shifted over the years. Some remain essential in 2026; one famous tag has been dead for over a decade yet refuses to disappear from advice articles. Here is the honest field guide: what matters, what is dead, and what is merely nice to have.
Title and meta description: still the headliners
The
The meta description (the ~155-character summary under the headline) is not a direct ranking factor, but it is your free advertising copy: it decides whether the searcher clicks your result or your competitor's. Write it like a pitch, not a keyword list. Every important page deserves a hand-written, unique description; duplicated or missing descriptions are one of the most common fixable SEO flaws.
Meta keywords: the tag that died
In the late 1990s, the keywords meta tag told search engines what a page was about — and was promptly abused into irrelevance by keyword stuffing. Google officially confirmed it ignores the tag in 2009, and Bing has indicated stuffed keywords can even count as a spam signal.
Yet the tag shambles on in tutorials and CMS templates, seventeen years after its funeral. If you see meta keywords in advice written for 2026, treat it as a signal about the advice, not about SEO. Delete the tag, or leave the field empty — either way, spend that energy on content instead.
Open Graph and Twitter Cards: winning the share
When someone pastes your link into WhatsApp, X, Facebook, LinkedIn or Slack, the preview card — headline, image, blurb — comes from Open Graph tags (with Twitter's card tags as a supplement). Pages without them get whatever the platform scrapes, often a cropped logo and a random sentence. For any page you want shared, these tags are the highest-ROI markup you can add.
The image does most of the work: a clear 1200×630 image with minimal text dramatically outperforms the default scrape. Set og:url to the canonical address so shares consolidate on one URL, and validate every important page with the platforms' card debuggers before launch — cached bad previews are a common and avoidable embarrassment.
- og:title and og:description: the share headline and blurb.
- og:image: the preview image — 1200×630px is the safe standard.
- og:type and og:url: article vs website, and the canonical share URL.
- twitter:card: usually “summary_large_image” for the big preview.
Robots and canonical: the traffic cops
The robots meta tag instructs crawlers per-page: noindex keeps a page out of search results (thank-you pages, admin screens, thin archives), while nofollow tells crawlers not to pass ranking credit through the page's links. Used surgically, it keeps your index clean; used carelessly — a stray noindex on a key page — it can make a site vanish from Google overnight.
The canonical tag solves the duplicate-content problem the web cannot avoid: the same content reachable via multiple URLs (tracking parameters, http vs https, print versions). Pointing each variant at one canonical URL tells search engines which copy counts, consolidating ranking signals instead of splitting them across duplicates.
The supporting cast
A few tags are pure infrastructure. Declaring UTF-8 charset prevents the classic mojibake of garbled characters; the viewport tag is what makes responsive design actually work on phones (and mobile-friendliness is a genuine ranking factor). Neither will win you traffic, but missing them can quietly cost you.
Then there is the graveyard: revisit-after (never honoured), distribution, rating, expires and their cousins — tags that crawlers never meaningfully used. If a tag's only documentation is a 2008 blog post, let it go. And remember that structured data (Schema.org JSON-LD) is not a meta tag at all, but it is the modern way to earn rich results — arguably more impactful than any tag on this list.
- charset (UTF-8): declares the character encoding — non-negotiable.
- viewport: makes the page render correctly on mobile; a ranking factor via mobile-friendliness.
- theme-color: colours the browser chrome on mobile — a nice touch.
- description, author: harmless; author is ignored by Google but fine for humans.
Frequently asked questions
Do meta descriptions affect rankings?
Not directly — Google has said for years that it is not a ranking factor. Indirectly, yes: it controls your search snippet, and a compelling snippet earns more clicks. Higher click-through can correlate with better performance, so write descriptions as ad copy for the page, not keyword stuffing.
How long should a title tag be?
Aim for about 50–60 characters (Google measures in pixels, roughly 600). Front-load the important words, make each title unique, and match the page's actual content — a title that overpromises earns the click and loses the visitor.
Should I still add meta keywords?
No. Google ignores it, and has since 2009; Bing has said it can treat keyword stuffing as a spam signal. Adding meta keywords today costs you nothing but gains you nothing either — spend the effort on the description instead.
What image size works for social cards?
1200×630 pixels is the widely recommended size for Open Graph images (a 1.91:1 ratio), under a few megabytes, in JPG or PNG. Keep key text and logos away from the edges, since platforms crop unpredictably, and always test with each network's card validator before publishing.
Conclusion
Meta tags will not rescue thin content or a slow site — no tag ever has. But the few that matter — title, description, Open Graph, robots, canonical, viewport — are the difference between a page that presents itself well and one that mumbles. Audit them once per template, write them by hand for important pages, and let the dead tags (keywords, revisit-after, distribution) rest in peace.