Meta Tags for SEO: Which Ones Google Reads, and Which It Ignores

Inside the head: the lines visitors never see

Every web page has two parts. The body is what people read. The head is a short block of code above it that tells browsers and crawlers how to handle the page: what to call it, how to scale it on a phone, whether to index it, which picture to show when someone shares it. Meta tags live there, written as a name and a value, like <meta name="description" content="...">.

People say "meta tags" loosely and mean everything in the head. Strictly, two of the most important lines are not meta tags at all: the title element and the canonical link element. We cover them here anyway, because that is what the searcher means, and because they do more for your rankings than every true meta tag combined.

One rule comes before all the others: the head must be valid. Google accepts only title, meta, link, script, style, base, noscript and template elements there. When it meets anything else, such as an img or an iframe a chat widget injected near the top, it treats the head as finished and ignores every element after it. If a tag seems to be ignored, look above it first.

Every tag Google lists, what it does, and an example

Google publishes a short page of the meta tags it supports. Here it is in table form, with the head elements that sit beside them and the tags it says it ignores.

Two more details from the same page. Letter case generally does not matter in meta tags, except in the site verification tag, which must match exactly. And Google says it ignores tags it does not support, so an unrecognized tag costs nothing but clutter. Google also ignores the lang attribute and detects language from the text; for language versions, see international SEO.

The meta description: an ad Google may decline to run

Google's snippet documentation says snippets are "primarily created from the page content itself." It uses your description when it might describe the page more accurately than text pulled from the body. Snippets also change by query, so a page can show your description for one search and a paragraph from halfway down for another. Writing a description is a request, not an order.

The length rule everyone quotes is not Google's. Google says there is no limit on how long a description can be; results cut it to fit the screen. So put the point first and let the rest fall off if it must. Google also asks for a unique description on every page, says programmatically generated descriptions are encouraged on large database-driven sites as long as they read like a person wrote them, and warns that strings of keywords are less likely to be shown.

The description does not move rankings. Google said so in a September 2009 announcement, and its current snippet guide treats the tag purely as snippet material. Its job is the click, which is why it pairs with our guide to click-through rate. To hide one passage from snippets without touching the rest, wrap it in an element with the data-nosnippet attribute.

Robots rules: one word in the head can remove a page from Google

The robots meta tag is the most powerful line most owners have never read. When nothing is set, Google may index the page, show snippets and follow its links. The rules below change that, and they are case-insensitive.

• noindex keeps the page out of search results. nofollow tells Google not to follow its links. none means both.
• nosnippet blocks text snippets and video previews, and also stops the content being used as direct input for AI Overviews and AI Mode.
• max-snippet:[number] caps the snippet at that many characters, and limits what AI Overviews and AI Mode can use. max-image-preview (none, standard or large) and max-video-preview set the media previews.
• noimageindex keeps the page's images out of the index. unavailable_after:[date] drops the page from results after that moment.
• indexifembedded lets content embedded in another page through an iframe be indexed despite a noindex on the source.
• noarchive, nocache and nositelinkssearchbox are retired; Google ignores them.

For files with no head, such as PDFs, send the same rules as an X-Robots-Tag HTTP header. When rules conflict, Google applies the more restrictive one. The wider crawling and indexing picture, robots.txt included, belongs to technical SEO. If AI answers are taking your clicks and you are weighing nosnippet, read AI Overviews took my clicks first, because the same rule also strips your regular snippet.

The title and the canonical: not meta tags, worth more than most

Every page needs a title element, and it should be unique and specific to the page. Google builds the title link from it most of the time, but also weighs the main heading, og:title, anchor text and other prominent text, and will rewrite a title it finds vague or stuffed. Length, formulas by page type, the brand name question and why Google rewrites are all in our title tags guide, so we will not repeat them.

The canonical link tells Google which URL you prefer when the same content lives at several addresses: with and without tracking parameters, filtered product listings, printer versions. Google calls it a strong signal, not a command, and accepts it only in the head. Google recommends absolute URLs and one method per page (the HTML tag or an HTTP header, not both), and it warns against using noindex to choose a canonical. As a matter of practice, also make sure the canonical target itself loads, stays put and is indexable; a canonical pointing at a redirect or a noindexed page asks Google to prefer a URL it cannot use. Our canonical tags guide covers the mistakes, and redirects covers the stronger signal for pages you are retiring.

Open Graph: written for the share, read by Google too

Open Graph tags decide what appears when someone pastes your link into Facebook, a messaging app or a work chat: the headline, the picture, the line of text. The protocol was originally created at Facebook, and it uses property instead of name, as in <meta property="og:title" content="...">. Without these tags, Meta says its crawler makes a best guess at the title, description and image. Best guesses are how a roofing company's link ends up previewed with its footer logo.

Structured data is a different tool for a different reader: JSON-LD describes the business, product or article to search engines, while Open Graph describes the share card to apps. See schema markup for that half.

Housekeeping tags, and the ones to stop maintaining

Viewport and charset are dull and essential. Google treats the viewport tag as a sign the page is mobile friendly, which matters because it indexes the mobile version (more in mobile SEO). Google recommends UTF-8 for charset and asks that the content value in http-equiv tags be quoted. The Next.js App Router adds both tags to every page by default, even when a route defines no metadata.

Meta refresh works, but Google says it confuses users and that a server-side 301 is the better choice, so treat any refresh tag as a redirect waiting to be done properly. The site verification tag must stay in place as long as you rely on it for Google Search Console access.

Then there is the attic. The keywords tag has no effect on indexing or ranking, according to Google, and it hands competitors your target list in plain text. Rel next and prev links are no longer used. Revisit-after, distribution and author-style tags from old templates appear nowhere on Google's supported list. Deleting them changes nothing in Google, which is the point: stop paying anyone to maintain them.

Read your own head in five minutes

1. 1

   Open the source

   Load a key page, press Ctrl+U (Cmd+Option+U on a Mac) and search the source for <title, description, robots and canonical.

2. 2

   Check for one of each

   Two titles, two descriptions or two canonicals usually means a theme and a plugin are both writing tags. Keep one.

3. 3

   Hunt for noindex

   If the robots tag says noindex on a page you want found, that is the first fix of the day.

4. 4

   Compare against Google's view

   In Search Console, run the URL through URL Inspection and view the crawled page, to see the head Google actually received.

5. 5

   Test a share

   Paste the URL into Meta's Sharing Debugger and check the title, image and description it reads.

6. 6

   Repeat by template

   Pages built on the same template share the same mistakes, so check one service page, one blog post and one product or location page.

On a site with hundreds of pages, do this by crawler, not by hand. The free site audit flags missing and duplicate titles and descriptions, and the website health check catches a stray noindex in seconds. On sites DataCram builds, tags live in code, and Iris, one of the Athena CMS agents, drafts titles, descriptions, social cards and structured data for approval. If you would rather hand the whole job over, our SEO programs include it. For how meta tags fit with headings, copy and links on the page itself, read on-page SEO, or start from the top of the SEO guide.