Meta tags: what search engines read before anyone visits the page

What a meta tag is#

Meta tags are in the page's code, in <head>, and do not show on the page itself. They tell search engines, browsers and sharing services what the page is called, what it is about, whether it may be shown in search results and which address is the real one. Google itself lists which tags it understands [1]. The tags that control how links look when they are shared have their own page: Sharing on social media.

Title: the page's name in the search result#

<title> is the most important tag. Google builds the heading in the search result (the title link) mainly from it, but can also use the page's main heading, other headings and link texts that point to the page [2]. There is no fixed length limit, but Google shortens the title so that it fits on the screen.

• Good: unique for every page, describes exactly that page, starts with the most important thing. “Carpenter in Umeå | Nordic Bygg AB” says more than “Home”.
• Poor: the same title on all pages, a row of keywords one after the other, or vague words like “Start page” [2].

Which title actually gets the most clicks we test in our experiment lab instead of guessing (see How we measure).

Description: the text under the title#

<meta name="description"> is a short summary of the page. Google mostly creates the text in the search result (the snippet) from the page's own content, but uses the description when it describes the page better [3]. There is no length limit, but the same description on every page does not help. The description does not affect the position directly; it affects whether people click.

Robots, canonical and hreflang: what may be shown and which address applies#

• robots (noindex, nofollow …) says whether the page may be shown in search results. The default is that everything may be shown; when two rules conflict, the strictest one applies [1]. A forgotten noindex after a new website is one of the most common reasons why a page disappears from Google.
• canonical (<link rel="canonical">) points out which address is the original when the same content can be reached from several addresses. It is a strong signal, and the address should be written in full [4].
• hreflang says which pages are language versions of each other. Every version should list itself and all the others; otherwise the tags are ignored [5].

Viewport and character encoding#

<meta name="viewport"> says that the page is built for mobile, and the character encoding should be UTF-8 so that letters such as å, ä and ö always show correctly [1]. Both are small lines that matter a lot: without a viewport the page looks like a shrunken computer page on the phone (read more about mobile and speed in Speed).

Myths: tags that do nothing#

• meta keywords is not used by Google Search and affects neither indexing nor position [1].
• A perfect description does not “rank” the page. It makes more people click when the page already shows.

This is what we do#

• Every page we build gets its own title and its own description, a self-referencing canonical, the right language and viewport, and we check them in the finished page.
• When we move an existing website to us, we fix duplicate and missing titles and descriptions without changing the addresses (see Switch without losses).
• In the free analysis you see which pages lack or share a title and description.