Titles, descriptions and snippets

Search results show a title link and a snippet for every page, and Google writes both from your page rather than copying your tags. Your markup is an input, not a command. The sections below answer the questions that come up most, in the order you meet them.

What does Google use to build a title link and a snippet?

For the title link, Google's documentation lists the sources it uses:

• the content of the <title> element
• the main visual title shown on the page
• heading elements such as <h1>
• the og:title meta tag
• other content that is large and prominent through style treatments
• other text on the page
• anchor text on the page
• text in links from other pages that point to it
• WebSite structured data

For the snippet, the description beneath, Google says snippets are primarily created from the page content itself. It sometimes uses the meta description when that gives users a more accurate description than text taken directly from the page, and it can show different snippets for different searches, because the best excerpt depends on the query.

The implication is that title, headline and opening paragraph should agree. If your <title> says one thing and your <h1> another, Google has to guess which to trust.

How long should a title or description be?

Google's documentation says there is no limit on how long a title element can be, but that the title link is truncated as needed, typically to fit the device width. It says the same of the meta description and the snippet. What fits depends on the device and the result layout, so there is no single character count that is always safe. Any exact limit you read is a rule of thumb from observation, not a published requirement. Treat it as a design target and nothing more.

Practical writing rules that follow from that:

1. Put the distinctive part first. If the end is cut off, the reader should still know what the page is.
2. Do not repeat the same keyword several times. Google says repeated words and phrases do not help the user and can make results look spammy.
3. Do not pad. A short, clear title is not "too short". Descriptions are different: Google's own examples include a two-word description, "Mechanical pencil", as too short to be useful.
4. Brand concisely. A short site name at the beginning or end, set off by a delimiter such as a hyphen, colon or pipe, is fine; a long tagline in every title wastes the visible space.

Why does Google show something different from what I wrote?

Google says that if it detects an issue on the page it may generate an improved title link from anchors, on-page text or other sources. The issues it names are:

Google presents these rewrites as automated efforts to represent the page better, not as a sanction, and the pages cited here do not describe them as penalties. They are still useful as feedback: if the generated title is better than yours, yours was unclear. Google also says that if you see a modified title link you should check your page for these issues, and otherwise consider whether the title in the results is a better fit for the query.

The French page with an English title

The language row matters for a bilingual site. A page written in French with an English <title> gives Google a reason to pick different text, and a French reader sees a mismatch in the results. Write each page's title in that page's language, and keep the proper nouns that really are the same. The related hreflang rules are in hreflang for a bilingual Canadian site.

What should I write instead?

A workable method, page by page:

1. State what the page is for. A title is a label, not an advertisement. "Setting up SPF records for a small domain" does the job; "Everything you need to know!" does not.
2. Match the headline. Keep the <title> close to the visible <h1>, differing mainly in the site suffix.
3. Make it unique. Two pages sharing a title compete for the same words and confuse readers who see them in a list.
4. Write the description as a summary. Google describes a good meta description as a short, relevant summary of what a page is about, like a pitch. Include the specific thing a reader will learn or get, not a slogan. Descriptions can also carry facts a snippet might otherwise miss, such as author or date.
5. Put the real answer high on the page. Because snippets mostly come from the page text, a clear first paragraph is also your snippet insurance.

Titles by page type

Mistakes that cost clarity

• Using the site name alone on every page, which gives search results nothing to distinguish.
• Putting the same text in the title, the heading and the description, which wastes the second and third.
• Generating titles from a template that appends a long tail to every page.
• Changing titles constantly. Google has to recrawl and reprocess the page to notice a change, which it says may take a few days to a few weeks, so frequent edits blur any comparison.

A before and after (fictional)

A fictional gardening-supplies page, example.com/soil/ph-testing/:

<!-- Before -->
<title>Home</title>
<meta name="description" content="Welcome to our website. We offer many products and services.">

<!-- After -->
<title>How to test soil pH at home | Example Garden Supply</title>
<meta name="description" content="Three ways to test soil pH at home, what the readings mean for vegetables and lawns, and how often to retest.">

The second pair says what the page contains. The opening paragraph on the page should echo the same promise, so that whichever source Google picks, the result is coherent.

Finding duplicate titles before Google does

On a built static site, a one-line pass over the generated HTML lists every title that appears more than once:

find dist -name '*.html' -exec grep -o '<title>[^<]*</title>' {} \; | sort | uniq -c | sort -rn | grep -v '^ *1 '

On a fictional six-page test folder where three pages were titled "Home", it printed a single line, 3 <title>Home</title>, and the other three titles were unique. That is half the pages carrying one title, which is the pattern Google's "micro-boilerplate" row describes. The command only sees titles on one line and misses pages whose title spans lines, so it is a first filter, not a full audit.

Can I limit or remove a snippet?

Meta tags let you limit what is shown, which is useful for paywalled, sensitive or partly private pages. Use one of the first two lines, not both, because Google applies the more restrictive rule when they conflict:

<meta name="robots" content="max-snippet:160">
<meta name="robots" content="nosnippet">
<p>Public text. <span data-nosnippet>This part will not be quoted.</span></p>

max-snippet: sets an upper bound on snippet length in characters (0 means no snippet, equivalent to nosnippet, and -1 lets Google choose), nosnippet removes the text snippet completely, and data-nosnippet excludes marked sections of a page. Google's specification says nosnippet and max-snippet apply to its AI Overviews and AI Mode as well, limiting how much of the content may be used as a direct input. A static thumbnail may still be shown under nosnippet when it results in a better experience.

data-nosnippet has its own constraints. It works on span, div and section elements, any value is ignored, and the HTML must be valid with every tag closed: an unclosed div marks everything after it. These are instructions to Google, not security: content that must stay private belongs behind a login.

How do I check what will actually show?

You cannot preview Google's rewrite exactly, but you can check the inputs and observe the outputs.

1. Check the inputs. From a terminal, confirm that the title and description exist in the HTML your server sends, without JavaScript:

curl -s https://example.com/soil/ph-testing/ | grep -i -E '<title|name="description"|<h1'

2. Check indexing. In Search Console, use the URL Inspection tool on the page to see the indexed version, and the Performance report to see which queries produce impressions and what the click-through rate is.
3. Compare after changes. Rewriting a title can take time to appear because Google must recrawl the page. Change one batch of pages, wait, then compare click-through rate for those pages against similar untouched pages. A single week of data on a new site is noise.
4. Search for it. Search for the page's actual topic in a private window and read the result as a stranger would.

Make sure the page is crawlable. Google warns that if it cannot access a page's content, it relies on off-page content such as anchor text from other sites to build the title link. To keep a page out of results use noindex, not a robots.txt block; sitemaps, robots.txt and Search Console covers the files involved.

Where does structured data come in?

Titles and snippets describe a page; structured data describes the things on it. Google's site name documentation says WebSite structured data on the home page is the most important signal if you want to state a site name preference, and its title documentation lists WebSite structured data as one source for title links. Keep the two consistent. Details are in structured data for a person and a small company.

Keep this

• Give every page one unique <title> that states what the page is.
• Keep the title, the main heading and the opening paragraph consistent.
• Write the description as a short summary; do not rely on a character count.
• If Google rewrites a title or snippet, read the rewrite as feedback on clarity.
• Verify the tags in raw HTML with curl, then watch Search Console rather than guessing.