Title tag SEO: nine sources Google can build your headline from

Title tag seo is the work of writing one element that Google reads first and can still overrule. Google publishes nine sources it may build a search result headline from, and your <title> is only the first of them. Write one distinct, front-loaded title per page and it usually survives; leave a template in there and something else on the page gets promoted in its place.

Before you start

This chapter is about the line people read before they decide whether your page exists. The line underneath it — the two sentences of grey text — is a different element with different rules, worked through in how to write a meta description. If you want the numbers on what real titles look like in the wild before you touch yours, we read meta title length across 27 homepages and found a median of 40 characters and no site whose title matched its own h1.

What title tag seo controls, and what it does not

It controls the input. It does not control the output. Google states the split without hedging: "Google's generation of title links on the Google Search results page is completely automated and takes into account both the content of a page and references to it that appear on the web." Your title element is the strongest input into that process and the only one you write on purpose.

Two consequences follow, and both surprise people. The first is that a rewritten headline is not a penalty; it is the documented behaviour when Google detects one of seven specific problems. The second is that there is nothing to appeal to: "While we can't manually change title links for individual sites, we're always working to make them as relevant as possible" (Influencing your title links in search results, read 2026-09-09).

You do not write the headline. You write the strongest candidate for it.

The nine sources Google can build a headline from

The documentation names all nine, in order, and the ordering is the useful part — the further down the list Google has to go, the less of the result was written by you.

Sources 1 to 4 are four separate fields that should agree with each other, and on most sites they do not. Source 8 is the one to notice: when a page is blocked from crawling, the documentation says that "we will rely on off-page content to generate the title link, such as anchor text from other sites" — other people's words become your headline. Which is also why how to write anchor text is a title problem as much as a link problem.

Do it: five steps

Each step ends in something you can look at. Step one is a measurement and it usually finds the problem on its own.

1. Pull the four fields for every important page. Title element, first h1, og:title, and the character count. The command is below. Done when you have one row per page and can sort them.
2. Sort the titles and look for duplicates. Repeated text is one of the seven documented triggers: "Avoid repeated or boilerplate text in title elements." Done when every row is distinct in its first 40 characters, not just somewhere in the middle.
3. Move the distinguishing words to the front. The tail is what gets cut — there is "no limit on how long a title element can be", but "the title link is truncated in Google Search results as needed, typically to fit the device width". Done when the first 40 characters alone tell you which page it is.
4. Make the title and the visible heading agree. Not identical, agreeing: same subject, same date, same season number. Done when no page has a title element naming a different thing than its own h1.
5. Cut the site name where it repeats. Google "may omit the site name from the title link, if it's repetitive with the site name that's already shown in the search result", so a suffix on every page is spending characters twice. Done when the brand appears once, at the end, or not at all.

What to ship: the audit command and the seven-symptom table

Paste your important URLs into a file and run this. It reads what a crawler reads, not what your CMS thinks it published.

# One row per page: length, title, h1, og:title
while read -r u; do
  h=$(curl -sL --max-time 20 -A 'Mozilla/5.0' "$u" | tr '\n' ' ')
  t=$(printf '%s' "$h" | grep -oiE '<title[^>]*>[^<]*' | head -1 | sed 's/.*>//')
  o=$(printf '%s' "$h" | grep -oiE '<h1[^>]*>[^<]*' | head -1 | sed 's/.*>//')
  g=$(printf '%s' "$h" | grep -oiE 'property="og:title" content="[^"]*' \
        | head -1 | sed 's/.*content="//')
  printf '%3d\t%s\t%s\t%s\n' "${#t}" "$t" "$o" "$g"
done < urls.txt

# Then find the repeated ones
cut -f2 out.tsv | sort | uniq -d

Google documents seven situations where it replaces the title, each with what it does instead. This is the table to read your audit output against.

A worked example of the last row is on the documentation page itself: its title element runs 128 characters and names the site three times, while its visible h1 is a plain six-word sentence. Measured on 2026-09-09 with the command above.

Three ways this goes wrong

All three pass a validator and look fine in a browser tab, which is why the audit reads the rendered HTML rather than the template.

1. Optimising for a character count that does not exist. There is no documented limit, only truncation at the device width, so a rule like "keep it under 60" is a proxy for a real constraint rather than the constraint. What matters is which words are in the first half, since that half is what survives on a phone.
2. Writing the title for the search engine and the h1 for the human. Two different sentences about the same page is exactly the inconsistency Google's process looks for, and when it finds one it picks — often the h1. You have then written the title twice and shipped neither.
3. Keyword stuffing that reads as spam. The documentation is blunt about the shape and the consequence: title text like "Foobar, foo bar, foobars, foo bars" "doesn't help the user, and this kind of keyword stuffing can make your results look spammy to Google and to users."

Where this stops

This chapter can get your title element into the best shape the documentation describes. Four things it cannot do.

1. Guarantee your title is the one shown. Generation is automated and there is no override, so the honest goal is to remove every documented reason for a rewrite, not to win an argument.
2. Tell you how often yours is being rewritten. No report exposes this. Searching for your own pages and reading the results by hand is the only method we know of, and it does not scale past a few dozen pages.
3. Move a page up. Nothing in the documentation connects title text to position, and we have run no test that separates the title from everything else that changes when you rewrite a page. What a better title changes is whether the result gets chosen.
4. Say what AI answer surfaces do with it. They cite pages by name, and none of them publish which field that name comes from. We do not know, and treating the title element as the source is a guess we are not making for you.

If the underlying problem is that a page is not appearing at all rather than appearing under the wrong headline, the title is downstream of that, and the gap diagnosis starts at the right end.

Frequently asked questions

How long should a title tag be?

There is no limit in the specification and none documented by Google. The result is truncated "typically to fit the device width", so the practical rule is that your first 40 characters have to work alone. Everything after that is a bonus that half your readers will not see.

Should the title tag and h1 be the same?

They should agree, and they usually should not be identical. The h1 sits inside a page the reader has already opened; the title has to work in a list of ten results next to competitors. Identical is safe. Contradictory is the failure case.

Why is Google showing a different title than mine?

Because it detected one of the seven documented issues, most often a half-empty template, a repeated title, or a title that does not describe the page. Run the audit above, find your page's row in the symptom table, and fix that specific thing. There is no appeal process.

Does the title tag still matter for ranking?

Google does not publish a ranking claim for it, and neither will we. What is documented is that the title link is the headline of the result and that Google builds it from your title element first. So title tag seo pays off in whether the result gets chosen, which is a click decision rather than a position claim.

Do I need og:title as well?

It is the fourth source on Google's list and the one social platforms read, so a mismatch between it and your title element gives the two surfaces different headlines for one page. Setting it to the same string is the cheapest way to keep them in agreement.