Google ignores your meta title when its systems decide that another piece of text on the page describes it better than the <title> tag you wrote. The tag stays in the code and still counts for ranking, but the snippet shows something else: the H1, a prominent piece of text, the anchor text of a link pointing to the page, or a shortened version of your own title.
It is not a penalty and it is not a technical error to fix. It is how title generation normally works, and it affects meta descriptions even more. Since 24 August 2021 Google has built the title links in its results in a different way. What follows is why it happens, how often it really happens on Google UK and how to tell whether the case in front of you is worth acting on.
Google does not ignore the title: it puts it up against eight other sources
The official documentation on how to influence title links lists nine places Google can take a result's title from. The <title> tag is the first, but it is one of nine.
- the content of the
<title>element - the main visual title shown on the page
- heading elements, starting with
<h1> - the content of the
og:titlemeta tag - other large and prominent text styled on the page
- other text contained in the page
- anchor text on the page
- text within links that point to the page
WebSitestructured data
When Google's systems pick one of these sources instead of your tag, the title in the SERP changes for every query, not only the one you are looking at. It is a choice tied to the page, not to the search: Danny Sullivan made that clear in the September 2021 post on the Search Central Blog.
The same post contains the figure everyone quotes and hardly anyone puts in context: after the corrections that followed the SEO community's complaints, Google says it uses the <title> element around 87% of the time, against 80% with the previous system. It is a global aggregate figure stated by Google, not measured by third parties.
How often it really happens: three independent studies, three very different numbers
Those who measured the phenomenon from the outside got percentages ranging from 33% to 63%. The spread does not come from the data but from what each one counted as a rewrite.
| Source | Sample | Market | Titles not shown as written |
|---|---|---|---|
| Google (Danny Sullivan), Sept 2021 | internal data, not published | global | around 13% |
| Ahrefs (Michal Pecánek), 2021 | 953,276 top 10 pages | English-language SERPs | 33.4% |
| Zyppy (Cyrus Shepard), Q1 2022 | 80,959 titles across 2,370 sites | United States | 61.6% |
| Portent, 2021 | 1,469,373 URLs from 32,000 keywords | United States | 63% (67% desktop, 59% mobile) |
Ahrefs treated a title as unchanged when the differences were minimal, for example punctuation or the addition of the brand name. Portent counted any different string as a rewrite, truncations included. Zyppy worked on its own crawl compared against the site: operator.
So 33% is the share of titles replaced in substance, and 63% the share of titles that appear different in any way in the SERP. They are two different questions and worth keeping apart, because only the first calls for action.
None of these studies was run on UK SERPs specifically, which is the limit that comes back every time you handle SEO statistics produced elsewhere. To see what happens here we repeated the measurement on a small sample, but one anyone can check in ten minutes.
What happens on page one of Google UK
On 26 September 2026 we took the organic results on page one of Google UK for the query seo agency london (DataForSEO, Google UK, searched from London, desktop), downloaded the source code of each page and compared the <title> tag with the title actually shown in the results.
Two of the nine organic results could not be compared: a Reddit thread blocks third-party crawlers, and one agency blog (elitedigitalagency.io) returns a "page not found" to them. That leaves seven.
| Site | Title tag in the code (characters) | Title shown in the SERP | Difference |
|---|---|---|---|
| Hydra Creative | SEO London | SEO Agency London | Hydra Creative (47) | SEO Agency London | repeated keyword and brand removed |
| The iBoost | Search Engine Optimization (SEO) Agency in London (49) | identical | none |
| SINE Digital | AI SEO Agency London & New York | GEO, AEO & AI Search Optimisation | SINE Digital (82) | AI SEO Agency London & New York | cut to the first segment |
| Fast Generations | SEO Agency London | Fast Generations (36) | identical | none |
| Cash Cow Marketing | How to Choose a London SEO Agency: 7 Things to Check Before You Sign - Cash Cow Marketing - Bespoke SEO & Digital Marketing (123) | How to Choose a London SEO Agency: 7 Things to Check … | truncated only |
| Chilliapple | eCommerce SEO Agency London UK | eCommerce Digital Marketing company (68) | eCommerce SEO Agency London UK | second segment removed |
| Pure Marketing | SEO London | London SEO Agency Without The Price Tag | Free Audit (65) | London SEO Agency Without The Price Tag | Free Audit | opening keyword segment removed |
Two pages out of seven show the title exactly as it is written in the code. One is only truncated because it runs far past the space available. The other four reach the SERP modified.
The 71% of our UK sample shown differently is in line with Zyppy's 61.6% and Portent's 63%, and with the 67% (six of nine) we measured on the same kind of query on Google Italy a few days earlier. With seven pages the margin of error is enormous and we do not draw a national statistic from it, but the order of magnitude holds: on most page-one results, Google shows something different from what the SEO wrote.
What Google cuts most is the repetition
In the UK sample Google never went outside the title: in every modified case it kept one piece of the tag and dropped the rest, and the piece it dropped was nearly always the one that repeated the keyword.
- Hydra Creative opens with "SEO London | SEO Agency London". Google keeps the fuller variant and removes both the shorter one and the brand, in a title of only 47 characters.
- Pure Marketing opens with the same "SEO London |" segment. Google removes it and starts the title from "London SEO Agency".
- Chilliapple says "eCommerce" twice. Google keeps "eCommerce SEO Agency London UK" and drops "eCommerce Digital Marketing company".
- SINE Digital has three segments in 82 characters. Google keeps the first and drops the list of acronyms and the brand.
The brand name is the other thing Google handles on its own terms. In the UK sample it disappeared from three titles out of seven (two removals and one lost to truncation) and survived in a short 36-character title. On Google Italy the pattern was even stronger: five of the nine differences concerned only the company name, four times removed, once added. Google has its own idea of where the site name goes and how much space it gets, and it applies it regardless of how long your title is. Fighting on this front is wasted time: the official guide asks you to brand titles "concisely" and leaves the final decision to Google.
The case that deserves attention came from the Italian test. One agency's title tag was 63 characters with four variants of the same keyword. Google discarded it entirely and showed the page's H1. That is not a normalisation, it is a rejection.
The six reasons your title does not make it to the SERP
The title is too long
This is the best-documented cause. In the Zyppy study, titles over 70 characters are modified in 99.9% of cases, and those over 60 characters in more than 76%. The band least likely to be rewritten is 51-60 characters, where the figure stops at 39-42%. Ahrefs measures the same thing in pixels: titles over 600 pixels went from 29.45% to 46.12% rewritten after the 2021 update, an increase of 56.6%.
In our UK sample the two titles over 80 characters (82 and 123) are the two that lose the most text. The details on length, with the pixel rules, are in the meta title guide.
The title is too short
The opposite figure is less well known and harsher: again according to Zyppy, titles of 1 to 5 characters are replaced in 96.6% of cases, and below 20 characters the rate is still over 50%. A two-word title does not give Google enough material, so Google goes and gets it elsewhere.
The same keyword repeated in variants
The Hydra Creative and Pure Marketing cases in the table: "SEO London" followed by "SEO Agency London". At the extreme, as in the Italian case above, Google treats the title as keyword stuffing and falls back to the H1. The official documentation says it plainly: avoid keyword stuffing and repeated or boilerplate text in <title> elements.
The title does not describe that page
Danny Sullivan listed four recurring patterns, all produced by badly configured templates rather than editorial choices.
- Half-empty titles: the template should build "Product name | Site name" but the first variable arrives empty and the SERP is left with "| Site name".
- Obsolete titles: the page says "2026 admissions criteria" in the H1 and "2025" in the title because nobody updated it. Google uses the date in the headline.
- Inaccurate titles: a static title on a page with dynamic content, such as "Giant stuffed animals, teddy bears, polar bears" when those products are sometimes not there.
- Micro-boilerplate titles: the same title repeated across a subset of pages, for example every season of a TV series with exactly the same title.
Titles duplicated across dozens of URLs are an on-page SEO problem before they are a snippet problem, and they show up clearly in a site crawl.
The title language does not match the page language
Hardly anyone mentions this cause, and it is the one that most often hits multilingual sites. In June 2022 Google introduced an algorithmic improvement that identifies documents where the title element is written in a different language or script from the content and looks for an alternative title that matches the main language of the page.
The two patterns affected are the bilingual title, meaning the same phrase repeated in two languages separated by a dash, and the transliterated title. On a site run with WPML or a similar translation system it happens in a mundane way: the English version of a page inherits a title in another language, or a UK organisation with an English and a Welsh version puts both languages into every title. If you are working on an international project, the check belongs alongside the hreflang check.
Emoji, symbols and decorative separators
In our Google Italy test one agency opened its title with a lightning bolt emoji. In the SERP it was gone, and with it the brand. The emoji was also the only element that made the title "unusual" enough to trigger a rewrite on a title of just 38 characters.
Rewrite or normalisation? The distinction that changes the priority
Take an example outside the agency sample. Macropix, a Milan-based LED screen manufacturer we have worked with for years and whose SEO journey we have described in detail, ranks third organically on Google Italy for "ledwall". Its title tag is "Led Wall: Cosa sono, Dimensioni e Installazione | Macropix", 58 characters. The title Google shows is "Led Wall: Cosa sono, Dimensioni e Installazione - Macropix".
The only difference is the separator: the vertical bar becomes a hyphen. Technically it is a rewrite, and in the Portent and Zyppy studies it would be counted. In practice there is nothing to do and nothing to lose.
It is worth pointing out because the advice to avoid the vertical bar in titles to reduce rewrites has been going round for years. In our UK sample the bar appears in five titles: one stays intact (Fast Generations) and four are cut. But in all four Google cut at a separator and kept whole segments: the bar is where Google trims, not the reason it trims. The separator, on its own, does not seem to decide anything.
The rule we use: if the title shown still says what you meant to say, leave it. If it says something else, or says the same thing but worse, then you have a CTR problem to measure.
How to check it on your site in three steps
Search Console does not show the title Google displays in the results: it reports impressions, clicks and position, not the snippet text. The comparison has to be built by hand, and it takes three operations.
- Extract the title tags. A crawl of the site with any tool that reads the source (Screaming Frog, Sitebulb, the Ahrefs or Semrush site audit) gives you the column with the real titles.
- Extract the titles shown. A
site:yourdomain.co.uksearch shows the snippets as Google sees them. For a large site, use a tool that queries the SERPs via API: comparing by hand works up to a few dozen URLs. - Cross-reference the two lists and sort by traffic. The rows that matter are those where the title shown differs substantially and the page brings in clicks. A rewrite on a page that generates no traffic is not a problem to solve.
At that point open Search Console, isolate the URL and compare the CTR for the months before the rewrite with the months after. If CTR is stable, the new title works as well as yours. If it has dropped, it is worth rewriting the tag.
Bear in mind that organic CTR is also affected by what appears above you in the SERP, from AI Overviews to featured snippets. Blaming a drop on the title alone without looking at how the results page has changed is the most common way to get the diagnosis wrong.
What to do when the rewritten title is worse than yours
There is no tag that forces Google to use your title. There are, however, levers that work in our experience, in this order.
- Align title and H1. When Google replaces a title, it uses the H1 in 50.76% of cases (Ahrefs data, 953,276 pages). If the two texts say the same thing, the replacement stops doing damage.
- Remove the repetition. One occurrence of the main keyword, one brand name, no list of synonyms.
- Stay between 40 and 60 characters. It is the band least likely to be touched in all three studies cited.
- Write the title for the question, not for the tool. If the title does not answer the search intent of the query the page ranks for, Google looks in the text for something that does.
- Report the case. Google keeps a public feedback thread on the titles shown in results, linked from the official documentation. It is not a fast channel, but it is the only official one.
What does not help: putting the title in capitals, changing the separator, adding the year, repeating the brand. None of these produced a measurable difference in our samples.
If the problem affects dozens of templates rather than a handful of pages, it calls for a technical SEO fix to the system that generates the titles, not manual rewriting. It is one of the checks we run at the start of every SEO project, because on an ecommerce site with thousands of URLs one faulty template is worth more than any page-by-page optimisation.
Frequently asked questions
No. The title tag is still read and still counts for matching the page with keywords, even when a different title appears in the SERP. The replacement only affects the text shown in the snippet, not how Google interprets the page.
Between 40 and 60 characters. According to the Zyppy study of 80,959 titles, the 51-60 character band has the lowest rate of changes (39-42%), while over 70 characters it reaches 99.9% and under 20 characters it exceeds 50%. The data comes from US SERPs.
Because it decides for itself where to place the site name and how much space to give it. In our test on seven page-one results for "seo agency london" on Google UK, the brand disappeared from three titles; on Google Italy, five of nine differences concerned only the brand. On short, well-written titles there is nothing to correct.
The title changes at the next crawl of the page, so anywhere from a few hours to several weeks depending on how often Google visits that URL. Requesting indexing in Search Console speeds up the check but does not guarantee the new title will be used: if the reason for the replacement remains, so does the replacement.
Yes, and to a greater extent: Ahrefs measured that meta descriptions are rewritten in 62.78% of cases. The difference is that the description is rebuilt according to the query, so the same page can show different text for different searches, while the title stays the same.
There is a consequence that gets little attention. Now that SERP titles are largely written by Google, the <title> tag has stopped being the text the user reads and has gone back to what it was in 2005: a signal for the machine. The text the reader sees is now the result of a choice among nine sources, and the sources Google prefers when the title does not convince it are the H1 and the body of the page. Of the two elements, the one that deserves more editorial attention today is no longer the title: it is the H1.
