A headline is the line that persuades someone to read what you have written. In blogging there is not just one: there are three, they live in different places, and Google rewrites almost two in three of them before showing them.
This is the part missing from most guides on the subject. They teach formulas, odd numbers, power words and the maximum length in characters. Then the article goes live, a different title from the one you wrote shows up in the search results and nobody can explain why.
Below you will find what the data says about which titles earn clicks, how often Google replaces them, and a four-step method for writing them and then checking them in Search Console rather than by gut feeling.
Headline, title tag and H1: three titles, three jobs
Before you write, you need to know which of the three you are writing, because the constraints differ.
- The title tag is the text in the page code. It is the main source of the blue link in the search results and you manage it from your SEO plugin: on WordPress with Yoast it is the "SEO title" field, separate from the post title.
- The H1 is the large heading at the top of the page. It is read by people who have already arrived, and Google uses it as an alternative source when it decides the title tag is not good enough.
- The social headline is the one that appears on LinkedIn or in a newsletter, usually taken from the
og:titlefield. No character limit imposed by Google, no keyword to respect.
Google Search Central's documentation on title links lists nine sources the search engine can use to build the title you see in the results: the title element, the main visual title on the page, headings such as the H1, the og:title tag, other large and prominent text styled on the page, other page content, anchor text on the page, text within links pointing to the page, and WebSite structured data.
Nine sources mean one thing: the title tag is a strong suggestion, not an order. Anyone who treats the "SEO title" field as the only place where the user's view is decided is working on one channel out of nine. If you need the full picture on that field, we cover it separately in the guide to the meta title, which is worth reading together with the one on the meta description.
Google rewrites 61.6% of titles
In the first quarter of 2022 Zyppy, Cyrus Shepard's company, compared title tags written by site owners with the titles actually shown in the search results, on a sample of 80,959 title tags across 2,370 sites. Google rewrote 61.6% of them. The study is public: Google Rewrites 61% of Page Title Tags.
On its own the figure is a curiosity. It becomes useful when you look at the breakdown by length, because it overturns the advice repeated everywhere ("stay under 70 characters").
| Title tag length | Share rewritten by Google |
|---|---|
| 1-5 characters | 96.6% |
| 20 characters or fewer | over 50% |
| 51-60 characters | 39-42% (the least affected range) |
| over 70 characters | 99.9% |
Above 70 characters a rewrite is practically certain. Below 20 it happens to more than one title in two, which few guides warn you about: a short title is not a safe title, it is a title Google considers thin and replaces. The zone where your text survives is narrow, and it sits around 51-60 characters.
The study also measures what happens to individual elements. The vertical pipe is removed or replaced in 41.0% of cases, the dash in 19.7%. Square brackets get rewritten 77.6% of the time, with the bracketed content deleted in 32.9%. Round brackets hold up slightly better: 61.9% rewritten, content removed in 19.7%.
The fix Zyppy documents is simple and hardly anyone applies it: align the H1 and the title tag. When a number appears in both the H1 and the title, it is kept in 97.3% of cases. When it exists in only one place, Google treats it as optional.
The 80% myth comes from print advertising in 1963
The most quoted statistic in articles about headlines is this one: "8 out of 10 people read the headline, only 2 out of 10 read the rest". It appears on seo.com, on WordStream and on dozens of blogs.
The source is Copyblogger's ebook How to Write Magnetic Headlines, where the sentence appears exactly like this: "On average, 8 out of 10 people will read your headline copy, but only 2 out of 10 will read the rest". No source. No study, no sample, no methodology.
The closest thing to an origin is David Ogilvy, who writes in Confessions of an Advertising Man (1963) that "on the average, five times as many people read the headline as read the body copy" and that, once you have written your headline, you have spent eighty cents out of your dollar. It is an observation about adverts printed in American newspapers sixty years ago, not a measurement of blogs. And a ratio of five to one is not the same as eight to two.
If you read that number in an article, you know the author did not check it. It is worth knowing because it is the same mechanism we have found in other figures repeated from memory across the industry, collected in our round-up of SEO statistics.
What the verifiable data says instead
Backlinko analysed 4 million search results, precisely 1,312,881 pages and 12,166,560 queries, using Google Search Console data supplied by Semrush. The study is available at Google Organic CTR Stats. Three findings relate directly to the title:
- Title tags between 40 and 60 characters get an 8.9% higher CTR than other lengths.
- Titles with positive sentiment beat negative ones by 4.1% in CTR.
- Titles phrased as a question do no better than the rest: 15.5% against 16.3%, a difference within statistical noise.
The last point is interesting because the question format is recommended in a large share of headline guides. The data does not back it up.
Then there is the joint HubSpot and Outbrain study of 3.3 million paid headlines, whose results HubSpot summarised in the title guide on its own blog. Two numbers clash with common sense: headlines containing "how to" performed 49% worse than those without, and the word "tip" reduced CTR by 59%. In the same analysis, titles with a clarification in square brackets such as [Guide] or [Video] performed 38% better.
Let us call it what it is: these are data on paid ads and a US audience, collected on content discovery platforms, not on organic search results. They do not transfer automatically. They are useful for asking a question that few people ask.
Square brackets: two pieces of advice that cancel each other out
Put the two studies side by side and a contradiction appears that guides rarely mention, because they seldom cite both.
HubSpot says square brackets raise CTR by 38%. Zyppy says Google rewrites 77.6% of titles that contain them and deletes what is inside in one case in three. The advice works where the title reaches the user intact, meaning social media, newsletters and ads. In organic search it has a one-in-three chance of being taken apart before anyone reads it.
In practice: put the bracketed clarification in the H1 and in the og:title, where it survives. In the title tag, if you want it, write it without brackets.
Data from a real site: ranking does not buy the click
All the benchmarks quoted above are American. We looked at what happens on a real site that we manage.
The sample: the 100 keywords with the most impressions for an Italian industrial manufacturer we work with, technical B2B sector, almost all commercial queries. 7,338 impressions and 67 clicks in total over the period recorded by Search Console, cross-referenced with the position tracked on desktop.
| Position range | Keywords | Impressions | Clicks | Actual CTR |
|---|---|---|---|---|
| 1-3 | 25 | 3,909 | 50 | 1.28% |
| 4-5 | 20 | 1,662 | 14 | 0.84% |
| 6-10 | 29 | 976 | 0 | 0.00% |
| 11-20 | 12 | 349 | 0 | 0.00% |
| Beyond 20 | 10 | 358 | 2 | 0.56% |
| Total | 100 | 7,338 | 67 | 0.91% |
Backlinko's international benchmark gives the first position an average CTR of 27.6% and the top three results 54.4% of all clicks. Here the top three positions bring in 1.28%. Seventeen of the twenty-five keywords in the top 3 collected zero clicks despite having impressions.
The title alone does not explain that gap. Other factors weigh in: ads above the organic results, Shopping blocks on product queries, AI Overviews that answer before the user scrolls, and the fact that many impressions come from worse positions than the tracked one. It is the same effect we have measured elsewhere, in the article on zero-click searches.
But the practical conclusion does not change: ranking first no longer guarantees traffic, and of the elements that decide whether the click arrives, the title is the only one you can rewrite in thirty seconds. The audience is not the problem. The problem is the line of text you introduce yourself with.
The four-step method
1. Read the search results before you write
Search for your keyword and look at the top ten titles. Not to copy them: to understand which promise everyone is already making. If eight out of ten open with "How to", the ninth that opens with "How to" disappears. The job of a title in the search results is not to be pretty, it is to be different from the nine around it.
This step assumes you have already checked the search intent: a perfect title on the wrong intent recovers nothing.
2. Write the version for the reader, then cut it down for Google
First the H1, without counting characters: it must say precisely what the visitor will find. Then the 51-60 character version for the title tag, which is the range where Zyppy measures the fewest rewrites and Backlinko the most clicks. The two versions must share the main keyword and any numbers, otherwise you lose the 97.3% protection.
3. Put the specifics where the value is
"A guide to on-page SEO" and "On-page SEO: 12 checks ranked by impact" take the same effort to write. The second tells you how many checks there are and how they are ordered. The difference between them is not style, it is the amount of information per character.
The recurring formulas that keep working are few: the number with a unit ("in 3 weeks", "across 100 keywords"), the explicit constraint ("without an ad budget"), the stated audience ("for B2B sellers"), the result with a figure. Power words alone are not enough, and some cost you: "secret", "ultimate", "incredible" raise expectations and lower trust when the page does not deliver.
4. Write five variants and pick the least predictable
The first title that comes to mind is always the one that would come to everyone's mind, and statistically it is the one your competitors already have on the page. Five variants take ten minutes. If you did step 1, you already know which ones to discard.
How to check whether the title really works
This is the part that closes the loop and that most guides leave out: after publishing, how do you know whether the title did its job?
Three checks, all in Google Search Console, no paid tools.
- See which title Google shows. Search for your page in an incognito window on the main query. If the text in the results is not what you wrote, you are in the 61.6%: compare H1 and title tag and align them.
- Compare CTR and position for the same page. In the Performance report, filter by page and switch on CTR and average position. A page stable in position 4 with a CTR below that of its neighbours has a promise problem, not a ranking problem.
- Rewrite and note the date. Change only the title, record the day, wait four weeks and compare CTR at the same average position. If the position moves a lot in the meantime, the test cannot be read: run it again.
The third point is a single-variable test, not a true A/B test: in organic search you cannot serve two titles at the same time. That is why it should be done on pages with enough impressions to make the comparison meaningful, and why it is worth choosing them by looking at your SEO analytics data first.
Four mistakes we see on almost every company blog
- The company name at the start of the title. It takes up the characters Google shows first and the user reads first. Google itself, in its title link documentation, suggests the brand separated from the rest by a delimiter, not placed in front.
- The same title on different pages. The documentation names it explicitly among the causes of rewrites: repeated titles make it impossible to tell pages apart.
- The title written last. Written after the article, the title describes what came out. Written first, it decides what has to come out, which is why it pays to fix it in the brief together with the target length of the piece.
- The promise the page does not keep. A title that brings clicks to content that does not answer produces immediate returns to the search results. You got the click, not the reader.
If the company blog you look after has dozens of articles with these problems, the job is not to rewrite everything: it is to sort pages by impressions and start from the top twenty. It is the same criterion we apply in our SEO consulting projects, and we have told the full story of an industrial case in the article on SEO for manufacturing, where a keyword with 4,400 monthly searches moved from position 88 to 2.
Frequently asked questions
It depends on which of the three titles you are writing. For the title tag, the 51-60 character range is where Google rewrites least (39-42% of cases according to Zyppy), and Backlinko measures the highest CTR between 40 and 60 characters. For the H1 and the social title there is no technical limit: clarity and precision are what count. Above 70 characters the title tag is rewritten in 99.9% of cases.
Because the title tag is one of nine sources Google uses to build the blue link in the search results, along with the H1, the og:title tag, anchor text and other text on the page. The Zyppy study of 80,959 title tags found rewrites in 61.6% of cases. The most common causes are titles that are too long or too short, text repeated across pages, keyword stuffing and a mismatch between the title and the actual content.
No. The post title becomes the visible H1 on the page, while the SEO title, the field Yoast calls SEO title, generates the title tag in the code. If you leave that field empty, Yoast applies a template that usually repeats the post title plus the site name. Filling them in separately lets you have a longer H1 and a 51-60 character title tag.
Yes, but put them in both the H1 and the title tag. Zyppy measured that a number present in both is kept by Google in 97.3% of cases, while a number present in only one place is treated as optional. A number works when it says something verifiable, such as the number of items in a list or a measurement, not when it is only there to impress.
In search results they remain a common format for informational queries, so they are fine when the intent calls for them. However, the HubSpot and Outbrain study of 3.3 million paid headlines found that titles with how to performed 49% worse than the others, on a US sample and on content discovery ads. The reasonable reading is that the formula alone is not enough: what sets a title apart is specificity, not structure.
In Google Search Console, in the Performance report, filter for the single page and compare CTR and average position. A page with a stable position and a CTR below that of nearby pages has a title problem. To test, change only the title, note the date and compare CTR after four weeks at the same average position.
Something the industry rarely says
The best title for ranking and the best title for clicks often do not coincide, and for some time now there has been a third audience that looks at neither in the same way. ChatGPT, Perplexity and Google's AI Overviews do not click on a blue link: they read the page and report its content. For them, a title that describes precisely what is inside is worth more than one that sparks curiosity, because curiosity is not a variable a language model weighs.
Which leads to a counter-intuitive conclusion: the old advice to "write clear, descriptive titles", which for years sounded like the boring tip you give beginners, is now the one that pays on all three channels. You can spend curiosity on LinkedIn and in newsletters, where the title arrives whole and a person reads it. In search, where a system that can rewrite or summarise you stands between you and the reader, precision is the only thing that gets through intact.
