The headline of an article is an essential component of your journalistic work and can determine how well the piece will be read. But does the number of words in the headline really matter? In the latest piece for our 'From Our Data’ series, this is exactly what we investigated.

We’ve done a lot of work on headline testing recently and between client-specific workshops and consultations and data investigations we’ve acquired a good sense of what makes a successful headline. Of course, it’s not possible to be prescriptive - a lot depends on the media brand, the topic, the timing, and so on - but winning headlines usually contain these active verbs:

Have, want, can, must, know

Powerful adjectives are also popular, such as:

First, big, new, best

Signal words guide the reader and therefore frequently appear in headlines that perform better:

This, how, what, why

In addition, sentiment words such as these also make for a successful outcome:

Happy, joyful, intense, angry, cruel, fatal

Editors often discuss the ideal number of words too. It can be frustrating when a content management system limits the number of characters. But sometimes there is also debate when an article with an extremely short headline performs exceptionally well, when conventional wisdom says the opposite. Is it really due to the headline's length? Let’s check what the data says.

The data research: 14 words is your average

Smartocto’s data team examined the number of words in winning headlines during a recent study of A/B tests, which allowed us to make useful comparisons.

(By the way… by ‘good performance’, smartocto doesn't just mean that the article is clicked on a lot. The system always considers loyalty in smartocto's feature, Tentacles. The system counts a click as a ‘loyalty click’ if the visitor stays on the page for at least 10 seconds afterwards, which is a useful indication that the article lives up to the promise of its headline.)

Want a deep dive on A/B tests? Check out this webinar. Or ask for a demo.

The smartocto labs department analysed nearly 30,000 winning headlines from the past year and categorised them into three groups. Below, you can see the categories and how many winning headlines fell into each category.

Here, the medium length category - those headlines between 12 and 17 words - is the standout. The average number of words that win is 14, as is the median. It would be easy to say that here's your answer: the optimal number of words in a headline is 14. But hold your horses, there’s a catch…

The average and the median of all losing headlines is also 14! The data team is adamant that you should look at other metrics in order to get a better, more rounded picture - and they should know. We’re not done yet…

Longer headlines perform better

It turns out that the average click-through rate (CTR - the percentage of visitors who click on the headline) for long headlines (more than 17 words) is the highest.

"You can draw all kinds of useful conclusions about that," says Rutger Verhoeven, co-founder and CMO of smartocto. "There may be more words in the headline that pique the curiosity of the visitor. Or it may be a type of article where the journalist has put in more effort, such as interviews with a quote. But still, the difference is significant."

What is noticeable as well, is that the average percentage of loyalty clicks is the highest for headlines containing more than 17 words. For headlines of medium length, it is the lowest. This means longer headlines perform the best, because a high CTR in combination with high loyalty, will eventually lead to better-engaged visitors.

"The real conclusion here is that you don’t need to be frugal when you consider a headline test with more than 17 words," says Rutger. "If a colleague says that your headline is too long, and visitors don’t have time to read it entirely, you can say with authority that the data doesn’t support that view. That said, we know that in some cases, a short headline can work very well too, so don’t make your headlines longer for the sake of it. It's not an absolute rule. We encourage you to think of this differently: find ways to examine hypotheses for your own brand. Try out different types of headlines in A/B tests and draw your own conclusions."

In a previous blog on headline testing, we gathered tips from media professionals on creating the best possible headlines. See below for the collected tips or read this blog for more context.