
The main purpose today fulfills headlines on the Internet: they should motivate people to click on it. An analysis of around 40 million English -language headlines from the past two decades shows that digital media formulate the titles of their articles more and more. Among other things, the headlines have become longer and more negative and contain frequent question words. This development is independent of the journalistic quality of the media. From the researchers’ point of view, however, it makes it more difficult for users to distinguish serious from dubious contributions.
It has never been as easy, fast and cheap to spread journalistic content as in the digital age. The Internet has become a huge marketplace on which the media compete for the attention of the audience. The equivalent to the calls of the market screens are the headings of online articles. You decisively decide whether an article is clicked or not. Publishers can easily determine how successful each individual heading is with the help of common analysis tools – and adapt their strategy if necessary.
New requirements and opportunities
But how did the online competition influence the design of the headings? To find out that, a team around Pietro Nickel from the Max Planck Institute for Education Research analyzed around 40 million headlines that have published English-language media since the early days of the World Wide Web in the early 2000s. With machine help, they evaluated the length, the emotional tonality and the set structure of the titles. “Our analysis shows that the language of online headlines has systematically changed over the years,” says Nickl. “Many of these changes indicate adaptation to the requirements and possibilities of the digital environment.”
The researchers found that the headings became more and more negative over time. This coincides with the results of earlier studies, which according to which headlines generate more clicks if they contain negative words. The length of the titles has also increased over time. In the early days of online messages, the headings were usually short, as is also common in the print area due to the limited space. “The digital format has the compulsion to express itself briefly and flush,” explain the researchers. Instead of short nominal sentences such as “Earthquake in Myanmar”, there were increasingly complete sentences with active verbs. Other linguistic means that are typical of clickbait also increased, including pronouns and question words. In this way, the headings should arouse curiosity without revealing too much.
Clickbait also with quality media
“The changes are not the result of individual editorial decisions, but an expression of a cultural selection process,” explains Nickl. “Certain linguistic characteristics prevail because they are more successful under the conditions of digital attention economy. They are used more often – if necessary, even without producers or consumers of these mechanisms.” According to the researchers, the recommendation algorithms of social networks can also contribute to this.
The headings evaluated for the study came from four international news portals-New York Times, The Guardian, The Times of India and ABC News Australia-as well as from the “News On The Web” Corpus, which contains around 30 million headlines from different countries. The researchers also included the title of the online portal Upworthy as an example of explicit clickbait. It showed that the headlines of quality media have become more and more similar over time. From the research team’s point of view, this development can lead to problems: “If the style of established media approaches those of problematic sources more and more, the limits are blurred-and this also makes it difficult to differentiate between reputable and manipulative content,” warns Nickl’s colleague Philipp Lorenz-Spren.
In order to create incentives for contributions that not only generate clicks, but actually provide valuable information, according to the researchers, it could make sense to establish alternative success criteria. The first approaches are already available: “The Guardian is now showing a list with the title ‘Deeply Read’ on his homepage in addition to the list of the most clicked articles,” report the researchers. This list includes contributions with which readers have been working for a particularly long time in relation to the length of the contribution – a possible reference to the quality of the article.
Source: Pietro Nickl (Max Planck Institute for Education Research, Berlin) et al., Humanities and Social Sciences Communications, DOI: 10.1057/S41599-025-04514-7
