What Makes You Want To Click This?

A University of Bristol study tells us the biggest rule of clickbait: No one wants to read about finance.
Courtesy University of Bristol

It’s the holy grail of Internet journalism: The click. How can I persuade you, in a few capitalized words, that this article will be worth the load time?

According to a new study from researchers at the University of Bristol’s Intelligent Systems Laboratory, the answer lies in using “sentimentally charged language” and skewing toward entertainment and crime.

The study looked at which stories were placed in the “Top Stories” and “Most Popular” feeds on 14 different news sites, including the BBC, NPR, The New York Times and Forbes. The researchers created models for the reader preferences of each publication based on an 18 month period, by looking at which stories were published on the home page (the top stories) and comparing that to what stories readers actually clicked on (most popular).

The “global appeal” of news articles increased when the language they used was more subjective–tabloids were found to be more appealing than the BBC, for example. Articles that were about “public affairs”–i.e. elections, the economy, business and politics–didn’t perform as well generally across all audience models as “non-public affairs” articles–those covering those always-fun topics like disasters, crime, fashion and weather.

There are a couple shortcomings to note: The models were based on articles’ headlines and short descriptions, but didn’t take into account other determining factors, like how great/terrible the stock photographs were on the articles. And since it couldn’t take into account the various demographics of readers of specific articles, the study had to assume that each publication’s readers would behave in the same way.

While this could prove fairly useful for SEO junkies, even if we knew everyone in the world would click on “31 Robotic Cats You Didn’t Know Were Sexting,” there’s no guarantee those people would actually finish the reading.