Measuring Online Reader Engagement

Contributed by: Mary-Katharine Phillips, Klaas Nelissen, Bart Baesens

This article first appeared in Data Science Briefings, the DataMiningApps newsletter. Subscribe now for free if you want to be the first to receive our feature articles, or follow us @DataMiningApps. Do you also wish to contribute to Data Science Briefings? Shoot us an e-mail over at and let’s get in touch!

This article also appeared on the blog of Twipe Mobile: Our research group collaborated with Twipe in the Digital Reader Engagement research project.

Can we measure and predict reader engagement? Twipe set out to answer this question, along with Mediahuis, iMinds Media, and two research groups from KU Leuven. In the two years since this Digital Reader Engagement (DRE) project began, we have learnt what happens in the body while reading news articles and turned this into our Google DNI-backed product EngageReaders which gives newsrooms actionable insights on a daily basis.

Recently, Professor Bart Baesens and PhD Student Klaas Nelissen, researchers at KU Leuven, sat down with us to discuss their findings and what lessons the media industry can learn from them.

Finding the treasures in your data

One key finding that both researchers wanted to highlight is how many in the media industry are ignoring, or not even aware of, the potential treasures hidden in the vast amounts of data they collect from readers.

The key message to the newspaper industry is to start looking at all the data you are gathering so you can better understand your subscribers’ reading behaviour.

– Bart Baesens, KU Leuven

Klaas Nelissen emphasized the need for the news industry to invest in understanding how people are reading in new digital formats. Instead of simply relying on page views equaling engagement, the DRE research project allowed for a clear understanding of what happens in the body while reading a digital newspaper, achieved through observing sixty readers while they read the tablet version of two Belgian dailies, De Standaard and Het Nieuwsblad. (An article discussing the findings of this experiment was recently published in the Journal of the Chartered Institute of Journalists.)

3 steps to get to know your readers

The first step in getting to know your readers is to understand where you are in terms of business analytics maturity. Take stock of your data awareness, do you collect reader surveys regularly? Have you instrumented your apps so that you have an understanding of reader behaviour?

Then in the second step, news organizations without much hard data already extracted from their digital publications or reader surveys can begin with qualitative experiments. By observing how your readers interact with your edition, you will begin to develop a general sense of your readers and their behavior.

Klaas Nelissen explained that during the qualitative portion of his research, observing the readers of two different newspapers clearly showed that there were clear differences between the two groups of readers. The differences were not only in how they talked about their experience with the newspaper, but also in broader ways, such as how they looked at the world in general.

I don’t need to know who you are. I only need to know how you swipe and interact with an article and where that article is located in the newspaper in order to know whether you’ll be engaged with the article.

– Klaas Nelissen, KU Leuven

Then news organizations can move to the third step, implementing a behavioural tracking aspect. This will give the newsroom clear insights to act on. To implement this behavioral tracking, it is important to first decide what level of analysis to conduct. Klaas Nelissen looked at how the level of analysis affects the research of reader engagement, moving from one reading session to an interaction with one article. By changing the level of analysis to one article, he was surprised to see that swipe actions were one of the most reliable predictor of engagement.

Ensuring privacy in data collection

Privacy is sure to be a growing concern in 2018, due in part to the looming General Data Protection Regulation affecting all companies collecting data of European Union residents. But this should not deter news organizations from developing data awareness and insights, thanks to Klaas Nelissen’s finding that swipe patterns are the best predictor of reader engagement. As such, behavioural tracking can serve as a privacy-preserving way to understand your readers.

Leading successful innovation

Tackling the question of digital reader engagement is no easy feat. Bart Baesens credits their success to two factors, the complementary team of researchers and Twipe’s dedication to innovation. By having a team of researchers that all had their own specific expertise, they were able to work together and cover more ground. As for Twipe’s dedication to innovation, he explained:

We were pleasantly amazed by the R&D commitment by Twipe and the way they keep on looking for new technologies and seeing how they can leverage those in their business projects. They are really cutting edge when it comes to analytics.

– Bart Baesens, KU Leuven

Next steps in digital reader engagement

The next stage in this multi-year research project is to link the findings from the biosensor data to specific swipe patterns. Klaas Nelissen will study how to connect the measurable impact of reading in the body with the swipe patterns of engaged readers, in order to predict reader engagement without needing to monitor what is happening in the body. By putting together the knowledge learned from how the body reacts when a reader is more engaged and the swipe patterns learned from the qualitative experiments, he hopes to be able to expand from small samples of readers to scaling their insights to potentially all readers of a newspaper, through behavioural tracking. We will be sure to keep you posted on future developments in digital reader engagement.