EngageReaders – BIGGIES Award-winning ePaper Analytics technology

Q&A with Danny Lein, CEO, Twipe Mobile

Big Data & AI for Media:

Why did you build EngageReaders? What is the media industry’s ‘problem to be solved’?

Danny Lein:

Many newsrooms these days are organised in terms of speed: there is the center of a newsroom with an online team, directly connected with its audience and on the other end there is the large team of journalists and editors in charge of the publication of a daily newspaper or weekly magazine. 

While the first team work on a stream of news and have extensive insights through analytics, engagement measurement tools and success on social media, the latter team work as they have been working for over a century: completely navigating in the dark on how the audience perceives and engages with their product. 

To bridge this gap, we initiated the Digital Reader Engagement project in cooperation with Mediahuis, publisher of 6 newspapers including De Standaard, NRC and De Telegraaf. The goal was to define, measure, and predict engagement of readers with their digital newspaper and to translate those measurements to key insights that can be used in the newspaper production process (both print and digital).

Today, EngageReaders is helping editors learn how to improve the engagement levels of both their print and digital newspaper publications based on scientifically grounded insights. Currently it is used in 10 newsrooms, and improves the engagement of editions such as Le Monde, Ouest-France, Aachener Zeitung, Berliner Zeitung, and many more.

Big Data & AI for Media:

What are the time frame and milestones for EngageReaders?

Danny Lein:

EngageReaders was developed in five phases, with the final phase consisting of ongoing improvements.

Phase 1: Scientific Research (2014 – 2015)

We started in November 2014 with a scientific research project in conjunction with the University of Leuven, the innovation hub Imec, and Belgian publisher Mediahuis. We studied panels of readers with sensors, tracking device interaction, posture, pupil movement, heart rate, blood pressure, and asked users for feedback on content. After all this data was gathered, an extensive analysis of all the user reactions and interactions was conducted. Prototype dashboards were developed and tested together with the newsrooms of De Standaard and Het Nieuwsblad.

From this research, the key learning was that time and positive affect are key metrics for understanding reader engagement with digital newspapers. Positive affect is influenced by content, composition, reading comfort and aesthetics of the digital edition.

Phase 2: Product Development (2015-2016)

With the support of Google DNI, we started to develop a Minimum Viable Product (“MVP”). We focused on three areas: measuring reader engagement, measuring and predicting attention time, and measuring positive affect. In close collaboration with the newsrooms of De Standaard and Het Nieuwsblad, we finetuned the prototype dashboards towards production ready tools.

During this process, we ensured that EngageReaders could be made available to any publisher. The technology had to be easily integrated within the apps and websites of publishers, not only in the apps of Twipe. This resulted in an open EngageReaders SDK and API.

Finally, we developed the first predictive model for reader engagement on digital editions. We are able to predict how much time readers will spend on articles, given the length, composition and structure of the publication. This predictive model gives the newsroom deep insight into over- and underperforming articles. It helps in making decisions on investing in content and improving the composition of the daily editions.

Phase 3: Pilot at Het Nieuwsblad (2016)

Toward the end Phase 2, we ran a 4-month pilot with one newspaper from Mediahuis, called Het Nieuwsblad–with a print circulation of close to 300.000 copies, this is one of the largest newspapers in the region. The purpose was to gather feedback on the tools we wanted to develop directly from the newsrooms. We asked the editorial team to integrate it into their daily routines, and then we gathered feedback on the usefulness of the tool. This feedback was integrated in the final version of the dashboards and tools.

Phase 4: Roll Out to Other Newsrooms (2017-2018)

Once the pilot phase was over, we aided Mediahuis in rolling out EngageReaders to additional newspaper titles over the next two quarters.

We launched last year in five newsrooms: Aachener Zeitung (Germany), Het Nieuwsblad (Belgium), De Limburger (The Netherlands), L’Avenir (Belgium), and La Montagne (France). This year we’ve continued to roll out EngageReaders in more newsrooms, including Le Monde, Ouest-France, DuMont, and MediaPrint.

Phase 5: Ongoing Improvements (2017-…)

Since we launched, we continue to make improvements to the product, based on our bi-annual partners meetings, where participating newspapers can discuss and vote on what new features are important to them. We foresee this phase lasting indefinitely so that we can always be improving and adapting EngageReaders to the real needs of its users.

Big Data & AI for Media:

What do the team and resources look like?

Danny Lein:

The team for each phase was slightly different, but for the main development the team was composed of two JavaScript developers, one Android and iOS developer, one big data expert, one data scientist, and one project manager. In addition, during the first phase we worked with researchers from the University of Leuven and Imec using Mediahuis’s newspaper De Standaard as the use case. In the latter phases we worked with the team from Mediahuis as well, with a dedicated Project Manager and Systems Architect, which allowed for a smooth technical integration within their infrastructure.

Our project benefited from co-funding from both Google’s Digital News Initiative and the Flemish government. In total this funding amounted to 1.5 million euro, with additional funding from Twipe and Mediahuis.

Big Data & AI for Media:

What are the challenges you faced?

Danny Lein:

The key challenge we faced was the issue of data aversion within newsrooms. This meant that it was difficult both in the development of the product, when trying to understand what metrics would be most useful, as well as in the implementation phase as there was resistance by some who did not want to have performance measures for their content. We learned that fully integrating any new dashboard into the daily routines in the newsroom takes a long time and requires champions who will push for integration.

While it was important to us that we develop EngageReaders as an SDK that could be integrated in to any ePaper app, this was not an easy process.

When we first started this process in 2014, we made the best possible choices with the technology available on the market. Since then the technology has evolved so rapidly that if we were to start today, we would have many more options to choose from. That is why it has been very important for us to ensure we are able to keep up with all technology changes.

Finally, the eternal difficulty with selling a product that gathers data is making the insights actionable. This may sound like common sense, but it is much easier said than done.

Big Data & AI for Media:

What is the business impact of EngageReaders?

Danny Lein:

One key business impact is that newspapers using EngageReaders have seen a reduction in churn, meaning more readers are staying engaged with the content and not leaving. By placing more engaging articles earlier in the edition, average time spent reading the editions have increased. We’ve also seen that publications with articles that are read longer have higher Net Promoter Scores by 3-5%.

To date, EngageReaders is the only tool in the industry focused purely on editions, using ePapers as a proxy for their print counterparts. This means that EngageReaders is the only tool available for newspapers to get daily insights into how their print product is read as well. We know that how people read digitally and in print are very similar, so newspapers can now take what they’ve learned from their digital edition and use it to improve their print offering as well.

Big Data & AI for Media:

What are the long-term challenges for EngageReaders?

Danny Lein:

Given the very specific niche in which EngageReaders operates, namely edition-based formats, and given the still existing data aversity among the industry, the business growth is moderate but steady. There are many competing analytics tools, however nothing is focused solely on editions. This is on one hand a challenge for us, but also an opportunity for us to grow.

While it has improved slightly, data aversion is still an issue when working with the newsrooms. This means we are pushed to look for further sources of value for EngageReaders, such as personalized reporting, ‘Rate My App’ function for satisfied readers, heatmaps for visual reports, churn reduction tools, custom surveys, and personalized news and notifications. This last feature is part of a new Google DNI project we are working on with a different publisher, called ‘JAMES, your digital butler’.

Big Data & AI for Media:

What are the recommendation you would give publishers in order to be successful with EngageReaders?

Danny Lein:

One key recommendation for success is to find your champions early on. Find the people who believe in the new product and are excited to integrate it into their daily routines, as they will be best suited for changing the minds of the skeptics within the organization.

Finally, since the launch, a group of launch partners have met every six months to share their experiences with this new innovation in reader analytics and to discuss their insights. This has helped to further improve our product offering, as well as allowing the publishers to learn from each other on how to integrate EngageReaders into their daily workflow.