German media collaborative shares 4 lessons in accelerating growth with data, AI
Ideas Blog | 03 April 2022
Using easy-to-understand metrics and dashboards that are embedded in the culture of the newsroom is key to DRIVE — a collaborative project of 20+ news publishers in Germany, run by dpa/SCHICKLER. The focus is on joint experimentation and learning for digital subscriptions.
“It’s an agile community that’s profiting from all the learnings,” explained Meinolf Ellers, chief digital officer of Deutsche Presse-Agentur (DPA). “If you aren’t capable of bringing these insights down into the reality of the newsroom, you will fail.”
DRIVE collects 2 million new events daily in its data warehouse, and then AI algorithms are used to analyse the data.
Christoph Mayer, partner and managing director of Artificial Intelligence at Schickler, guided INMA members through some of the learnings the DRIVE collaboration has achieved through four phases of maturity:
- Step 1: Analyse
- Step 2: Experiment
- Step 3: Transform
- Step 4: Automate
“The question comes, what does that actually mean for the newsroom?” Mayer said. “Finding insights for the newsroom is the end goal.”
These four stages can give other news media organisations an indication of the steps they can take to optimise for their products. DRIVE started with defining core metrics, he said. Mayer believes in focusing on one core metric, even though that might include several smaller sub-metrics.
“You shouldn’t give subscription number goals to the newsroom, because an editor cannot really generate subscriptions,” Mayer said. The focus of the newsroom should be on producing great content based on the insights.
A new metric
DRIVE defined a main metric on which to focus: media time.
“We have decided not to look at clicks or number of users, but we are looking at time they spend on the Web site,” he explained.
This approach works well because the core metric includes others that heavily correlate with the actual targets of getting conversions and keeping subscribers.
“Media time metric heavily correlates with those two. I suggest you find a metric that correlates with your goals,” Mayer advised.
With media time as the core metric, the end result is a goal for the newsroom of writing great content. For article performance, an article generates high media time if a lot of users click the article (meaning the topic is of interest or the teaser created curiosity), or if users read the article for a long time (meaning it is well-written and/or has a good length).
“We always try to boil it down to the question, what is the decision of the editor?” Mayer said. “The decision space isn’t as huge as you might think. We try, in the end, to boil it down to recommendations for the newsrooms that are actually within their decisions. They cannot generate more users, for example. Those are all targets, but you generate recommendations that are in the decision space people are actually working in.”
Print and digital differ
One big thing to consider is the difference between print and digital. Print is a push product, whereas digital is a pull product. When a subscriber receives a print newspaper, they can’t ignore it. On the other hand, they can subscribe digitally and ignore the product.
This means that transforming the print-oriented process to the digital world is not enough, because most print articles do not work in the digital subscription world.
“In digital, the individual article is much more important,” Mayer said. “That means two things for digital subscription. We need visibility of content — people need to find the content — and we need different types of content that work in the digital subscription space.”
They conducted a study to find out how far down people scroll and how positioning of articles on the Web site or in the app affected viewability and media time. Their hypothesis — that visibility would be highly influenced by the positioning of the article — proved to be true. Visibility decreases dramatically based on the position of the article. Click rates also decrease. Most digital articles are never seen.
“That’s really a big problem in digital, because most users open the app but don’t scroll down. If they don’t find something highly interesting to them in the first screen, they will leave the site,” Mayer said.
This means there should be a priority on positioning and how to design the digital site.
“It’s important you show them relevant content within the viewpoint of the user when they first enter the site. This is where personalisation can really come into play,” he added.
DRIVE is developing algorithms for personalising the user experience, heavily involving publishers and newsrooms in this process. Different types of personalisation include things such as a front page box with six article recommendations and clearly marked “recommendations for you” sections.
“Personalisation is a strong way to show the user more of what’s actually relevant to them,” Mayer said. “What you have in print will not work in digital. An endless stream of content is not working anymore.”