Overview of this campaign
Each participating publisher has set its own north star targets for digital subscriptions they want to achieve by the end of 2024. As a general idea, the publishers aim to cover by then 100% of their newsroom cost with digital subscription revenues.
Today, they reach less than 20% of these targets. Conversion from free to paid users is promising, but churn rates are still a challenge. In order to sustainably increase the subscription figures, focus must be placed on the attractivity of the offers.
Personalisation, individual user journeys and products are key elements to this goal. The foundation: data driven approach. Algorithms learn individual user behaviour and adapt the offer to the expectations of the reader. But it is a tedious and costly way. Each publisher has to recruit own data science teams, has to build the data infrastructure and to go through all the experiments. Only to realize that she lacks the critical amount of data.
As most publisher faces similar problems, the German news agency dpa and the leading German publishing consultancy Schickler have called to life a cooperation between publishers with a simple idea: share resources, share data and share insights. The data experts from Schickler set up a joint data warehouse which collects in real time first party customer data from all the news sites. Every single customer action is recorded. All content data is included, every article, live blogs and picture galleries. This giant database gathers more than a billion datasets in less than three months. This is the raw material for self-learning algorithms, analysis, tests and dashboards.
Eight publishers have already joined the cooperation, more are expected. They have created a fascinating culture of sharing, helping, and completely open discussion. All the data is transparent to everyone in DRIVE, dashboards and results are shared and there is a great willing to help each other on the way to digital success – an amazing success story in an industry not know for open books and cooperation.
Results for this campaign
The pilot phase has been initiated in Summer 2020 by dpa and Schickler with three regional publishers: Mittelbayerische Zeitung (Regensburg), Badische Zeitung (Freiburg) and Aachener Zeitung (Aachen), all between 90.000 and 130.000 in print circulation. In the second half of 2020, five new publishers have joined: Lensing Media (Dortmund), Aschendorff (Münster), Südwestpresse (Ulm), Zeitungsgruppe Ostfriesland (Emden), and Rheinpfalz (Ludwigshafen). We expect to increase the cooperation to up to 15 publishers in 2021.
The technical infrastructure has been built in less than four months. We employ state-of-the art data engineering using highly scalable infrastructures in Google Cloud Platform. Specifically we collect raw usage data using the open source Snowplow tracking engine, store all data in Google BigQuery and perform analysis using DBT and dashboards in Apache Superset.
We are establishing a consistent KPI system within the newsrooms of the cooperating publishers in order to compare tests and results among them. "Media Time" per user has proved to be the most important driver to build usage patters. These patterns are the fundamental prerequisite for sustainable subscriptions. We only consider the Media Time of prime target groups and take out fly-bys and irregular users. Every week, newsroom representatives from all publishers gather in a video call to discuss dashboard analysis and test insigths which aim to improve the Media Time per user in the core target groups. As every publisher works with the same database, results are comparable.
The second main area of expertise will be the monetizing game. For the first half of 2021, we have planned to build algorithms for individual pricing, converting and retention programs. The procedure is simple. We first prioritize the topics among all publishers. Then 2-3 publishers carry out the experiments (mostly A-B tests). The effects are analysed by the central data science team and the results are shared among all publishers in the cooperation. With this divide-and-share-approach, we can increase the number of tests and experiments to the benefit of everybody.