Overview of this campaign
The homepage of Blick.ch is the most important entry page for the Blick news plattform. Serving over a million users per day, interests in topics can be quite diverse. The most prominent placements to show articles are however limited which leads to potentially interesting content being missed by users because it is too far down on the page or simply not visible. Instead, they get to see content which is not interesting or relevant to them (e.g. football news for non-sports readers).
Through a smart use of data, users should get to see more relevant articles to them in more prominent positions. At the same time, editorial staff can define core content which all users get to see thereby preserving the unique profile of Blick and preventing the creation of undesirable information bubbles and echo chambers where users only get affirmative or one sided news.
All these measures should lead to more returning users and longer sessions because users find more relevant content catered toward their interests while preserving the editorial fingerprint.
This would also free up valuable time for editors to focus on creating new content and increase content quality instead of focusing on micromanaging the homepage.
Results for this campaign
Powered by the personalization engine developed jointly by Ringier and RASMAG as part of STAR collaboration, available in Ring Publishing platform, the Blick homepage is now partially personalized.
Editorially defined core content is intelligently mixed and ordered with a pool of articles written by the staff published within the last 48 hours or imported from wire services. To further ensure that breaking news or similar reporting is served to all users fixed positions are defined.
Users are now clustered into five segments according to their interests and reading behaviour. For each of these segments, a separate homepage is created. These consist of the fixed content, the core content and articles which are chosen after a semantic analysis and categorization based on the preferences of the segments. To optimize the automated content selection, articles are then constantly tested to optimize clickthrough rates thereby improving the selection process and the personalization algorithm. The overall success of a segments homepage is measured against the manually curated version. This version is shown to 10% of the users to serve as a baseline.
Through this, the clickthrough rate for the whole homepage has increased on average by 4%, the session duration by roughly 5%. On the personalized positions, the uplift is at around 10%. Users as a whole get to see more of the editorial content of Blick (35 items opposed to 20 items on the manually curated baseline version).
As a side effect, the newsroom can use the gathered data to discover hot topics. By permanently automatically testing new content across all segments and monitoring this traffic in real time, site editors can identify content which is unexpectedly sought after. They can react accordingly with more reporting on these issues or enhancing existing content.
Fuelled by the first results, a more granular segmentation of the user base is planned to increase precision of the personalized content selection. Also expanding the number of personalized positions seems promising.