The "Article Score" was developed by NZZ's data and audience management team. Each article can score a maximum of 1000 points. The score is divided into three individually weighted sub-metrics: attractivity, engagement and conversion. These metrics are backed by measurable numbers an article generates every hour, such as clicks, engagement minutes and subscriptions.
Besides visualizing the sub-metrics, the main goal of the tool is notifying managing editors about anomalies detected in the data. Anomalies are significant differences in the sub-metrics: If, for example, an article has a high engagement score but is low in attractivity, the article usually demands special attention. An editor then should probably try to promote it better on the available publishing channels. For the MVP we decided to describe the detected patterns instead of offering concrete action steps.
The screen above shows seven anomalies by highlighting them in blue and offering a short, descriptive paragraph. Through the support of a foldable layout, anormal rows can be easily brought to the top – while the top five stay expanded permanently. Once a managing editor has done a content-related change like the adoption of an article title, the check on the right can be clicked which makes the article aggregate data based on the new set conditions.
The color coded sub-metric bars come with adjacent information blocks: on the left the usage of the articles publishing channels are indicated (N for Newsletters, P for Pushes, S for Social Media, CH/DE for the front pages). On the right, the article's generated conversion numbers of registrations (R) and subscriptions (A) are revealed. In the list's right-most column live the user group batches and indicate special achievements when reaching the majority of a group like females, students or german-based readers.
To be continuously displayed on NZZ's big newsroom display, I created a dark mode version of the dashboard. Besides using stronger color contrasts, the typography is more readable from a distance this way. The sub-metric color's new brilliance allowed to downscale and stack together the bars to better visualize the Article Score's total.
Currently I can only show the concept and design overview of the current articles. I also designed an article detail view and a historic overview for mainly prospective purposes, but that part of the project is confidential at the moment.