Our focus will be on data science and metrics for the newsroom in a privacy-by-design environment. The morning sessions will be followed by 16 data analytics, data science, and data engineering break-out topical roundtables.
A welcome get-together cocktail will be organised the evening before the event, Thursday 20 October, 19:30.
The preliminary programme below is subject to periodic updates. Check back in soon to view our latest updates!
After so many online sessions, it’s time to get together with fellow members of the Data Science Expert Group for a friendly networking opportunity.
Please confirm your participation: RSVP
Enjoy a taste of French coffee and Parisian croissants. We recommend that you join us at 8.30 a.m., as the programme is subject to further development.
An overview of the topics favoured by our readers is an essential requirement for an analytics team, but it is far from easy…
The first challenge is to categorize all the content produced daily. Although an effort is made during the production to attach meta-information to them (for example, categories, tags), creating clusters of articles that cover the same topics can be complex. We will demonstrate how we use recent advances in machine learning and input from journalists to develop new, meaningful categories. Finally, we will discuss how we evaluate the potential of these clusters by considering the uncertainty around future performances.
Regulatory measures by legislators, software providers and manufacturers are fundamentally changing the way audiences can be targeted, measured and tracked. The target scenario drawn up by nxt.statista is inspired by its conversation with publishers and has moved towards developing a holistic data strategy. It does not view the touchpoints, data points, and use cases of “subscription generation” and “advertising” in isolation from one another but combines them and lets them benefit from one another. On a strategic level, it shows which addressability pillars for advertising will be available in the future and which data will be required internally for maximizing control and user experience.
Finding the right balance between being informed by the data and respecting editorial choices. The use of audience data in determining the business model.
Reader behaviour and willingness to pay for traditional newspapers have been under radical change due to the emergence of free or cheap online sources, leading to a decreasing number of subscribers in recent years. The F.A.Z. Churn Prevention Model deals with the high churn rates of its print newspaper and e-paper by identifying customers willing to churn before they actually decide to cancel their subscription. With more than 50 variables, the model predicts which customers are likely to churn with a probability of 90%. The model is embedded in the F.A.Z. data lake and sales systems to automatically provide current churn scores. In recent market tests, prevention measures based on churn scores have led to up to 40% lower churn rates.
People seem particularly willing to pay for experiences, like travel and sports events, even at higher prices, after being deprived of them under pandemic restrictions. Subscriptions to news products were no exception. But in 2022, several companies say they have already noticed higher prices hurting demand, at least for some of their products. So what will happen to our content offerings this year, and what can we learn from the experience of Mediahuis and Mather Economics when moving into the pressure for real-time adjustments to increased inflationary stresses?
We want to allow participants to join the discussion group of their choice. If one of the round tables on the agenda conflicts with another topic of their choice, in the same time slot, we will repeat it in the second round of round tables on the same topics. In all, we will have 8 topics discussed twice during the afternoon.