The Talpa Network ratings predictions experience

• Challenge: Manual TV ratings predictions at Talpa Network faced challenges regarding efficiency and precision, impacting the optimization of ad sales. 
• Solution: Ratings Artist
  • Daily touchless operations with human-in-the-loop approach 
  • Accurate ratings predictions for multiple channels, break types and target groups 
  • Integrated with BI Studio, providing a user-friendly interface for data visualization and reporting 
  • Empowering the team to monitor, manage, and override predictions with ease 
• Impact: Streamlined ad sales processes, advertiser satisfaction and increased profitability.  

One of the most intriguing applications of AI and machine learning in media and entertainment is predicting TV ratings to understand the audience better and improve content monetization strategies.   

Talpa Network, a Dutch media conglomerate and prominent player in the media and entertainment industry, faced challenges in optimizing ad sales due to the manual and time-consuming process of forecasting TV ratings for commercial breaks. The existing methods involved running averages in Excel, manual corrections for outliers, and heavy reliance on gut feelings. The process needed to be faster and more precise. 

They called on MEDIAGENIX to help devise a machine-learning model, specifying that this model had to meet three critical requirements: 

1. Performance: The machine learning model’s predictions had to match or exceed the accuracy of human experts.  

2. Human in the loop: The model cannot account for every contextual factor, so the experts would still have the authority to override predictions.  

3. Touchless operations: Predictions needed to be generated daily, incorporating the latest actual viewership data without human intervention.  

MEDIAGENIX collaborated closely with the Talpa Network team, which brought a wealth of expertise to the table. They aimed for a model that performed similarly to the team of experts, allowing automation to take over a significant portion of their work and freeing them to focus on more critical tasks.  

AI predictions tailored to unique nuances 

This resulted in Ratings Artist. Rating Artist leverages a machine learning model that considers a vast array of metadata, including content metadata, climate information, holidays and calendars, schedule-related details, and historical performance. The model is designed with daily touchless operations for up-to-date forecasts and a human-in-the-loop approach, allowing experts to override predictions. This human expertise ensures that the predictions are also tailored to the unique nuances of Talpa Network’s business.  

The process now operates daily at Talpa Network. To ensure the results are comparable to or better than human experts, the model’s predictions are compared to the actual ratings received for commercial breaks.  

Understanding the model’s decision  

To provide transparency and build confidence in the model’s predictions, MEDIAGENIX implemented a system to help understand the model’s decision-making process. 

This system breaks down the main factors influencing each prediction and their direction, such as the weather, cast popularity, broadcast start time, channel, programme runtime, and historical ratings. This insight helps broadcasters comprehend why the model predicted a particular rating for a specific commercial break and allows them to make more informed decisions refining their strategies based on the model’s reasoning.  

Every time the model makes a prediction, it also gives a confidence score between 0 and 95%, expressing the level of certainty associated with each prediction. This allows the forecasting team to manage by exception and concentrate efforts on where the biggest improvements can be made.   

User-friendly interface for ratings management  

While Ratings Artist handles the data-driven predictions, BI Studio enhances the visualization and reporting. It equips analysts at Talpa Network with a ratings cockpit to monitor and manage everything by exception. BI Studio provides various visualization tools and features, allowing the team to track performance, compare forecasts with actuals, and conduct in-depth analyses.  

The Talpa Network team utilizes BI Studio to quickly identify and manage predicted ratings that require attention. Through this tool, they can easily override ratings where necessary, thus improving efficiency and precision.  

Significant improvement in efficiency and profitability 

The implementation of Ratings Artist resulted in a significant improvement in efficiency and profitability for Talpa Network. Predictions for four channels, three break types, and 17 target groups are now generated daily, enhancing the network’s ability to make swift and accurate data-driven decisions. The integration of human expertise ensures reliability and tailoring to Talpa Network’s specific business nuances.  

The use of BI Studio further enhances the visualization and reporting aspects, allowing the team to monitor, manage, and override predictions with ease.  

This combination of AI-driven predictive ratings and robust data visualization tools has empowered Talpa Network to maintain advertiser satisfaction and streamline ad sales processes, showcasing their commitment to innovation and data-driven success in broadcasting.