Science
Machine Learning Models
New technology that applies machine learning algorithms to real-world music and data problems.
Music and Data Converge
Improvements in machine learning—more accurate models, algorithms, and decision-making—align with the music industry’s transition to digital services.
Today’s music visionaries are building a future where machine learning technology will predict, scale, and automate artist investments, upending traditional marketing and A&R forever.
FIGURE 1.
Lift Model
Global artists have multiple marketing campaigns active at all times. But without the ability to isolate paid marketing lift from external influences like fan-generated content, it’s hard to know whether ad spend truly moves the needle.
A lift analysis provides a powerful way to quantify the value of music advertising on an artist’s business, independent of all other activity.
Using a mix of historical data sources, a lift analysis works by forecasting two possible outcomes—the change in key business metrics over time, both with and without ad spend—then calculating the difference in performance between them, which represents lift.
FIGURE 2.
Time Series Forecasting
Current marketing forecasting solutions use past data to assume future results. But these approaches rarely produce accurate results due to a lack of context around buyer behaviour.
Machine learning techniques take into account additional variables, including seasonality and discounts, to build forecasts that better reflect the unique trends of artists and their fanbase.