Data Sources and Methodology
We worked with several years of weekly streaming data for well-known Christmas songs, covering 2017 to 2025. The dataset included information such as the date, week of the year, season, track, artist, and weekly stream counts. We cleaned and organised the data so each row represented one track in one week, making it easier to see long-term patterns and seasonal changes.
Our goal was to understand when people start paying attention, when interest peaks, and when it drops off. To do this, we used a time-series model that looked at trends, seasons, weeks of the year, and differences between tracks and artists. We also ran checks to make sure the results were reliable and compared peak performance across different years. This approach gives advertisers a clear picture of what drives audience behaviour and helps them plan campaigns at the moments when listeners are most active.