Carefully analyzing billboard hits over the years can reveal useful patterns in music consumption. Artists and record label executives can act on these to ride the wave of a trend or get off before it becomes unfashionable.
Clear Trends In Billboard Hits
Here's an example of a clear trend that emerged from the analysis of 20800 songs (1806 unique) from the Spotify Top 200 over 2 years, from week to week. The Y-axis percentages shows proportion of songs in the chart with a particular tempo range.
As you can see, 90-99BPM is clearly getting less popular, and 120-129BPM is clearly growing in popularity. This type of clear and useful trend appears quite often in the Top 200 set, across several of our classifiers (We'll have another article about that!)
But it's a totally different ball game with Viral music.
We analyzed 5200 Viral songs across 2 years, from the Spotify Viral Top 50, and found that the data was much more random. Here are some examples of that, including contrasting data from Spotify Top 200 music in the same period.
Every dot in these graphs represent the percentage preference for a particular tempo range or genre during a particular week in this 2 year timeline.
As you can clearly see - the Top 200 data is tightly grouped and shows a fairly steady ~10% decline across the two years, whereas the Viral 50 data is, quite simply, a complete mess. No trends are clearly visible and the preference for this tempo range varies wildly from week to week. This is largely true across many of the most popular tempos ranges:
Deriving Insight From Chaos
However, the Viral 50 does show meaningful patterns in Genre popularity in some ways:
The Folk genre enjoys greater popularity on the Viral 50 charts than the Top 200 charts on average. This varies wildly, and sometimes Folk does not appear on the Viral 50 charts at all.
At first glance, it appears to be mostly a mess, but take a closer look: At the same time each year, around Christmas, Folk genre songs are significantly more represented on the Viral 50 charts than normal.
Let's take a look at another seemingly chaotic dataset:
Here you can see that the Top 200 exhibits much more predictable behavior in the preference for Latin Genre music, while the Viral 50 data is extremely variable. Overall, it is safe to say that the Latin Genre is less represented in Viral 50 on average.
There is one possible trend that may be worth taking a closer look at in an expanded study: Right after the New Year, Latin drops in popularity, and then climbs back up by more than 5% around March/April.
Overall, what we can see from this data is that Top 200 trends tend to exhibit slightly more controlled and predictable behavior. Looking at a graph of the data is often sufficient to make basic determinations on past and upcoming trends.
However, data from the Viral 50 set tends to be much more highly variable.
This is perhaps due to the organic and not necessarily strictly musical reasons for music to go viral.
On closer inspection, these may reveal important trends, but overall, they mostly point to the fact that virality is quite unpredictable.