Hit Potential Technology
All bold assertions must be rigorously tested in a reproducible manner. Our assertion remains that our Hit Potential Algorithm is able to determine whether a piece of music is likely to be commercially successful or not, based on the audio alone.
This time, we ran a much larger set of 14312 songs to verify the accuracy of our initial results. We picked 4 separate and commercially successful (measured by virality or number of streams) sets of music and compared them with each other. Duplicate songs were removed.
Examined Sets
- Global Top 200 (352 Songs)
- Latin America Top 200 (903 Songs) - These also comprised a Trigger City set in another study.
- USA Viral (1543 Songs)
- Southeast Asia Viral (2842 Songs)
We also compared these with a ‘control’ set of 8672 uncurated user-generated songwriting submissions (UGC). This is a heterogeneous set where the commercial success of each song is not known nor measured by any other metric than the Hit Potential score.
Higher Score = Higher Commercial Success
We performed a one-way Anova on the dataset and found that they were all significantly different from each other (p<0.0001).Â
Global Top 200 had the highest average hit potential score (75.12), followed closely by Latin America Top 200 (71.93), a result that corroborates Chartmetric’s findings that songs which chart in Trigger Cities are more likely to also make it to the Global charts.Â
These 2 were followed by the USA Viral set (69.42), and then the SEA Viral set (66.65). Predictably, the UGC set scored the lowest (54.46) on average, demonstrating that higher Hit Potential scores are correlated with greater commercial success.
Global Trends vs Niche or Regional Trends
The scatter plot above compares the distribution of Hit Potential scores within each set.Â
While there is a huge range of scores in each set, the Global Top 200 set has the tightest concentration of high-scoring songs. This set is arguably the most curated set, representing the most-streamed songs in the world (versus within a single country).
Notably, the UGC set, while being mostly concentrated in the lower range, clearly has some songs that score in a range more akin to the majority of data points in the Global Top 200. These outliers that may indicate the potential presence of undiscovered hits.
The other commercially successful sets have a number of songs that fall far below the Global Top 200 mean, suggesting that those particular songs are charting in those areas due to regional tastes rather than global mass market appeal.
Hit Potential, then, is most useful as an indicator of global trends and likelihood of global commercial success specifically.
A Cutoff Point
This figure shows the percentage of songs within sets that received a score greater than 65. 91.48% of songs in the Global set received a >65 score, contrasting dramatically with the UGC set, where only 13.94% of the songs did. This may serve as a useful cutoff point for A&R executives poring through thousands of song submissions to drastically reduce their workload by allowing them to concentrate on the best of a large batch.
Additional Figures
The above figure shows the distribution of the UGC set (yellow) versus the individual commercially successful sets.
The above figure shows the distribution of the UGC set (yellow) versus all the commercially successful sets combined.
Conclusion
Clearly, the Hit Potential Algorithm is a powerful tool for identifying commercially viable music, though it should be reemphasized that it performs best as an indicator of global mass market appeal, versus niche or local trends. In fact, looking specifically at songs with lower Hit Potential scores that still make it to a region’s charts may be a useful exercise in identifying and quantifying such niche trends.
Want to know how your song will do? Check out our Hit Potential Songwriting Competition!
