How accurately can you forecast a song's hit potential?

Hyperlive predicted streaming and sales performance of a billion dollars-worth of singles by 10 of the worlds biggest superstars with  over   80% accuracy .

Hyperlive predicted streaming and sales performance of a billion dollars-worth of singles by 10 of the worlds biggest superstars with over 80% accuracy.


Read the full press release for more detail



When it comes to predicting how many streams or sales a new song has the potential to receive, there are many different approaches you might take. Simply asking industry experts or focus groups what they think is an obvious tactic. Or you might compare features of a new track with those of previous hits under the assumption that what’s been successful before will be successful again.  An alternative approach that has gained traction in recent years is to base predictions upon extra-musical factors, such as volume of social media activity, number of fans or followers, or stream/sales over a period of time to see which songs pop out as potential hits, then channel marketing and promotion resources appropriately. 

At Hyperlive, our approach is like no other. We predict hit potential using an advanced model of the musical experience that, based upon computationally-extracted musical features, captures what drives people’s engagement with music on a fundamental level, and how that guides their broader preferences. This allows us to predict a song’s hit potential before it’s even released, reducing risk, increasing resource allocation efficiency, and adding both speed and accuracy to the decision-making process.


To demonstrate the effectiveness of our algorithm, we predicted the performance of all singles released by 10 of the worlds biggest artists of the 21st Century. The 200+ tacks analysed — from the likes of Taylor Swift, Rihanna, Ed Sheeran, and Bruno Mars — have accrued the equivalent of 180+ Billion streams/1.2+ Billion single sales since their release, grossing in excess of a Billion dollars in the process. By analysing nothing more than each track’s audio signature, we predicted the number of streams and sales it was likely to have accrued, and on the basis of this classified it as having been a Hit Song, a Major Hit Song or a Cultural Phenomenon. Predictions were then compared with actual performance. 

Our algorithm predicted actual performance with 84% accuracy overall, and correctly identified every single track defined as a Cultural Phenomenon. Moreover, for tracks incorrectly classified, predicted streams/sales nonetheless fell within an average of +/- 25% of the actual range.


The major benefits of our data-driven algorithmic approach are many:

OPTIMISE RESOURCES. By selecting the optimal combination of songs to include on an album and, on the basis of their Hit Potential score, those to release as singles, we enable more efficient allocation of marketing and promotion resources.

MINIMISE RISK. Our comprehensive analysis reduces risk associated with releasing — and putting substantial resources behind — songs that are less likely to perform well as singles.

MAXIMISE ROI. By optimising resource allocation and reducing associated risk, we enable a more precise estimation of return on investment.

INCREASE VALUE. Additionally, by releasing Album Tracks and Singles most likely to accrue the largest number of points over time, associated copyrights will naturally increase in value.


Read the full press release for more detail


About Hyperlive. Music isn’t something we just listen to — music is something we experience with our heart and soul. Hyperlive captures this experience by modelling a range of neurobiobehavioural responses to music as well as the psychological processes that underpin them. This gives us a deep understanding of what drives musical engagement on a fundamental level, allowing us to quantify, model and predict that engagement — and the musical features that motivate it — with unmatched levels of precision.


Got some songs you'd like us to analyse? We'd love to hear from you! 

Geoff Luck