Taste profiles, most often used to serve up recommendations and connect listeners to other users with like-minded tastes, has the potential to serve another purpose -- predicting which new customers will become high-value ones.

According to a recent white paper by the Echo Nest, it's now possible to separate the whales from the minnows, the big spenders from the indifferent tourists. The paper, titled "How Music Services Can Acquire, Engage and Monetize High-Value Listeners," posits a method for predicting each user's future value to a music service.


The attractiveness of being able to zero in on high-value targets is obvious. Music services can devote more resources to understanding and tailoring products to this key group while also delivering more effective advertising and other messaging to a more engaged audience. And if a service is able to do so early on, then all the better.

The paper proposes a three-step method for predicting which new user will become most valuable. Step one is to build a taste profile for each user. Such a profile would measure five attributes: adventurousness (the tendency to stray outside of musical comfort zones), diversity (the variety of preferred genres and styles), freshness (their preferences for new artists and sounds), locality (the geographic diversity of favorite artists) and mainstream vs. obscure (the affinity for known artists versus obscure artists).

Step two involves deciding what's valuable to the music service. That can be visits, shares, time spent listening, purchases or conversions from free to the subscription tier. What's "valuable" will depend on the business models and goals of the particular service.

Step three is to correlate listeners who exhibit high-value behaviors with taste profiles. The music service can use those profiles to identify which new users, based on some initial interactions, would most likely be more valuable.

"We can start to predict whether new users will be a high-value listener and focus our efforts on keeping those people around," Echo Nest CEO Jim Lucchese says.

At this point, it's tempting to jump to the end of the paper to see which attributes signal the highest value. Here's the rub: The answers will vary for different services. For a service catering to an engaged, hipster audience, designing a diverse, adventurous set of recommendations may be the way to achieve higher conversions, Lucchese says. A service created for more casual, mainstream listening would require more familiar, lower-diversity playlisting.

The study cites a report from Midia Consulting that estimated between 60% and 80% of people who register for the free tier of a music service are inactive, meaning they haven't used the service in at least 30 days. The Echo Nest approach suggests that services focus on the incoming users who are likely to convert to high-value listeners, ignoring the rest.

But the Midia report makes the opposite conclusion: Why not try to engage the rest? On-demand music services have struggled with relatively low adoption rates with an estimated 20 million subscribers worldwide, according to IFPI. That's compared with the 864 million pay-TV subscribers globally, according to eMarketer. When looking at it this way, those 60%-80% inactive, low-value users can instead be seen as "priceless qualified marketing lead databases," writes Mark Mulligan, the report's author, suggesting that the music industry could do well to figure out ways to activate a much broader customer base while also engaging the music buff.