Skip to main content

Pandora’s podcast genome project will help you find your next Serial

Pandora’s podcast genome project will help you find your next Serial


Algorithm uses over 1500 attributes to find your perfect match

Share this story

Photo by Amelia Holowaty Krales / The Verge

Pandora’s podcast genome project, which aims to do for podcasts what its music genome project did for music, has launched in public beta. The new initiative hopes to introduce listeners to podcasts they might enjoy based on prior listening habits, and was first announced earlier this year.

Although the project includes curated elements from an in-house team, much of the work is shared with an automated system which draws upon up to 1500 attributes to build an algorithmic picture of your preferences. These attributes include everything from the ratings you give to podcasts, to which episodes you skip, and which you replay more than once. Pandora even claims that its natural language processing is clever enough to identify a podcast’s topics on an individual episode level, allowing you to dip into a gaming podcast for a single episode on your favorite game, for example.  

The initiative comes as the podcast market is maturing, even if it still has a long way to go. Podcasts earn far less in advertising per hours consumed than most other content mediums and still sit in something of a demographic niche. Pandora’s algorithms aren’t going to make podcasts more profitable, but they at least have the potential to make discovering new podcasts a lot easier — a key sticking point to attract new listeners.