Podcast JTBDs

podcast  recsys  search  information-retrieval  ]

07 Nov 2020


Freshness

Music has various jobs, all stimulating the emotional brain and many triggering a sense of familiarity. We often listen to the same tracks again because music has a quality of teleporting the listener to a familiar state of mind - a nostalgia or a mental flow state. Playlist Curation and Repeat Play features, which are now table stakes in every music app, enable users to persist in that flow state.

Podcasts in general do not stir the same brain signals. Listening to a podcast can be considered a more frontal brain activity, demanding a ‘lean-in’ experience. The functional, emotional and social aspects of a podcast JTBD are all centered around ‘getting smarter’. A podcast episode can be considered more similar to a book in this aspect. Once a book is read, it is seldom revisited but only to recall insights, which are saved as highlights and notes.

If there is strong support to the Book X Podcast similarity, how could the revisiting/replaying experience be made better for podcasts?

It would be interesting to compare and contrast -

No Image


Serendipity and Discovery


During a session, a listener skip through music recommendations, spends 5-10 secs on each track before finally arriving at the right one for the mood. The transaction cost incurred by the listener while sampling through music recommendations is negligible, thereby increasing the session length and chances of arriving at the right recommendation. This behavior is harder to replicate in podcast discovery. Podcast episodes can vary from 15-120 mins in length, and choosing one to listen can be a more planned activity. A greater upfront investment makes the user less tolerant to podcast recommendations.

Does this make humans more tolerant to algo-generated music recommendations as compared to algo-generated Podcast recommendations?

How can the podcast discovery behavior be made more tolerant?

Average session time will be a bad metric to compare the listener behavior across these audio formats. Rather comparison across

No Image