I am thinking of building a curation product. I totally understand the value it holds for music fans, and curators, but where do the biggest opportunities lie that may not be obvious at first glance?
What should we keep in mind in order to scale a curation product later, even though early on we are focusing on a small and specific niche?
Music curation is something that greatly interests me as well, and one of my projects slated for 1-2 years out will be in that area.
Don't forget that -- apart from being united by musical taste -- fans usually have MUCH else in common as well. Musical taste is a great predictor of other demographic features such as age, nationality, race, gender, region, economic class, and a million much subtler social characteristics.
Fans of Steely Dan or Yo Yo Ma are different overall than fans of 50 Cent, Garth Brooks, or Lady Gaga.
That's obvious, and you know it already. But keep in mind that sites like Facebook are valuable for advertisers, yes, for the eyeballs but partly because they can CHARACTERIZE those eyeballs and predict their behavior as consumers.
You may be making a site that's laser-focused on music as such. But monetizing music has become notoriously difficult in the post-CD era. So you may want to look at non-musical ways to monetize music, based audience characteristics that various fan groups happen to share.
If you're wondering about a brand name or domains, call me.
The best music curation is done by people. Having worked in the music industry for a number of years including radio station, live sound, stage, concerts, and music production, potential customers love to find a good station that they can just hit play and enjoy. If you have a good DJ you will have a big company. There are lots of ways to go about doing it to from doing it entirely by yourself with developers to using services like Live 365. In my experience, customers always ranked music picked by a DJ higher than those picked by Pandora, Google, iTunes etc.
The downside to curation systems is they are designed to find things you enjoy which means your music is consistently going to be the same. DJ's are able to introduce you to music you might not have discovered otherwise.
As a side note I am not an old guy reminiscing about the good ol days. I'm in my 20's with active Pandora, Google Play All Access, and iTunes accounts. I love being able to discover a song through the radio or being occasionally surprised by Google (it's algorithm isn't as good as Pandora so it sometimes thinks I would enjoy some pretty strange songs) and then adding it to my Google Play account.
I currently run a SaaS in the music tech space (VL Group) and we are seeing an increased focus on curation. Existing curation services are owned/operated by either the labels or the streaming services. I believe the next wave will decouple curation from platform and content. This will allow tastemakers to monetize their value directly. This requires the existence of a playlist without the requirement of a platform that will allow creators to collect the data, drive new use cases and monetize. Spotify playlists are the key to driving their brand trust and I think that this tool will be sought out by other entities (brands, tastemakers, venues).
There is also a huge problem in new music discovery that needs to be addressed. Purely algorithmic and human solutions have been the two camps in this battle but I think the answer is somewhere in between. In addition to this blended approach I also think the we need to re-think what the basis of the recommendations are. Since there is a big problem with the meta data associated with tracks we will need to look at other sources for matching. A site like MusicBrainz is crowdsourcing a lot of data but that is just the first step. Songs can be matched based on any number of characteristics other then genre, artist, etc. In today’s music environment genre is increasingly meaningless and the reality is that I may have a match based on instrumentation, lyrics (all songs with the word rain in them on a day of downpours) with machine learning growing rapidly I think that understanding the listeners “intent” is a missing element.
Happy to talk more about what we are seeing in the industry or any of the thoughts above.
