A Personalized Song List For You

Our app “FilmiFilmy” was the first to recognize that there was a treasure of hindi film music on YouTube but the challenges in discovering songs was a big hurdle to use. We aggregated the songs, converted the roman song titles to hidni (the first app in the world to show hindi song titles) using some cool algorithmic tricks and also added a hindi voice search (the first app in the world to support hindi voice search). This week, I have heard that YouTube Music is trying to do many of the things we have already done in the app. Its a great vindication of our idea but also a threat to see a big gorilla come into this space.

So how do we stay ahead of YouTube? Or force them to consider a partnership given our innovation is significantly ahead of them?

The solution lies in the magic keyword of internet “personalization”. FilmiFilmy currently has zero personalization. It treats all users as “same”. It is a smart catalog aggregator with a smart hindi UI and hindi search. It has some tools to select your favorite songs into a playlist and share them but this is quite generic.

What should be happening is that we should “know” our users and their tastes. Since have FB login, we should be using this to build a persona of the user based on their FB profile, likes/dislikes especially in music. We should be tracking what they have searched, viewed, listened to, shared, selected in the app. We should be tracking what they have shared on their social profiles. We should be using this to build a model of their musical interests.

The profile of their musical interests should be used to create a song feed that is relevant and personalized for the user. When they start the app, the app can start from their last search, the last song shared/played, the song last commented on facebook (sort of like re-targeting) or from the cohort of like minded people – what the social vector is listening to. Or the top song played on the platform in the last 24 hours or similar time-period. All of this should be managed/configured by the user through a preferences mechanism. The user declares their choices/priorities and the platform uses their engagements on the app and outside in the public domain to meet their needs. The existing catalog should get reduced to a search function, at best.

This seems like a cool idea. But is it only limited to our app? music? Or is this a recipe for a broader app-personalization technology that all consumer apps should be using – whether the app is music, news, books, articles or shopping or ordering food, tickets, concerts – anyplace where a consumer persona helps an app provide better service and smart assistance is game for this technology. Wait, isnt this was Cortana, Siri are all about? Becoming the smart digital assistant to the user. Seems like there is validation of the idea!

What happens after the user is presented the first song / recommended song? If you get it right, the user plays the video, listens to the song and is ready to accept the next recommendation. If you get it right, the user browses onto the next recommendation and if the next recommendation is not relevant, switches to search to find a more relevant starting point. I could live with that!

So, the ideas are in. Now to build this capability. And perhaps find some partners who can pitch in to fund the development of a larger persona and interest database that everybody can use for recommendations. Now that will take some doing.

BTW, if you look at the previous post, you will see that I promised to build the feature to pull images from the cloud into our gifting app, ThotOfU. Happy to confirm that both iOS and Android versions have now fully functional ability to use any of your own pictures for use in a greeting card. Try it and see how simple it is to pick an image from your flickr / instagram account and create a greeting card with it.