Andy Baio has a very thoughtful piece on how the App Stores (plural) could gain by adding a true social layer and help users find better apps for them.

Andy is right. Very little has been done to change the rules.

All the updates we’re seeing with the App store are about preserving the same model based predominently on editorial curation. Even IOS8 is mostly about cosmetic changes and not about a deep revamp that will change the rules. Andy is suggesting to add some identity layer, a way to follow users, get the recommendation from your friends and organizing a feed of information in the App store the same way Apple tried to build it for music and Ping (but with a failed implementation).


Good. But can that really work?

Building a social networks around apps is not the best way to find relevant apps. Actually that can be the worse too.

For the context, we built four years ago our initial discovery app, Appsfire centered on that wrong premise. We thought it would be great. An endless number of startups tried that too after us and failed and not because of implementation but because people actually do not care enough about that .

When people say they want to know which apps their friends like, they actually mean apps that are right for them (and possibly liked by their friends). What users want are simple strong signals and relevancy in a very digestible way. This is why top ranks are so popular. Because this is a strong simple signal. And a strong signal has little to do with “social” recommendation and more about tastes and affinity but also with the interface and user experience. Unlike music where the act of listening/consuming can be associated to some sort of positive signal, the act of downloading means nothing.

Regularly i check Spotify friends activity for new music to listen. I get nothing of relevance. As incredible as it seems: my friends are not a source of good recommendation for music (or books or movies). The same goes with any type of media format. You may be friend with someone but may not share the same tastes on everything. 


This means that in order to be relevant, you need to get users to contribute to “qualitative” data about. That could work if the app review/rating system was simpler, more inviting and built around this “new identity layer” described by Andy. But here is the truth: you don t review the majority of the apps you like (or at all). Why? because this is not a natural thing to do and you just don’t have time.

On mobile, things get worse: users don’t have time and want an easy way to get relevant information. Add a twitter like social layer in the App Store, as is, and the worse will happen: an endless noisy feed, impossible to digest. 

What really matters is not the social activity: what matters is a way to include social signals as a way to get better recommendations or make better decisions on what to install. The core is not about what people around you like. The core is about what you really like and how the app stores can adapt to fit you, not the other way round, with as less efforts as possible. 

A good way to start would be for the App Store to give some controls back to the user on what they want and don’t want to see. The app stores would become powerful if they could help me personalize my interface, either by letting me do it or learn from it. Personally i don’t want to see Games even if they come from my friends, but still the App Store is “forcing” me to view those selections.  On Google play the very first thing you’ll see is about apps that are recommended for you. Not the top ranks.

Android Apps on Google Play

Imagine also some sort of match making based on tastes: I am not sure i want to follow all my friends, or any of them, but the app store (knowing my tastes) could help me match with people of similar interests and provide a digest of things of interest. Finally the app store would look better, a bit like Google play and some of the mock ups designed by Andy, if it was “humanized”. Every app page would “show” which of your “app friends” own this app (or liked it).

SwiftKey Keyboard Android Apps on Google Play

Finally the idea of treating a developer as a “person” to follow as part of a social system is great. But again the level of noise has to find a way to get filtered (imagine following 50 developers each  with a new app every months) based on your tastes.

So what defines taste? Price sensitivity, Main activities (sports,..), Family status, interests in life….Instead of trying to guess them the app stores could just ask me and serve recommendations based on my tastes.

Match making, Personalization and quality filtering are the real difficulty: We want a store that is right for us, for our taste and for our daily usage. This can only happen when Apple will give give control back to the user and let them customize their store experience; Apple has to also apply their massive amount of data to match apps with users through a simple and easy to digest interface.  That would be a true advancement in the world of App distribution. It would remove so much friction and useless downloads.

Apple (and Google, and Amazon..) has to become the enabler to filter the noise. Not just a beautified catalog controlled by some editors. The myth of solving app discovery and discovery at large by adding a social layer is just superficial. Even Facebook which is all about social information applies a considerable amount of information in filtering the noise though a match making algorithm.  And undertaking such a change is massive and requires a lot of commitment, time and ressources.

Friending the App store by adding a social layer is only one of the dimension to a much more important project to make the store more personal, and more relevant. And in all honestly Google play seems to understand this better than Apple, although a lot remains to be done





  • Rodrigue H

    Glad to read this! We’re doing exactly the same thing for entertainment recommendations. And we have the same arguments as in the only way your friends favorite movies (songs etc..) end up being recommended to you is if you have the same interest (and very specific to genres, actors etc..). For example if Ouriel and I like the same action movies but I never reviewed or shared a thriller movie, I’ll never get a thriller movie recommendation coming from you. Once the recommendation scoring is done mainly on your own interest and social is more of a tool to improve that scoring (and not become the base for it) – interesting things happen! (example on our service, people review things without sharing to the public 7x more then sharing it with friends/ public, good for our user’s privacy but bad for our growth, but its ok!)
    Every month or so I see new competitors pop out solely based on “friends recommendations” – none of them have shifted to recommendations based on interest yet, but it’ll happen!
    I didn’t name the service we created on purpose, I’m not trying to advertise here, just saying Amen to what you wrote about and wanted to emphasize on this phrase “As incredible as it seems: my friends are not a source of good recommendation for music (or books or movies).”