PlayDrone is a fabulous experiment, conducted between April and November 2013 about the Google Play Store. If you haven’t already, you should read the full paper [pdf] where computer scientists Nicolas Viennot, Edward Garcia and Jason Nieh of the University of Columbia describe their methodology and findings. What they did is just a great engineering-backed investigation work, and though others might have done similar things in the past, or even currently with the Play Store, the scale of the PlayDrone experiment as well as the simple fact it has been shared with everyone make this project unique and invaluable.

The PlayDrone is basically a tool scrapping the PlayStore in order to analyze the content all the free apps it can find through code decompilation. The points below are my own conclusions about the experiment, and therefore they do not reflect thoughts or involve the original authors.

Google’s lack of attention

The experiment shows that many thousands of developers have used secret tokens required by third party services like Amazon AWS or Facebook Login in their app, thinking this wasn’t a threat. But it is a threat, because it’s damn easy to download, decompile and exploit a non-obfuscated app.

I like the philosophy of letting users decide if a product is acceptable or not, rather than having a strict validation process. That’s why I like the Play Store philosophy more than iTunes’ way of dealing with app submission. But having no proper background check at all is I think a big mistake. It is one thing for Google to let everyone with $25 become an Android developer, and submit apps without any trouble, it’s another not to collect basic data about the apps being submitted to detect trends, errors, malwares, you name it.

The PlayDrone team has helped Google to address these problems, which is the least we could have expected and a proof that Google is not particularly checking what is being submitted. I find that highly surprising from a leader in that field.

The Play Store’s depth

The PlayDrone team also compared apps to find similarities, and clones. 25% of the free applications were similar (p. 9). What does this mean? Competition is high, which is not a new fact, but more than that, it means some verticals must be quite deep. Clash Of Clans clones are an example, Battery savers another, and the Flappy Bird frenzy yet another one. Now that there are 1.5M apps on the Play Store, differentiation and discovery are going to be an even harder challenge, pushing developers to re-think the way things are done, or perhaps look into less crowded app verticals – yet linked – such as apps for cars, Living rooms, and wearables.

Paid apps / Free apps ratio

If you check out the paper’s table 2 on page 6, you can see the free app and paid app count by Play Store category. The authors note that interestingly enough, the personalization category is the biggest of all (unless we consider Games as a comparable category). What is even more interesting is to consider which category has the highest percentage of paid app, or in other words, in which category launching a good free app might be a disrupting and potentially winning strategy. The top 5 categories with the highest paid app percentage are: Books & reference (39%), Personalization (36%), Medical (32%), Travel & Local (31%) and Comics (31%).

All these categories are relatively small apart from Personalization, which suggests that paid apps may get some traction in this specific category. Confirming this hypothesis would require a much more in-depth look.

Games. Despite the highly popular freemium business model, paid games accounted for 20% of all games, a ratio still really close from what is seen for non-gaming apps (22%). A notable similarity as gaming is such a specific vertical, that a bigger difference (either ways) could have been expected.

And to finish on this matter, subscription-based industry are not relying at all on paid apps, as proven by the News & Magazines category, having just above 5% of paid apps. I hope for them they have good in-app conversion rates…

Google Play insights

There are many ways to gain insights from what’s happening on Google Play, whether it is via Google’s own products (Analytics) or external tools such as AppAnnie, MixRank and countless others. PlayDrone is just an example of what some are doing. The fact so many tools exist shows that the official Google tools are not sufficient to gain enough knowledge about the Play Store. Of course, a lot needs to be kept secret, otherwise the Play Store charts would be full of scammy apps. But for sure, Google needs to raise its game when it comes to Play Insights, or others will do it.

A lot more interesting stuff can be figured out from the PlayDrone’s paper, again, if you haven’t read it, go for it. The PlayDrone source files are also available on Github.