For a self-identified "data scientist", this is a pretty naive approach to "finding out what worked for others". The methodology doesn't distinguish in any way between "colors that help make apps successful" and "colors that are just used often by everybody".
It could be that there really are some colors that will help an app to be successful in the app store, but without comparing the data from successful apps with that from a control group, there is no way to identify them. If the goal is simply to make an icon that wouldn't look out of place in the top 300, this is a good formula. But that doesn't strike me as a very interesting problem.
I'm the "self-identified data scientist". I agree that there are quite a few holes in my analysis, but I was not aiming for an actual increase in sales just by designing the icon this way. I however wanted to have an icon which looked good enough and could blend in, at least from a color point of view.
So I guess it might not be an interesting problem to you, but I actually had quite some fun thinking about it. Which is the most important for a side project, right? ;)
Confused about the blending in part. One of my biggest gripes is continually missing an app icon because it's blue (or otherwise similar to another, but generally blue). We just had a conversation in a similar thread the other day - "Make Your App Icons Boring" - http://news.ycombinator.com/item?id=4219164
You also don't show the icon at the small size that we'd see on our devices. Does it translate well?
As someone else commented, your title is a bit misleading. Perhaps try the same analysis for the bottom or a random subset of 300 apps and see what comes out. Dataporn can be nice, but you should be a little more rigorous with your statements.
My guess is that while that might help with the initial evaluation of the app, and possibly even smooth the initial sale (familiarity and all that), looking like other icons is a net negative for continued usage for your app.
I routinely try to check in with Foursquare on my phone and accidentally open Facebook, because their icons are similarly colored. Meanwhile, the "Words with Friends" icon is not something that could be easily confused with anything else, and I've never had a problem finding / using it.
Once your brand is established, those problems probably go away, but I know I've dropped a number of apps before because I wasn't using them because they were too difficult to easily identify in a menu lineup.
If you want to develop camouflage then this would be a good way to go about doing it. If you want to develop a brand identity that stands out, this is precisely the wrong way.
Distribution of hues of the aggregate of colors present in the top 300 iOS apps icons.
What would be more interesting is if this were compared to the distribution of hues in a large collection of unsuccessful apps too. If the histograms were somewhat similar, then the hue is probably of little significance.
Couldn't you argue that _not_ having the most widely used color would be a great differentiator for your app? People have been complaining for the longest time how pages and pages of apps on their iPhone are just pure blue icons. Why be another blue icon lost in a sea of iPhone apps?
I can't comment on approach or validity, etc. I do have one observation though, the icon looks masculine. I would either make it androgynous or add a feminine addition to the icon so that people don't pigeon hole it as a "guy's" app or something like that.
In some ways, a smaller data set is harder to deal with than a larger one. For example, the surprising effectiveness of Bayesian statistical analysis is mostly due to the huge amount of data you have to throw at it.
Oh, it's blue. I seem to be the only one bothered by it but starting any iOS or OS X device for the first time bombards you with a lot of blue icons. I'm starting to equate blue icons with brown video games, more of the same.
Aside from anything else, really? Your app allows a user to send regular, scheduled text messages to friends saying "Where are you?" so that you get the benefit of knowing where they are? That sounds horrible for the recipients of these messages that I effectively have no way to opt out of, without telling my friend to knock it off. I could choose to ignore the request, sure, with the potential for passive-aggressive drama to follow.
What are the use cases for this app that "Hey, where are you?" won't suffice for? "Here, click on this link and I'll see, on a map, verified by your phone - and this website will also get to know."
Good job. My thought is after Facebook came to limelight, most of the icons are made in cyan-blue. No wonder your result, as well as your icon is towards that.
It could be that there really are some colors that will help an app to be successful in the app store, but without comparing the data from successful apps with that from a control group, there is no way to identify them. If the goal is simply to make an icon that wouldn't look out of place in the top 300, this is a good formula. But that doesn't strike me as a very interesting problem.