Our last app update (version 2.3.1) brought with it a very important change for TimeTune. We finally retired our old ‘tt’ icon and changed it for a new minimalist ‘clock hands’ icon:
Why is this change so important? Because the new icon performs 11% better than the old icon. This translates to more downloads, more daily active users and therefore more revenue.
In this post we’ll explain how we got to the new icon and the lessons we learned on the road.
This will be our second post dedicated to app icon experiments on Google Play (but our findings can be applied to the App Store as well). You can see our previous set of ‘hard-learned’ lessons here (you should read it if you want to avoid the same rookie mistakes we made!).
GRAB A PENCIL
Store listing experiments are there to help you test all your crazy ideas. You never know which icon will perform better until you have tested them all in the wild. And you might get surprised.. as we did.
When we started our app icon experiments, we had a few goals in mind:
- Find the main shape for the icon (and refine it progressively).
- See if we could find an icon that could combine the notions of time + fine-tuning.
- Try to find a unique and recognizable identity for the app.
And so we started sketching ideas on paper. We draw a lot of icons (and when we say a lot we mean a lot, in the range of several hundreds). We tried all imaginable shapes and combinations related to time, fine-tuning and the ‘tt’ letters (keeping the letters was certainly an option).
Only the drawings that looked good on paper went to the next phase: sketching on desktop. It’s funny how some icons that look awesome on paper lose all their charm on the screen. They seem to auto-eliminate themselves.
In the end, we reduced our selection to seven icons:
Some of these ideas only appeared after plenty of sketches had been done. Had we desisted the search, we wouldn’t have found them at all.
So if you’re starting to design your app icon, grab a pencil and get ready to draw a lot. The process is not easy and you’ll get frustrated many times (especially if you’re not a designer!).
DON’T BE BIASED
Out of the seven selected icons, three of them are based on an hourglass shape. Why? There’s no other way to say it: we were biased. We liked the hourglass shape and we wanted it to win (in fact, we believed it was going to win).
But the app icon experiments proved us wrong.
It’s easy to fall in this trap when trying to find an icon for your app. We think we know what’s going to work and what isn’t based on our own prejudices. We almost discarded testing the winning icon for being too similar to a clock app!
This was a particularly tough lesson to learn: don’t be biased, give a fair chance to all the ideas.
HAVE A CONFRONTATION
After learning from our initial mistakes, we realized that testing icons one by one doesn’t always give you the full picture. Did we stop each experiment at the right time? Were the results stable enough? When percentages are similar, which icon is better?
For example, these were the results after testing each icon individually for several weeks:
The percentages represent how well each icon performed against the old (then current) ‘tt’ icon. And yes, all the icons performed better. Until then we weren’t really aware that our ‘tt’ icon was so bad 😥
From the results we could already see that icon #2 was a good performer. But was that result too good to be true? Had we stopped that experiment too soon? (and most importantly, why the heck didn’t an hourglass shape win?).
We needed a way to verify the result. That’s when we decided to confront the two best performers on a single experiment: icon #2 and icon #6 (plus the base ‘tt’ icon).
And this was the final result (we stopped the experiment when we saw totally stable results, which happened after four weeks):
(Again, the percentages represent how well each icon performed against the base ‘tt’ icon).
The most important discovery here was not that icon #2 outperformed icon #6, but the fact that it did it in a very consistent and regular way, increasing the difference day after day. And this is something we couldn’t see on the individual tests.
So if we had to start the app icon experiments all over again, we would use confrontations from the beginning, testing icons in pairs (besides the current base icon) and choosing the winners as contenders in subsequent rounds. This will give you a better picture of which icons perform better.
DON’T MISS OUT ON A GOOD CHANCE
OK, we found a shape that performed well. The ‘clock hands’ icon could be the base shape for the new TimeTune icon.
Because… we couldn’t use it just like that, could we? The icon was too similar to a clock app icon! But on the other hand… it performed 11% better than the old ‘tt’ icon. So what to do?
Our initial intention was to tweak the ‘clock hands’ icon progressively, performing more and more app icon experiments until we could find the ‘perfect’ icon for TimeTune.
But seeing that the process could take serveral additional months, we finally decided to retire the old ‘tt’ icon on the upcoming version and leave the tweaking process for later.
A full 11% increase in downloads is a chance too good to waste!
And the thing is that -after updating the new icon on the app, on Google Play and the web- we love how it looks! This made us realize how blinded we were with other icons (including the ‘tt’ icon).
So don’t miss out on a good chance. If you find an app icon that performs much better than the current one but it’s not your ‘perfect’ icon yet, weigh pros and cons and see if you can make it your new base icon. You’ll have time to tweak it later.
ONCE FOUND, KEEP SEARCHING!
Our new icon will stay for a while. Knowing that it’s a good performer, now we have time to calmly experiment with tweaks and modifications. Or even change our focus for a bit and test other features that need some improvement as well: screenshots, feature graphic, app description, etc.
But the fact is you don’t know if there’s another icon variation waiting there to be discovered. A new icon that performs much better than the one you just updated.
That’s why you should always be testing one marketing feature or another with store listing experiments. Keep searching and you’ll find more gold!
We hope our findings can help you improve your search for the ‘perfect’ icon. If at least we helped you avoid our same mistakes, that’s more than enough for us.
Thanks a lot for reading and good luck on your quest! 😃