Yet another success story

On Habré, success stories or failure of indie developers of mobile games and applications are now quite popular. On this wave, I want to share with you a success story (and failures, which is already there) of a single Android application from a large company. I work as a product manager at Avast, my brainchild is the Avast Battery Saver app. As the name implies, the application saves battery power.

Well, now the story itself - a long text and a lot of pictures.

In addition to the antivirus itself and security in general, on Android we are gradually expanding to other areas. The Tools category on the Google Market is huge, with great competition and hundreds of millions of users oil . In this category, applications for saving battery power are very popular, which is not surprising - most people are unhappy with the battery life of their batteries.

When I began to study competitors, I found an interesting thing: in most cases, the main function that saved battery power was the task killer. If you suddenly think that removing applications from RAM can help, try a simple test. Install any task killer, and click on the main button - Optimize / Clean. The application will happily close everything up and show how many hours your phone will last longer. Wait a few seconds and press this button again. Nothing will happen, and this is right, because you just closed everything - it means everything is in order. Now erase the data of this application or reinstall it. Then a surprise awaits you - after clicking Optimize, about as many applications will close again as they did for the first time. The solution is very simple - almost all of the closed applications wake up again after a few seconds using Wakelock. And in the Optimize button, just a timer, without which it will lose all meaning.
')
Perhaps this is an echo of habits on Windows: when the computer starts to slow down - time to close unnecessary programs. Android does it for you, unloading from the memory primarily of applications that have not been used for a long time.

Then it seemed to me that everything would be very simple - to make an application that actually saves battery power, and crush all competitors. The bosses supported my enthusiasm and plan to seize the world .

We decided to save the battery by automatically switching various profiles: at night, to reduce the screen brightness, to turn off mobile data at work, on Wi-Fi street, to turn off everything except the GSM module with a low charge. In addition, when the screen goes off, turn off the Internet, and turn on every 30 minutes to download mail and so on.

Quickly enough, we already had a working prototype - without graphics and with a bunch of incomprehensible inscriptions. But he worked and saved battery power! We tested it quite simply - we wrote a separate application that includes the screen, downloads data, launches games, and so on. Then they drove this script on the phone with our application for a day, and a day on the same phone without it. On different phones and scenarios, the charge savings were different, but almost everyone was quite tangible - on average, it turned out 20%.

Then they made nice graphics, the application began to look like this:



We launched the first version in three countries - the Czech Republic, Vietnam and Italy. Check if everything works correctly and how people like this version. Of course, we did user tests on users, but when a person sits in your office and gets paid for it, he learns more about all the functions, reads texts and does not remove the application at all 5 seconds after installation. In addition, we conducted user tests “in the wild” - we approached people at McDonalds and asked us to test our application for a few minutes.

And here begins the main difference between a large firm and a startup or indie developer. In a large company there are:
• PR department, which will competently compile a press release and send to journalists and bloggers
• many satisfied users who can be notified of a new product
• well, finally, the strength of the brand - in addition to people who are looking for the name of the brand, it also gives the installation from the section "other applications from the developer"

After the release, everything went more or less smoothly. Hundreds of organic plants per day, several good reviews in the Czech Republic and Italy. PR has brought us an additional several thousand downloads, plus we have been promoting the application in the desktop antivirus for several days. In the Tools category in the Czech Republic, our application climbed to ninth place, in Italy, to thirty-sixth.

The developers, meanwhile, sighed after the release and began to finish everything that was postponed until later. And then I noticed a very unpleasant thing - a rather high number of deletions (Daily Uninstall) compared to the installations. The situation was complicated by the fact that no one could confidently name the cause. In Google Analytics, a specific screen or event from which users are leaving is not visible. In the Google console, you can see only the total number of installations or deletions per day, but it is completely incomprehensible on which day the user deletes the application. Fortunately, we have our own analytics system, which has a cohort analysis by day, like Flurry. And here we were waited by the second unpleasant surprise - about 40% of users deleted the application immediately on the second day after installation.

We stopped progress and began to look for the cause. We tried several ideas, mainly to reduce too aggressive savings: if the user turned on the mobile data, and the profile is set “off”, leave it as the user decided, before turning on the new profile. Or do not turn on the emergency mode (when the battery charge drops to 20%) until the screen goes out. And even show how the application extends battery life:


To my great surprise, all these measures not only did not help, but even by a couple of percent worsened the withdrawal of users.

Then we changed the user's path a bit - hid some profiles.



Some indicators have improved, for example, a larger percentage of users began to get on the screen with profile settings. But it did not affect the withdrawal. The most successful was the release, the cat on the first screen showed the main functions and replaced the text with a more understandable one, for example, “Data on background” was renamed “Internet when screen is off”.



But he radically decided nothing.

Here we decided to conduct user tests again and find out in more detail whether users understand the concept profiles in general and some functions in particular, for example, the synchronization period when the screen is off. And then at last the light at the end of the tunnel began to glow — many users did not really understand what was happening.

When it became clear what the problem was, it immediately became easier to move on. We maximally simplified the problem screen with profile settings. Instead of two functions, Synchronization Period with the screen off and Wi-Fi Auto Mode (turns off Wi-Fi when there are no familiar networks in the area), the Internet optimization has been made a little mysterious. And all that was incomprehensible was hidden in additional parameters.

In addition to all these changes, the design was redesigned to synchronize with our other applications.

And here the difference between a large firm and an indi-developer appears again - tight deadlines. Release in two weeks to catch the Mobile World Congress in Barcelona. Plus, it is necessary to release a long-planned function Power-consuming applications.

This function measures which application has consumed how many clock cycles on the processor, how long Wi-Fi or mobile data has been used, and then converts it into clear percentages of charge consumption using a complex algorithm.

At the same time, we found out that the majority of users poorly understand the difference between the application in the forgrand and on the background (I don’t know how to call it in Russian). Therefore, instead of two different lists:



All applications were pushed into one list and hidden system:



Instead of the usual Scrum Sprint, we switched to the Stakhanovist regime “to finish this to Barcelona.” Big release rolled out on Thursday at 20%. And on Friday, in Google Console, I tremblingly pressed "all countries". Installations immediately grew from a few hundred to a few thousand a day, and this is without marketing support, only organic. Finally, the issue with the withdrawal of users was resolved.

That's the whole story, I will be glad to answer questions in the comments.

Well, for a sweet - current design: