Frédéric Block

A few weeks ago, I blogged about user retention and how to try improving it. As I stated, we used these methods to analyze and improve user retention on InboxZero..

What and how did we do ?

Retention : 2 to 4%...

Currently InboxZero is still in launch phase. This means we release a version, perform marketing around it and analyze user retention / behaviour afterwards to determine whether we are heading in the right direction or not.
The latest available version (2.5) achieved a user retention of 2 to 4%... Compared to the 5% to 10% an app should try to achieve, InboxZero is clearly too low. Why is this ? How can we improve it ?

App stability

The first thing you tend to think is 'my app is too buggy'. That's why users are not keeping it. 
Of course, InboxZero is not perfect. There are bugs, there are functionalities not working as they should. We know that and we keep improving this. This is 'basic stuff', common to any application...
So, while app stability can certainly be a major reason not to keep using an app, you should not stick -too long- on this one.

But which are the flaws ? And how did we find them ?

Flaw 1 : onboarding

InboxZero has its own servers. So we have data we can analyze. Using this, we noticed 30% of users who installed the app never actually at least read one mail. Based on internal and external feedback, we understood user was not ready to wait before he can test / use the app. But since InboxZero has to process all user's mails to be able to build the contact's conversations structure, user could wait a long time (up to one hour) before being able to use the app for the first time.

We decided to introduce 2 work-arounds :

  • The account data processing will be split in two phases.

The first phase, which can be -technically- made faster, should be done after 2-3 minutes for most users.
Afterwards, the engine will keep working but user will already be able to use the app.

  • We will introduce an interactive onboarding. The user will use the app in a training mode, with fake data, to get accustomed to what the app can do, and how to do it... This will take approximatively 2 minutes..

Thanks to this onboarding, we hope most of these 30% of users will keep the app long enough to really try & enjoy it...

Flaw 2 : app intuitivenes

"The app has great features... maybe too many... and basic ones are difficult to reach."
This is the feedback we got from our user's tests. We performed this 2 ways :

  • We worked with an external partner, Anais Digital, specialized in ergonomics auditing, to help us in finding out what was good or bad within our app.
  • We organized tests with users who never used the app before. Those were asked to perform some basic tasks (such as install  and configure the app, send a mail, find a mail,..). This was filmed and personal behavior was afterwards analyzed to highlight difficulties.

Based on this, we decided to re-design many screens with focus on the main key features of our app. Most advanced -and scary- features will be 'hidden', meaning not as visible as the major ones.

Flaw 3 : Use your data

Building a 'lighter' screen can look simple. It is not. How can we know which features are important to users, which are not ? By using the tracking figures we have on each possible action within the app, we were able to quickly define what was rarely or never used. 
If the underlying feature was important, we had to find a better way to expose the action. 
For unimportant features, action could simply be removed.

Flaw 4 : Comparative testing

In the process of deciding how to re-design screens, we had to take decisions. The choice between two -or more- ways of improving something can be difficult. Maybe we went in the right direction, maybe not... So we sometimes decided to develop different solutions... and we will compare whether the app retention in one case can prove to be different from the other...

The next release of InboxZero (v3.0) which is based on these is expected to ship in Q4 2017. I hope you will enjoy taking the first 5 minutes to test and like it :-)