Inside AppMetrica
September 26 2018

AppMetrica On-Going: news for fans of qualitative report data

In this installment we’ll talk about how to avoid losing important app events (with a bonus tip from Max Kozhnov at Azur Games), plus new user privacy features in the AppMetrica SDK and a couple of nice additions to the Profile card.

Disabling statistics collection in the SDK

Most likely, your app already has a “disable sending statistics” option. Sometimes it’s directly offered to the user during onboarding, but it’s almost always available somewhere in the settings.

This can complicate analytics: A new user hasn’t yet agreed to transmit their data (they’ll probably agree after completing a couple of steps in the tutorial), but you need to collect valuable data about tutorial completion right now.

The AppMetrica SDK 3.2 (for iOS and Android) has a new way to enable and disable sending statistics, as well as a flag for activating the “don’t send statistics” option.

It’s pretty simple:

  1. The SDK initializes with the “don’t send statistics” flag.
  2. All the important events that you track in the app are transmitted to the AppMetrica SDK, but they are stored locally and are not sent to the server.
  3. As soon as the user agrees to send statistics, call the “enable sending statistics” method in the SDK. All of the accumulated events are sent.

To learn more about how to use this method and the flag, read the documentation.

Force sending the SDK event buffer

Here’s a little tidbit that you probably already know: events passed to the AppMetrica SDK are almost always sent with some delay, because they are first grouped into packets of several events. If there aren’t enough events, the packet is sent when the session ends or the timeout occurs. This buffering keeps the reports up to date in real time, while also saving significant amounts of traffic and battery resources on the user’s device.

However, there are cases when a user exits the app too soon, and doesn’t come back — or even deletes it!

These scenarios need to be analyzed as thoroughly and accurately as possible. This is the type of situation where events are essential, because you can use them to reproduce the user’s behavior and learn what caused the user to leave.

This is what Maxim Kozhnov, Head of Analytics at Azur Games shared with us:

The tutorial and onboarding levels are critical. In my experience with various projects, a significant percentage of “lost” users are abandoning the app at the trial or tutorial stage — the user just closes the app during or immediately after onboarding, and that’s it.

If that’s all the further they went in the game and deleted it, the result is that we don’t get any of the events accumulated during the tutorial. But this is a key spot for analysis and improvement. So although it might sacrifice some battery life, events need to be dispatched immediately at this point. You need to configure the SDK so that events are sent off right away, without waiting for a packet to accumulate.

In the AppMetrica SDK 3.2 (for iOS and Android), there is a new method for force sending the event buffer. Use it in critical user scenarios, like funnel purchases in an e-commerce app, or onboarding in a game. This is crucial.

Search in the user profile

User profiles are a fairly recent addition to AppMetrica, but we already have a useful update to this feature: now you can search for events on a timeline in the user profile.

Analyzing the timeline of events is like watching the user interact with the app. All the events in the session, including crashes and opened deep links, are laid out in order. This is an excellent tool for qualitative app analysis. You can use the search function to filter events so that the timeline only shows the ones you want to study.

As an example, we can look at how an individual AppMetrica user works with the report on operating systems.

First we need to find a session where the OS report was opened, and then we can study the details, like what filters were used to create the report, and how long the AppMetrica user was studying the report.

Try it out for yourself — we think you’ll like it.

Welcome back from vacation!

The AppMetrica Team