Mobile: Uninstall Tracking

Uninstall tracking is commonly used to understand mobile churn. This metric can identify actions that lead to uninstalls, and it serves as an immediate indicator of users leaving. While Mixpanel doesn’t offer uninstall tracking out of the box, it is possible to build custom scripts that collect uninstall information and import it into Mixpanel. Uninstall data can then be viewable in Mixpanel reports.

It is important to assess KPI’s, business goals, and alternative tracking methods before building uninstall tracking, as there are often more meaningful metrics that are much easier to implement.

Tracking Uninstalls

It is possible to detect whether or not an app is installed on a mobile device by sending a push notification and verifying that it was successfully delivered. If the push notification isn’t delivered, then the app may have been uninstalled. The list of users that couldn’t receive the message can be collected and imported into Mixpanel. This will allow the viewing of uninstall information in Mixpanel reports.

The process is similar across mobile platforms, and includes building and hosting a script that can send empty push notifications, collecting the users that do not receive it, and formatting that list. While this process is the same across platforms, there are some differences. Depending on platform, the necessary script will need to execute the following actions:

iOS:

  1. Export all People profiles from Mixpanel and collect the device IDs of all users.
  2. Build and send a push notification with no payload (i.e. a silent push notification) to all users.
  3. Collect all user devices that do not receive the empty push notification. This is done through the The Apple Push Notification service (APNs), which will:
    1. Prompt the client to collect information from the application.
    2. If the application is no longer installed on the client, then the client will return an error to the APNs.
    3. An error will return to the developer through the APNs Feedback Service. The APNs feedback will be in JSON format, and report a specific error signifying that the application is no longer installed.  
  4. Identify and create a list of all Mixpanel profiles that no longer have a device ID installed.
  5. For each Profile that has uninstalled the application, update the "$ios_devices" reserved People property to remove that device_id from the list. Also, set a new People property named "Date of Last Uninstall" and increment a People property named "Number of Uninstalls".

Android:

  1. Export all People profiles from Mixpanel and collect the device IDs of all users.
  2. Build and send a push notification with no payload (i.e. a silent push notification).
  3. Collect all user devices that do not receive the empty push notification. This is done through Google’s Firebase Cloud Messaging (FCM), which will:
    1. Prompt the client to collect information from the application.
    2. The client will query the local Package Manager to see if the target application is installed to accept the message. A false flag will be returned if the application is uninstalled.
    3. The client will return a message to the FCM server, which will in turn report a JSON formatted message to the developer indicating that the app has been uninstalled.
  4. Identify a list of Mixpanel profiles that have device IDs that are no longer installed.
  5. For each Profile that has uninstalled the application, update the "$android_devices" reserved People property to remove the that device_id from the list. Also, set a new People property named "Date of Last Uninstall" and increment a People property named "Number of Uninstalls".

Alternatives to Tracking Uninstalls

Understanding churn is often the main motivation behind tracking mobile application uninstalls. While this can be an informative metric, there are often other metrics that better encapsulate user behavior and factors that lead to churn.

Common goals associated with understanding churn through uninstalls include tracking churn in real time and building correlations between churn and user actions. Fortunately, uninstall tracking is not the only way to do this. Retention reports offer the same, if not better, insight into churn. Some specific metrics that highlight this include the tracking of:

True Churn

Uninstall tracking is often viewed as the true mechanism to track churn. It is not a perfect solution, however.

While it is true that many users uninstall an application because they are leaving the product, there are many users who uninstall an application but are not churning. For example, a user who purchases a new device and clears the old device of data will be tracked as an uninstall. This user can install the application on the new device, but will be counted as someone who has churned.

Tracking user retention is superior in this regard, as there will be no data discrepancies stemming from false-positive uninstall events. While retention is not as specific of a metric, it still provides accurate user engagement information without introducing as many data discrepancies.

Real Time Churn

Another argument for tracking uninstalls is that it provides immediate, real-time churn data. Barring edge cases, this is true, but it won’t actually help prevent churn. Users who stop using an application but still have it installed are at risk for churn. If they uninstall the application, they have left the application and will be harder to win back.

Alternatively, tracking retention includes users that are at risk for churn, while also encompassing those who have uninstalled the application. This all-inclusive method of tracking can facilitate the creation of campaigns, product changes, and other actions that can keep users from leaving an application in the first place.

Events That Lead to Churn

Creating actions to retain users is often the main value behind understanding churn. Therefore, uninstall tracking is often built with the intent of correlating specific actions with uninstall events. This is an actionable insight that catalyzes change needed to retain users. This being said, the correlation between user actions and user retention can track user engagement just as well as that of user actions and uninstalls.

Mixpanel’s Signal report is a perfect way to determine correlation between events and engagement. Determining correlations between user actions and retention will expose the features that are driving customers away from an application.

 

While uninstall tracking can serve as a method for understanding user engagement and churn, it is often not the best one. Mixpanel Retention can provide a more encompassing understanding of user engagement, and it can do this out of the box.

Find your answer?
0 out of 0 found this helpful

Comments

0 comments

Article is closed for comments.