Predicting User Loss, for Enhanced User Retention - Huawei Developers

An app's ability to retain users has a major impact on its profitability. Retaining users has always been a major challenge for developers, given the presence of competitor apps, and the willingness of today's users to shop around to find the best possible experience.
Therefore, developers have needed to invest more in marketing in order to acquire users. Furthermore, paid traffic has proved even harder to retain, as such users do not tend to install apps solely based on their original needs. Fortunately, there are effective ways to prevent user loss and maximize the value of paid traffic.
This article will highlight a few case studies which illustrate how the Prediction service can help you predict user loss and retain users.
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
The figure above shows how a game app was able to improve its user retention, by following just a few steps in the Prediction service. The app's original retention rate was low, and thus its developers enabled the Churn prediction model in Prediction to identify users who demonstrated a high churn probability. They then utilized Prediction to gain actionable insights into the characteristics and behavior of these users, saved them as an audience, and then pushed tailored messages to engage with them on a more targeted basis. This process can be broken into several different stages, and we'll address the key stages for the Prediction service.
1. Precise user targeting
If a user is about to churn, they will likely become less active. The user's activity can be measured based on behavioral data, such as the daily launch rate, launch frequency, and use duration.
Let's take ARPG games as an example. Typical active players of such games tend not to launch the game many times a day, but will stay in the game for relatively lengthy periods of time after each launch. For games in this genre, we regard players who launch the game 2-3 times and play about 5 rounds a day as active players. If an active player has not launched the game very often over the course of a week, or even failed to sign in for two consecutive days, the Prediction service, based on the churn prediction model trained via machine learning, will determine that the player has a high churn probability in the next week.
The Prediction service uses the active user data over the most recent two weeks to train the churn prediction model, which is used to predict the probability that active users of the app in the past week will be lost during the next week. It's important to note that users who are inactive the next week, or uninstall the app are considered churned users.
The Prediction service does not require any high-level algorithm expertise to implement user targeting. Instead, prediction tasks are automatically executed in the background on a daily basis.
2. In-depth insights into audiences
For a targeted audience, the attributes and behavioral characteristics are analyzed, in order to implement a more actionable operations plan for better user retention.
*The data provided is for your reference only.
*The data provided is for your reference only.
This figure is an example of the background data in the Prediction service. It shows that users with high churn probabilities still have game sessions from the past seven days, though these are infrequent. In addition, Prediction enables you to query such users by device model, making it easier to reach them via periodic message pushing, in HUAWEI Push Kit, which users their most recent usage times.
3. Reaching users more effectively
For a targeted audience whose characteristics have been learned, you'll need to implement the right measures to retain its users. HUAWEI Push Kit, mentioned above, is the most handy method for doing so.
An audience generated by the Prediction service can be directly applied by HUAWEI Push Kit to target users. In the case of retention, push messages are sent specifically to users who demonstrate high churn probabilities. Common examples of this include the pushing of time-limited game gift packages, new game characters skills, and even simple greetings.
Prediction tasks incorporate data that is updated on a daily basis. As a result, push messages can be sent periodically to continuously engage recent users with high churn probabilities.
Acquiring traffic is more challenging than ever, but thanks to intelligent data technologies, you can now conduct refined operations that help you extract the most value from existing users. The Prediction service and Push Kit are indispensable for helping enterprises effectively improve ARPU, while also promoting digitization and intelligence across the board.
To learn more about Prediction, please visit our official website.
For more details, you can go to:
l Our official website
l Our Development Documentation page, to find the documents you need
l Reddit to join our developer discussion
l GitHub to download demos and sample codes
l Stack Overflow to solve any integration problems
Original Source