Predictions

Predictive analytics for mobile apps

Schedule Demo

Predictions is a crystal ball for mobile marketers. It lets you to look into the future and see which users will churn and which users will convert. With Predictions, you can shift from being reactive to proactive about how you engage with customers. It allows you to influence users’ future actions by sending the right message, to the right user, at exactly the right time.

 

Predictions_1New

 

Using Predictions is easy. You start by defining what you view as churn or conversion. A conversion can be any action (or set of actions) that a user takes in your app. Churn can be a complete lack of activity for a set number of days or you can define churn as a user not performing a specific action for a set number of days. Once you’ve provided your definition and Predictions has analyzed all your data, you get three sets of insights:

 

 

Predictions_2

 

 

User Grouping: Predictions divides your users into three groups based on their likelihood of churning or converting. Use these groups as audiences for precisely targeted campaigns. For example, you can send a campaign designed to re engage users with a high likelihood of churning or incentivize users on the verge of converting.

4 Ways App Marketers Can Prevent Churn & Increase Conversion

 

Predictions_3

 

Related Behaviors: Learn exactly how the behavior of users who churn/convert is different from those that don’t. See the impact of specific events and event attributes. Most importantly, identify inflection points – the number of times a behavior is repeated before there is a change in a user’s likelihood of churning/converting. Are users who receive three push messages in their first week using your app more likely to churn than those who receive four? With Predictions, you’ll know the answer.

 

Predictions_4

 

 

Related User Attributes: Not only does Predictions show you Related Behaviors, it also shows how a user’s individual characteristics impact their likelihood of churning or converting. Use these insights to identify your most/least valuable users and optimize your acquisition strategy. Additionally, these insights can be used to pinpoint issues with your app’s user experience.

How Product Owners Can Use Predictive Insights to Optimize the App Experience

 

Take Control of Your Company’s Future

Predictions can have a dramatic effect on your company’s churn and conversion rates. Some of our early customers have seen improvements of 7%, 24%, and even 66%! To learn more about what Predictions can do for you, click here.

 

“With Localytics Predictions we have discovered many interesting metrics that affect churn that we wouldn’t have otherwise known. Being able to see and understand the factors that influence churn with Localytics Predictions lets me plan effective campaigns without asking for help from my business intelligence team. As internal resources can be hard to come by, it is great to be able to rely on Localytics to find that information for me.”

LOGO_hungryhouse-55pxErik Kubik
Online Marketing Executive

 

Customer Success Story: Onefootball Reduces Churn by 7.5%

Onefootball is the best app for football fans worldwide. Based out of Berlin, Onefootball is helping fans of over 100 international leagues stay connected to the game with live scores, game highlights, and the latest football news.

Problem: Onefootball wanted a way to identify users with the highest risk of churning, and proactively target them with personalized messages to keep them active in the app during the offseason.

Solution: Using Localytics Predictions, Onefootball was able to identify user behavior patterns that were indicative of churn, which they defined as 30+ consecutive days of inactivity. The Predictions product produced an algorithm that was able to predict a user’s likelihood of churning- low, medium, or high.

Results: Predictions was a huge success for Onefootball, resulting in a 7.5% reduction in churn.

Read their story ›

 

 

FAQs

 

Localytics Predictions gives you the ability to forecast how app users are likely to behave in the future, and proactively communicate with users based on that forecast. Localytics will use historical user behavioral data from your app to create unique algorithms that will help you predict which of your users have a low, medium or high likelihood of churning or converting (aka “likelihood segments”).

We’ll show you these likelihood segments directly in the Localytics Dashboard. We’ll also show you the top related behaviors and user attributes associated with your churn and conversion likelihood segments, and provide you with the tools you need to act on these likelihood segments so that you can prevent users from leaving your app, or alternatively-drive users to convert.

Identify users that are at a high risk of churn, and offering them an incentive — such as access to a new feature — to continue their relationship with your brand and your app.

Discover the key behaviors that are early indicators of user retention; one example of this is from Facebook’s growth team — which found that users who add 7 friends in their first 10 days are more likely to be retained.

Discover the key user attributes that are related to purchase behaviors; for example, users on iPhone 6 Plus devices and AT&T as their network carrier have a higher conversion rate than your average user.

There is a separate, MAU-based add-on cost for Predictions. Please contact your Account Manager or fill out the form above for more pricing details.

We consider a variety of data inputs including the recency, frequency, timing, and volume of user sessions and events, and the interaction of these elements over a user’s lifetime.

Yes. Predictions are created specifically for your users. We will work with you to help you define what churn or conversion means for your business and the algorithms we run will be based off of these definitions.

We recommend at least 2 months of user behavior data at a minimum, but in some cases we can start making predictions using less data. Please know that 2-6 months worth of historical data is ideal and can help us create more accurate algorithms.

We highly recommend setting aside a control group when creating campaigns based on Predictions in order to measure changes in your target metric. Separately, all of our predictive models target an F-score of 0.75.