Flurry Demographic Estimates is a new feature of Flurry Analytics that can help you understand the demographics of your application users. By using the collective knowledge of all the data in the Flurry system, Flurry Estimates calculate an estimate of the age and gender breakdowns of your users.
Viewing Flurry Estimates
To view the Flurry Estimates for a given application, simply navigate to the Audience tab and select either Age or Gender.
Figure 1. Selecting Age or Gender from the Audience Tab
If you already were collecting your own demographics data you will notice there is now a drop down that you can use to switch between your Reported Data and the Flurry Estimates for each. If you don't currently report age or gender data you will be taken to the Flurry Estimate by default.
Figure 2. Selecting Flurry Estimates from the selection drop-down
As you can see in Figure 2, the Flurry Estimate is an easily understood percentage breakdown of your users.
Frequently Asked Questions
How does Flurry estimate age and gender?
When you integrate the Flurry SDK into your application you have the option for using the setGender() and setAge() functions to set the age and gender of the user if you collect that as part of your application. If you use these optional functions the Flurry system will compute analytics for your application showing the breakdown of genders and ages.
However, not all applications collect user age and gender as part of the application experience, making it difficult to collect such demographic information. Flurry Estimates solves this problem by using the collective knowledge of all devices tracked by Flurry Analytics to estimate the demographics of all applications. This estimation is the result of research done by Flurry over a long period of time and utilizes numerous machine learning techniques designed to work with the unique aspects of mobile application usage and consumption.
How accurate are the estimates?
The accuracy of Flurry Estimates will vary by application based on the number of users with known demographics across the Flurry network. You can see in Figure 2 that there is a breakdown of "Known" and "Unknown" users for the Flurry estimate. The higher the number of "Known" users the more accurate the Flurry estimate will be.
What do "Unknown" and "Known" mean?
As mentioned above, there is a confidence threshold required to make a Flurry Estimate. There will be users of your application where the Flurry system is not confident enough in its prediction to make an Estimate and these users are considered "Unknown" and excluded from the metrics. All of the users included in the estimate are considered "Known".
Keep in mind that the "Known" users may not be a representative sample of all users of your application.
How can I pass Flurry age and gender for my reported data?
Flurry Estimates will never replace directly collected demographic data from your users. If you find the demographic estimates to be useful we encourage you to investigate using the setGender() and setAge() functions in the Flurry SDK. More information is available in the documentation for the Flurry SDK you are using for your application.