Predictive Audiences is a Google Analytics feature that employs machine learning to anticipate future user behavior. This tool lets you build audiences based on predictive metrics. These predictive indicators are computer models that estimate future behavioral probabilities based on previous behavioral data from users. For example, you can build a “likely to buy in the next 7 days” prediction audience. This demographic contains people expected to purchase during the next seven days.
What kinds of audiences can be predicted
Let’s first look at the three indicators used in the GA4 predictive audience:
Purchase Probability: This refers to the likelihood that a user who was active in the previous 28 days will record a specific conversion event (such as a purchase) within the next 7 days. This measure assists in determining which users are most likely to purchase shortly.
Churn Probability: This is the likelihood that a user who was active on your app or website in the previous 7 days will no longer be active in the following 7 days. This indicator is used to identify users who are going to churn, allowing measures to be taken to enhance retention.
Predicted Revenue: This is the revenue predicted to be generated in the next 28 days from all purchase conversions made by users who were active in the previous 28 days. This measure is useful for forecasting future revenue and designing marketing and sales strategies.
Currently, only purchase/ecommerce_purchase and in-app purchase (in_app_purchase) events are supported for purchase probability and revenue prediction metrics. Although Google Analytics will still process e-commerce purchase events, using purchase events is now recommended.
Based on the three indicators of purchase rate, churn rate, and predicted revenue, GA4 provides users who are likely to complete the purchase within 7 days, users who are likely to complete the purchase within 7 days (users who purchase for the first time), users who are likely to churn within 7 days, users who are likely to churn within 7 days, and predictions of 5 types of audiences in total, including users who are likely to churn (users who have completed purchases) and users who generate the most purchase value within 28 days.
How eligible are you to use GA4 predictive audience
To use the Predictive Audiences feature in Google Analytics, there are a few things you need to meet:
Data quality and quantity: There needs to be a certain number of positive (such as completed purchases) and negative (such as incomplete purchases) user behavior instances so that the model can learn to distinguish different user behavior patterns. There needs to be at least 1,000 returning users who triggered the relevant prediction conditions (such as purchase or churn) during a consecutive 7-day period within the past 28 days, and at least 1,000 returning users who did not trigger these conditions.
Model quality maintenance: The quality of a predictive model must remain stable over some time to be able to generate effective predictive indicators.
Google Analytics account settings: Your Google Analytics account needs to be configured correctly to collect relevant user behavior data. For example, if you want to use a purchase probability metric, you need to make sure you’re tracking purchase events.
Privacy and Compliance: Ensure your data collection and use comply with all applicable privacy regulations and standards.
Once you meet these criteria, you can create and use predictive audiences in Google Analytics to make your marketing campaigns more targeted and effective.
How to build your predictive audience
Like other audiences, if we need to use Google Analytics audiences, we need to associate Google Analytics with the Google Ads account so that the audiences we create in Google Analytics can be automatically synced to the Google Ads account.
To create a predictive audience, it’s easy. We only need to enter the created interface through the path Admin-Data display-Audiences-New audience. Then we can see the predicted audience by selecting Predictive in Use a Reference. If we meet the conditions for use, “Ready to use” will be displayed below the audience.
Click on the predicted audience we want to create, and enter the audience settings. Google Analytics has set the conditions for us, and we can name the audience and keep it directly.
Of course, we can also customize the prediction conditions. We only need to click on the blue text in the default settings, such as “95th percentile,” to enter the setting interface.
You can set the desired prediction conditions according to your own needs. When setting the prediction conditions, we can observe the icon on the right so that the set results meet our needs.
The difference between GA4 predicted audiences and other audiences
There are several key differences between predictive audiences and other types of audiences in Google Analytics:
Based on predictive models：
Predictive Audience: Use machine learning algorithms to predict users’ future behavior based on their historical behavioral data. For example, predicting which users are likely to purchase or churn in the future.
Other audiences: Typically, based on static criteria such as the user’s geographic location, pages visited, device type, etc.
Predictive Audiences: Audience members may change dynamically as the model continues to learn and the data is updated.
Other Audiences: Once defined, membership generally remains unchanged unless the criteria are manually updated.
Data analysis depth：
Predictive audiences: Using deep data analysis and pattern recognition, complex relationships and trends in user behavior can be revealed.
Other audiences: Rely on more intuitive, superficial characteristics of user behavior.
Predicting audiences: Especially suitable for dynamic market environments and complex user behavior patterns, such as purchase predictions for e-commerce websites.
Other audiences: For broader scenarios, such as marketing campaigns targeting specific regions or optimizing for users on specific devices.
Overall, predictive audiences provide a more dynamic, in-depth, and precise way to identify and target potential customer segments through the use of advanced data analysis techniques.
Application scenarios of GA4 predictive audience
Imagine if you could know in advance which users are most likely to purchase in the next 7 days, or which users are on the verge of churn, how would you act? We’ll take you through how to accurately identify and influence these key user groups through Google Analytics’ predictive audiences feature.
Improve conversion rate of e-commerce website: Use predictive audiences to identify users “likely to make a purchase in the next 7 days.” Target these users with customized promotions and coupons to increase the likelihood of purchase.
Reduce user churn rate: Identify users who are “likely to churn within the next 7 days.” Implement re-engagement strategies for these users, such as sending emails reminding them of items in their incomplete carts, or offering special discounts.
Optimize mobile application promotion: Find out new users who are “likely to make their first purchase in the next 7 days” through audience prediction. Show these users guided ads that introduce your app’s core features and benefits.
Improve advertising efficiency: Identify users who are “expected to spend the most in the next 28 days.” Spend more of your advertising budget on this group of users to increase ROI.
Increased customer loyalty: Analyze users who “may churn but have a high purchasing frequency before”. Target these users with loyalty rewards or exclusive services to enhance their brand loyalty.
Through these examples, you can see how predicted audiences can help companies more accurately target potential high-value users and adopt targeted marketing strategies to improve results.