By - Neal Cole

Using Predictive Audiences in Google Analytics 4:

Remarketing can boost ecommerce sales

In October 2020, Google launched its machine learning web analytics platform, Google Analytics 4. GA4 not only offers advanced web analytics, but its predictive audiences can be used to identify which users are most likely to convert or churn.  This allows marketers to create more effective campaigns in Google Ads for both remarketing and re-engagement.

GA4 could also allow marketers to make higher bids for keywords if they are triggered by a predictive audience that has a higher propensity to convert. If you haven’t already upgraded to Google Analytics 4 this may be a reason not to delay anymore. Predictive audiences deliver a valuable tool for conversion rate optimisation to help grow the business and improve conversions.

 

1. What are predictive audiences?

 

GA4 automatically enriches data using machine-learning to predict the future behaviour of your visitors. GA4 currently has three predictive metrics for building predictive audiences.

 

Purchase probability: The propensity that a user who has been active in the last 28 days will convert in the next 7 days. Only purchase and in_app_purchase is currently supported.

Churn probability: The propensity that a user who has been active within the last 7 days will not be active in the next 7 days.

Revenue prediction: The estimated revenue from all purchase conversions in the next 28 days from a user who has been active in the past 28 days.

 

2. Prerequisites:

Given predictive metrics rely on machine-learning, GA4 requires a certain number of conversions to activate these metrics. To successfully train the predictive models, over a 7-day period GA4 needs a minimum of 1,000 returning users who had previously converted (purchases or churned users) in the past 28 days, and also 1,000 non-converters over the same period.

 

These volumes of conversions must be maintained over a period of time for a GA4 property to be eligible for predictive metrics. If your property meets the criteria, the predictive metrics will be generated for each active user once per day.

 

3. Create predictive audiences:

To build predictive audiences go to your GA4 property and navigate to ‘Audiences’ and click on ‘New audience’.

1. Audiences in Google Analytics 4

This takes you to the ‘Build a new audience’ interface where you will have the option to create an audience from scratch or use ‘Suggested audiences’. Select the ‘PREDICTIVE’ tab on the Suggested audiences section.

2. Build a new audience for predictive audiences

Depending upon the volumes of data being received by your GA4 property, you will get an eligible or not eligible to use message below each audience.

3. Predictive Audiences

There are currently 5  suggestions for predictive audiences:

    • Likely 7-day purchases. Users with a high propensity to purchase in the next 7 days.
    • Predicted 28-day top spenders. Users who are estimated to generate the most revenue in the next 28 days.
    • Likely first-time 7-day purchasers. Users with a high propensity to make their first purchase in the next 7 days.
    • Likely 7-day churning purchasers. Purchasing users who are unlikely to visit your site or app in the next 7-days.
    • Likely 7-day churning users. Active users who are unlikely to visit your site or app in the next 7-days.

You cannot modify the predictive condition for each audience, but you can add new non-predictive conditions, such as Device category equals mobile.

 

4. How to use predictive audiences?

When you create predictive audiences, make sure you use a Google account that has permissions for your Google Ads account. Predictive audiences will then be automatically shared with all Google Ads accounts linked to your GA4 property. Google suggests two ways of using predictive audiences in remarketing campaigns and re-engagement campaigns.

 

4.1 Remarketing audiences:

Predictive audiences are ideally suited to remarketing campaigns because GA4 uses machine learning to identify deep patterns of behaviour that are unique to your site or app which indicate  the user is likely to convert. The ‘Likely 7-day purchasers’ are the ideal audience for a remarketing campaign. A persuasive follow-up message delivered to these users could be the trigger they need to complete a purchase.

 

4.2 Re-engagement campaigns:

Re-engagement campaigns can help maintain engagement with your business among users who are showing waning interest in your products or services. The ‘Likely 7-day churning users’ are a cohort of users who need to be re-engaged and would benefit from a strategic message or special offer to reverse a decline in engagement.

 

4.3 Strategic bid adjustments:

 

You could use predictive audiences that are likely to purchase to trigger keyword bids. This should allow you to place higher bids on keywords because these users will have a higher propensity than normal to purchase.

 

4.4 New customer campaigns:

 

Predictive audiences can also be used to grow your customer base. Try using the ‘Likely first-time 7-day purchasers’ audience as a means of attracting new customers across the Google Display Network, Gmail, YouTube and the Search Network.

 

Conclusion:

 

Predictive audiences offer marketers an opportunity to improve targeting and conversions in the digital space. Testing of predictive audiences is required to build up data to prove their effectiveness, but as the machine learning processes more data it should become more accurate at predicting outcomes.

 

  • About the author:  Neal  (@northresearch) provides web analytics and CRO consultancy services and has worked in sectors including financial services, online gaming and retail. He has helped brands such Hastings Direct, Manchester Airport Group Online and Assurant Inc to improve their digital marketing measurement and performance.