A Deep Dive into Predictive Metrics in Google Analytics 4 (GA4) - Digital Analyst Team

Get in touch

Awesome Image Awesome Image


A Deep Dive into Predictive Metrics in Google Analytics 4 (GA4)

Writen by Magesh

comments 0

Predictive Metrics in Google Analytics 4

Released in 2020, Google Analytics 4 has businesses staring down a long list of endless possibilities and opportunities. Thanks to the features that come with the new property, people can now enjoy the benefits of Machine Learning.

GA4 takes the regular Universal Analytics and takes it a step further with its Predictive Metrics, a feature used to quantify the advancement of a conversion. GA4 utilizes Machine Learning to collect and analyze data on your customers or users. The analysis results help predict future purchasing behaviour.

Types of Predictive Metrics in Google Analytics 4

Predictive metrics enable you to pinpoint which users are highly likely to convert or make a purchase. The opposite is also true; the metrics allow you to identify which users have a low probability of converting.

GA4 uses three Types of Predictive Metrics:

Purchase Probability

Purchase probability is the likelihood that a user active in the previous 28-days will trigger a purchase within the following seven days. You should note that GA4 supports only three events: ecommerce-purchases, in-app purchase events and purchases.

Churn Probability

The likelihood that a user who was active in the previous seven days will be inactive in the following seven days.

Revenue Prediction

It predicts the revenue anticipated in the next 28 days from active user purchases in the previous 28-days.

Types of Predictive Metrics in Google Analytics 4


Since GA4 is a sophisticated system tied to machine learning, you must fulfil a set of requirements to train the predictive metrics.
These qualifying requirements are more of a list of standard prerequisites that need to be met to trigger predictive metrics. They are:

  • The least possible number of positive and negative samples is 1,000, with the least number of non-operative or inactive users as 1,000 as well. By positive, we mean that you have 1,000 purchasers or conversions, while negative means that the other 1,000 users haven’t purchased anything.
  • Google requires that Model quality be maintained over a particular duration. By model quality, we mean that users’ traffic triggering purchase events remains regular over time. Although Google failed to specify the period required, the usual number of days is often 28.
  • The GA4 uses three types of predictive metrics- Purchase, Churn and Revenue Prediction- as highlighted above. To trigger these metrics, you must set up the purchase event and send it to the GA4 property.

Remember, once you’ve met the prerequisites, you’ll receive predictive metric updates daily. However, when the numbers fall below the standard requirements, you stop getting updates.

How To Use Predictive Analytics

Head to GA4. Once there, head over to Suggested Audiences to use the templates in Audience Builder.
You’ll have three choices: General, Template and Predictive. Click on Predictive to create your Custom Audience. You can group them to suit your preferences. For example:

  • Likely 7- Day Purchasers
  • Likely 7- day Churners
  • Likely 7- day Churning Purchases

Google Analytics 4 will provide the below options for you.
Below is how you can use the property:

  1. Go to GA4. Under the Analytics report menu, click Explore.
  2. From the many exploration templates, click the (+) plus sign to choose a blank template.
  3. Under the newly opened console, click Tab Settings on the drop-down under Technique.
  4. After clicking the drop-down, you’ll be given several options; choose User Lifetime.
  5. Under Variable Tab, click the (+) plus sign under Metrics.
  6. A search overlay will appear with a list of options below it. Click on the Predictive Metrics option.
  7. After you’ve done the above, a drop-down menu under the Predictive Metrics option will appear. It will have Churn, Purchase and In-App Purchase Probability.
  8. You can check them and select whichever sub-metrics fall under the main metric options. The sub-metrics are options in the form of different Percentiles, including an Average Sub Metrics Option.
  9. Double click on any of the metric choices. Double-clicking will add them to the Analytics report.

That’s it; you have created a User Lifetime report using predictive metrics.
Predictive Analytics in GA4

Keep Using Predictive Metrics in GA4

Google Analytics 4 utilizes machine learning to learn its users’ behaviours and then uses that information to help you understand and predict their future behaviour. Such data can be leveraged by marketers looking to maximize their campaign efforts or increase their engagement efforts.

Whichever the case, you’ll be able to identify which groups to focus more of your efforts on or when to launch a product.

Tags : ,