Predict user behavior with Google Analytics 4

A new way to target even more efficiently

With the introduction of Google Analytics 4 as the new tracking standard, users of the tool now have many new options for collecting, analyzing and activating data. One feature is particularly exciting: predictive metrics, i.e. measured values based on forecasts. With this function, the collected data can not only be used for reporting purposes, but also for many other marketing measures.

Simply analyzing how many visitors have bought a product on a website is old hat and has long been part of the standard data collection repertoire. But what if we could also get predictions about who is most likely to make a purchase in the future - and if we could target them specifically with the help of Google Ads? That's right, our Google Ads would have a special effect on this target group. After all, these users only need a little nudge to turn their purchase intent into a purchase. With Google Analytics 4, this is easier than ever before. Machine learning makes it possible.

Using probabilities to maximize sales

Predictive metrics can be used to predict which visitors are most likely to plan a purchase. Users who are highly unlikely to visit a website again can also be identified. Google Analytics 4 uses three different predictive metrics for this purpose:

Purchase Probability

• The so-called purchase probability states: A user has been active in the last 28 days and will make a purchase within the next seven days. GA4 takes into account the events "e-commerce purchase", "in-app purchase" and "purchase".
  ‍

Churn Probability

• This probability is an indicator that a user has been active in the last seven days and will be inactive in the following seven days.
  ‍

Revenue Prediction

• This is a forecast of how much revenue can be expected in the next 28 days from the purchases of users who were active in the previous 28 days.
  ‍

‍

Can my website use predictive metrics?

GA4 creates predictions with the help of machine learning. Therefore, Google Analytics must first learn: How do users who make a purchase interact with the website? And how do those who are inactive afterwards interact? To do this, Google needs continuous data on which to base its forecasts. A few basic requirements must be met for this:

• Within the last seven days, at least 1000 users must have made a purchase and at least 1000 users must not have made a purchase.
• The model quality must be constant. This means that the number of users who make a purchase or do not make a purchase fulfills the above requirements over a period of at least 28 days.
• The purchase event must be implemented on the website and send data to the GA4 property.
  ‍

More precise targeting for even greater campaign success

With these forecasts, target groups can be created and used as targeting for Google Ads: so-called predictive audiences. Depending on the objective, there are various options for designing successful campaigns. For example, predictive audiences can be used to define a target group consisting of users who are most likely to make a purchase. By integrating this target group into Google Ads, you can increase your reach. This is because the campaigns are specifically targeted at new users who are similar to people who are already likely to buy there. This is a much more efficient method of increasing the target group than simply increasing budgets, buying more keywords or expanding the target group criteria.

On the other hand, predictive audiences are also valuable for identifying users who have already engaged with the website or made a purchase and are at risk of churning. A target group of users who are likely to be inactive in the next seven days can improve customer retention. Re-targeting users with special content or offers prevents them from churning and builds a loyal customer base at the same time.

Predictive Metrics and Audiences add a really powerful function to GA4, transforming it from a mere analysis tool into a data activation tool. We at antwerpes healthy media will be happy to help you check whether this new function is suitable for your website. We will also support you in setting it up and, of course, in campaign planning so that you can utilize predictive audiences as efficiently as possible.

Please feel free to contact me: