GA4 is more than just a new version of UA.
The name "Google Analytics 4" suggests that it is a simple upgrade of Universal Analytics - but it is not! GA4 is a new data model. As such, the change is similar to a new system. To work well with it, you should know the basic differences in metrics. Here are some of the most important ones.
Not all users are the same
In Google Analytics 4 (GA4), you can view a number of different user metrics, the default of which is the Active Users metric. Active users are those who have triggered an engaged_session event. Confusingly, the default reports refer to these as "users" when they are actually active users.
For a detailed breakdown of different user metrics, it is recommended to use exploratory data analysis. It distinguishes between "total users", "active users" and "new users".
"New users, as the name suggests, trigger the first_open event. The "total users" metric is most similar to the "users" metric in Universal Analytics. However, a difference of up to 10% from the old system is considered normal.
Session to Session with Interactions
A session begins when a user accesses a page and no other session is currently active. It ends when the page is left or after 30 minutes of inactivity. This was already the case with Universal Analytics and remains the case with GA4, with one modification: the session duration is stopped if the user does not interact with the page. This has the advantage of making the data much more accurate. A session with interaction is only counted if the session:
⏱️ lasted at least 10 seconds or
💕 at least two page views or
🤩 includes a conversion.
As a result, the "Average Session Duration" is also changed to "Average Interaction Time". The average session duration represents the total time from entering the page to leaving it, and this time continues even if the page is not in focus. In contrast, interaction duration is based on the engagement_time_msec parameter that is passed along with interactions.
If the page is not kept in focus, no interaction time is measured - the time measurement only starts with the first interaction. Thus, interaction time is the pure time spent on the page and interacting with it, while session time looks at the total time spent on the page.
This has the advantage that it is now possible to analyze more precisely how long users really engage with the site.
What happened to the bounce rate?
The bounce rate is no longer included in GA4's standard reports, nor does it exist in the way it did in Universal Analytics. Instead, the Engagement Rate is reported. The engagement rate is part of the standard reports and shows the percentage of sessions that had an interaction. To be counted as an interaction, the user must be active on the page for at least 10 seconds, visit 2 pages, or trigger a conversion event.
The bounce rate in GA4 would be the opposite of the interaction rate - the percentage of sessions where there was no interaction. So bounce rate plus interaction rate is always 100%.
However, there is one difference: while the bounce rate in UA had no time value, the engagement rate is time-bound. This value can be edited manually, which makes the analysis more flexible. That's why you can't compare GA4's bounce rate with UA's.
Dimensions: First User Interaction & Session
The dimensions Source/Medium, Source, Medium and Campaign are still available in GA4. But here, in the dimension selection, additional attention must be paid to what precedes it. There are 2 possibilities:
Session
First user interaction
The good thing is that both dimensions do what their names say they do. Just don't confuse them in the settings.
💡Also interesting: In exploratory data analysis, you can select the "Source/Medium" dimension under Attribution even without a previous session or first user interaction. This usually results in an error message because the data is estimated based on the selected attribution model. Therefore, we recommend that you select either First User Interaction or Sessions, depending on what you want to query.
Modeled data completes the user journey
In GA4, under Management ⇾ Properties ⇾ Identity for Reporting, there is an option to specify which methods Analytics should use to associate events with users. The Merged option also includes modeled data. However, selecting this option is not enough to use it.
What is modeled data? If a user accepts all cookies, their journey through the website can be well tracked. However, if cookies are rejected, a simple signal is sent to Google that someone is on the site. This signal, also known as a ping, contains no data other than the fact that someone was there, a timestamp, and the referring URL. Based on the behavior of the people who accepted the cookies and pings, a prediction is made. This calculates the likely path of the people who rejected the cookies.
Google's calculation is rather conservative. It assumes that people who reject cookies convert less often than those who accept them. So it is an estimate that enriches the existing data set. As a result, the data is closer to reality, despite the drop-off due to declined cookies.
The downside is that there is no separate analysis in the reports. This means that if the modeled data is turned on, it is no longer possible to filter it out from the actual data.
Checking the box in the settings is not enough to enable it. To use modeled data, you must first have enough base data. In addition, Google Consent Mode must be integrated. This allows Google to track consent and see how many people have declined cookies. If this is built into a page, it should be mentioned in the privacy policy.
Our conclusion
Due to the new way of measuring in Google Analytics 4 and the partly different definitions of the metrics, many values cannot be compared 1:1. Moving to the new system means that we have to say goodbye to some old metrics, but we have many new, exciting ways to read data.
We are happy to help with the change to GA4!