Data models are just something for real data nerds?
Not this time. For Google Analytics 4 understanding the data model is important for everyone.
Why?
Because moving from Google Analytics Universal to Google Analytics 4 is basically moving to a new analytics service. It’s not an incremental update but a completely new setup.
The major change is the data model.
That is why I did a podcast episode about the data model change. Link in the comment.
In GA4 we have now, what we know already from Firebase, Amplitude or Mixpanel: an event & user (unique identifier) based model.
What does that mean?
First of all the user (or account, or device) becomes more important because it replaces the good old session (mostly) - so adding user attributes is a good idea because all events inherit them (example: has_subscribed_to_newsletter, current_plan).
Then events become first class citizens and are pushing the good old pageviews to second class (a pageview is just a specific event). And now we can use the more organic model of event name and event attributes. No more data stuffing in the event label dimension.
But GA4 keeps some weird stuff around events:
- To see the attributes in the reports you have to assign them to custom dimensions - (honestly, this drives me crazy)
- why did they not call it properties like everyone else does. No in Ga4 they are called attributes…
So how does it change. In most cases not so much. Here is a simple example:
We have a newsletter subscribe page, where a user can subscribe to a newsletter and select some topics she likes
In Universal analytics we would have implemented it like this:
Category: Newsletter
Action: subscribed
Label: topic 1, topic 2
In GA4 it would be this:
Event name: newsletter_subscribed
newsletter_topics: topic1, topic2
Not so much different but much clearer at least when it comes to the topics.