Learning Google Analytics

Chapter 5. Data Modeling

Data modeling has the potential to be the most technical aspect of a project, since this is typically where you would see machine learning or advanced statistics working across your data. It could also be as simple as making a join across two datasets. It is here where the magic happens in a data project, turning that raw data into information and from there into insight. However, data modeling shouldn’t be the end goal—that is reserved for when we export that insight to the data activation channels, be it analysis in a one-off report or pushing to your data activation channels. Data modeling is a means to an end, not the end to the means. It should be about how you extract value out of your data and never about using the latest technique—it may be that the task that will extract the best value is a simple join rather than a sophisticated neural net. I also think of data modeling as the place where you put your own unique business logic that defines your advantages over your competition. Here’s where you can be creative and bring your own competitive advantage and experience, tailoring how your data is used for the eventual end goal of helping your customers and your business.

There are many ways to model your data outside of GA4, which we will dig into, but we first turn to what GA4 provides natively within its own platform.

GA4 Data Modeling

When using GA4, you can take advantage of some data modeling that comes baked into the product that saves ...