rw-book-cover

Metadata

Highlights

While UA had a steady, albeit less steep, learning curve, GA4’s has been steeper and less well-supported, both by Google and the broader analytics community since it simply was a new tool nobody had any experience with. (View Highlight)

The plentitude of blog posts addressing specific issues with GA4 and its usability is a testament to the tool’s current shortcomings (View Highlight)

The second table, users_* focuses on user-specific data, updating similarly when there’s a field change. Unlike the Pseudo ID table, this one can include data for unconsented users if a user ID is present (which might be considered somewhat problematic from a privacy perspective). (View Highlight)

While the exports contain a lot of expected data points like user properties, device information, and geo data, I am mostly hyped over the Audiences, Lifetime and Predictions fields. The Audiences fields allow you to query a list of User or User Pseudo IDs that are part of a specific audience. This is a great way to save time when attempting to recreate complex audience definitions from the GA4 Audience Builder in BigQuery. Previously you had to export the audience definitions from the GA4 UI and then recreate the actual audience in BigQuery. Now you can simply query the audiences.id or audiences.name fields and get a list of User or User Pseudo IDs that are part of a specific audience. Furthermore, you might want to share these IDs with other tools in your data activation stack, like a CRM or a DMP. The Lifetime fields contain user-level totals for revenue, engagement time, numer of purchases, etc.. This is super helpful if you plan to apply behavioral segmentation techniques to your data. Essentially, you can use these fields to create a user-level RFM (Recency, Frequency, Monetary) model, which can be used to identify your most valuable customers and create personalized experiences for them. Lastly, the Predictions fields contain user-level predictions for the likelihood of a user to perform a specific event. This is a great way to identify users that are likely to perform a specific action, like making a purchase, and then target them with a specific message or offer. Again, having this data at your fingertips out of the box saves you the time and effort of recreating these predictions in BigQuery using BigQuery ML or building out your own pipelines using something like Vertex AI. (View Highlight)

New highlights added 2023-12-27

in 2018, Google re-strucutred its marketing platform, which resulted in GA becoming a part of the Google Marketing Platform (GMP). In the GMP, GA takes the role of a central measurement platform able to share audiences and conversions with the marketing tools in the platform to help optimize ad budgets (View Highlight)

The Lifetime fields contain user-level totals for revenue, engagement time, numer of purchases, etc.. This is super helpful if you plan to apply behavioral segmentation techniques to your data. Essentially, you can use these fields to create a user-level RFM (Recency, Frequency, Monetary) model, which can be used to identify your most valuable customers and create personalized experiences for them (View Highlight)