And the Threshold? Nobody talks about the threshold.
The threshold: That GA4 limitation whereby entire rows of data are “hidden” when the system sees so few users that it thinks you could identify specific individuals.
This applies to both the granular and aggregate systems, which is why we haven’t mentioned it in the previous points. However, it is another limitation to keep in mind, and it doesn’t affect the third system we’ll discuss now.
Exports to BigQuery: The Power of Raw Data
You’ve probably heard that BigQuery is “what you absolutely have to use to get the most out of GA4.” I couldn’t agree more, and at the same time, I couldn’t disagree more with this statement.
BigQuery offers a third reporting system by exporting the raw data captured by GA4 to a cloud database. This system is ideal if you need to work with data without the sampling, cardinality, or retention limitations that affect other systems.
The advantages of using BigQuery include:
There are no limitations on sampling, cardinality, thresholds, or data retention. You have the data here, and you can only limit it if you want. So GA4 doesn’t downgrade your data quality for various reasons.
It allows you to cross-reference GA4 data with other datasets, giving you a more complete view. It’s a SQL database, which means you can upload data from your business or campaigns and get a much more complete picture of what’s happening on your website or app.
It gives you the freedom to create your own custom metrics and analyses. Again, you can do whatever you want with the data, and if, for example, conversions are per user for you, you can create that metric, which doesn’t exist in GA4.
However, this system also has its challenges. Enough to make you not want to use it.
The database doesn’t offer Google’s consolidation and machine learning capabilities. You won’t have blended user data, Google Signals dimensions like gender, age, or affinities, or any of the technologies Google uses to enhance your data. If something isn’t captured through direct hits, you won’t have it in BigQuery.
Attribution is more limited. While last-click attribution data is now available, it still isn’t fully synchronized with Google Ads, Campaign Manager, and other tools. There’s also no session attribution data modeling, nor is there a data-driven attribution model (which is the most commonly used). You won’t be able to report any of this with BigQuery.
