User Stickiness in GA4

1. The Case of the Missing Fields

GA4 Reports and Explorations aren’t just two different ways of looking at the same table.

They actually support different fields. If you’ve built a standard report using specific dimensions or metrics and then try to “open in Explorations,” you might notice some data simply vanishes.

When a field isn’t supported in the Exploration interface, GA4 just drops it.

If your visualisation depends on that specific metric, it won’t appear at all.

2. Filtering Isn’t Universal

We often think of “search” and “filter” as the same thing, but in GA4, they have very different personalities.

In a standard report, the search box (the one sitting between your graph and your table) acts as a “contains” match.

It isn’t case-sensitive.

But when you move to Explorations, the rules get stricter. Filters there are case-sensitive and require you to choose a specific match type (like “exactly matches” or “begins with”).

Expert Tip: If you’re searching for “Email” in a Report, you’ll see results for “email” and “Email.” In an Exploration, you’ll only see what you typed.

3. The Retention “Hard Stop”

This is a big one that often catches people out. Standard Reports can often access data further back in time because they rely on aggregated data tables.

However, Explorations are limited by your property’s Data Retention settings.

If your property is set to the default 2 months and you try to run an Exploration for a 6-month period, you’re going to see a lot of zeros where the old data used to be. Standard reports might still show the totals, but the granular Exploration won’t play ball.

4. Behavioural Modelling and Machine Learning

If you have Consent Mode set up, GA4 uses behavioural modelling to fill in the gaps left by users who declined cookies.

When this is active, the machine learning algorithm has to work across two different data sets:

1. Aggregated tables for your standard Reports.
2. Raw event and user-level data for Explorations.

Because these two data sets have different structures, the algorithm might produce slightly different results for each.

The more users you have declining consent, the higher the probability that these two numbers will drift apart.

5. The “Freshness” Factor

You know how it is when you’re trying to report on a campaign that only launched yesterday? Data in Analytics comes from various systems and isn’t all processed at the same time.

If you are running queries for the last 48 hours, you’re essentially looking at “unsettled” data.

Reports and Explorations might hit different processing systems at different times, meaning a refresh five minutes later could change the gap between the two.

6. Thresholding and Low User Counts

If you have a low user count in a specific date range, Google might withhold certain data to protect user privacy. This “thresholding” can be applied differently depending on whether you’re looking at a summary report or a granular exploration, especially if signals are enabled.