Data thresholds are a privacy feature in Google Analytics 4. When a report or exploration includes data that could reveal the identity or traits of individual users, GA4 withholds rows below a system-defined threshold. They most often appear when Google signals is enabled, when demographic dimensions are in use, or for small segments. The threshold values are set by Google and cannot be changed.
How to tell thresholds are applied
GA4 shows an indicator near the report title when thresholds are active. Totals may look lower than expected and some rows, such as rare search terms or small countries, may be missing. The data is collected; it is withheld from that view.
What triggers them
Thresholding is tied to identity and sensitive breakdowns.
- Google signals enabled and included through the reporting identity
- Demographic dimensions such as age, gender and interests
- Small user counts in the selected date range or segment
- Site search terms and similar text dimensions with rare values
Reducing their impact
Switching the reporting identity to device-based often removes thresholds caused by Google signals, at the cost of cross-device stitching. Widening the date range raises counts above the threshold. For exact numbers, the BigQuery export contains the raw events without thresholds, though it does not include Google signals data.
Example
Missing rows in a regional report
Suppose Hearthgrid, a facilities management tool, filters a report to a small market and sees almost nothing. The thresholding indicator is on because Google signals is active. Changing the reporting identity to device-based restores the rows, and the team notes that cross-device user counts will differ.
Common mistakes
Assuming data was not collected
Thresholded data exists but is withheld from the view. Check the indicator before debugging tags.
Enabling Google signals without a reason
If you do not use its features, signals mainly adds thresholding. Turn it on deliberately.
Questions
No. Google sets and adjusts the thresholds. You can change reporting identity, date ranges or data sources to work around them.
