Data Thresholding

Data thresholding is a privacy control in Google Analytics 4 that withholds rows from a report when the underlying user counts are too small. Google applies it to stop anyone reading a report from inferring the identity or sensitive attributes of an individual user. The data is still collected and still counts toward totals. It is simply hidden from the view you are looking at.

Google is explicit about the limits of your control here: “Data thresholds are system defined. You can’t adjust them.”

Why Data Thresholding Exists

Thresholding solves a re-identification problem. If a report shows one user, aged 25 to 34, female, in a small city, interested in a niche topic, that combination can point at a real person. Withholding the row breaks the chain.

The risk rises with every dimension added. A single demographic breakdown is usually safe across thousands of users. Cross that with city, device, and landing page, and the group in any one row shrinks fast. GA4 evaluates each card and each row independently, which is why one table in a report can be thresholded while the table beside it is complete.

The trade-off is real. Small sites and short date ranges lose the most, because their raw counts sit closest to the threshold in the first place.

What Triggers Thresholding

Thresholding is triggered by the combination of sensitive dimensions and low user counts, not by low counts alone. Google names these triggers:

• Demographics dimensions: age, gender, and interest categories in any report or exploration
• Audiences built on demographic data: an audience defined by age or interest carries the same restriction into every report it appears in
• Search query data: query terms are withheld when the total user count is insufficient
• Narrow date ranges: a 7-day window on a low-traffic property can fall under the threshold where a 90-day window does not

Demographics and interests come from Google signals, which is why “thresholding Google signals” is the way most people first meet the problem. Turning Google signals on unlocks demographic reporting and introduces thresholding to those reports at the same time.

Reporting identity matters less than it used to. The reporting identity setting still shapes how GA4 counts users, but switching it no longer exempts a report from thresholding.

How to Tell Thresholding Is Applied

Look for the data quality icon at the top of the report. Hovering it returns the message that Analytics “has applied thresholding to one or more cards in this report and will only display the data in the cards when the data meets the minimum aggregation thresholds.”

Three signs point at hidden rows even without the icon:

1. Row counts that do not add up to the card total.
2. A dimension breakdown that returns fewer rows than the same report over a longer date range.
3. Numbers that change materially when a demographic dimension is removed from the same query.

Thresholding applies to standard reports, Explorations, and the Data API. It is not confined to the UI, so a connected dashboard or a scheduled export can inherit the same gaps without showing any warning.

How to Reduce Data Thresholding

Thresholding cannot be switched off. Google does not offer a setting, and the threshold values themselves are unpublished. What can change is the query you send.

1. Widen the date range. The cheapest fix. A month of data often clears a threshold that a week does not.
2. Remove demographic and interest dimensions. Most thresholded reports become complete once age, gender, and interests are dropped. Ask whether the breakdown is load-bearing for the decision at hand.
3. Check audience definitions. An audience built on age or interest carries thresholding wherever it is used. Rebuild it on behavior, such as pages viewed or events fired, when the demographic filter is not essential.
4. Reduce the dimension count. Each additional dimension splits users into smaller groups. Two dimensions usually survive where four do not.
5. Use the BigQuery export. Event-level export is not thresholded. Google signals data is not exported to BigQuery at all, so demographic fields will be absent, but the rest of the data arrives whole.

One widely repeated fix is now outdated. Switching the reporting identity to Device-based used to suppress thresholding in many reports. It no longer does so reliably: a report using Google signals features such as demographics is still thresholded under a Device-based identity.

Thresholding vs Sampling vs (other)

These three GA4 limitations produce similar symptoms and have separate causes.

• Thresholding withholds rows for privacy. The data exists and is excluded from your view.
• Sampling estimates results from a subset of events when a query exceeds processing limits. Explorations show a sampling notice with the percentage used.
• The (other) row collapses low-volume values once a table exceeds its cardinality limit. That is a volume ceiling, not a privacy control, and it is explained under dimension cardinality.

The practical tell: thresholding usually clears with a wider date range, sampling clears with a simpler query, and (other) clears only by reducing the number of distinct values.

Frequently Asked Questions

What is data thresholding in GA4?

Data thresholding is a privacy control that hides rows from GA4 reports and explorations when the number of users behind them is too small. Google applies it so no one can infer an individual’s identity or sensitive attributes from a report. The events are still collected and still counted in totals, so the property is not losing data, only hiding it in that view.

Why does GA4 say “thresholding applied”?

That message means at least one card in the report contains a row too small to display safely. The usual cause is a demographic or interest dimension combined with a short date range or a low-traffic property. Hover the data quality icon at the top of the report to confirm, then widen the date range or remove the demographic breakdown to see whether the rows return.

Can you remove data thresholding?

No. Thresholds are system defined and there is no setting to disable them. You can only change the query so it stops meeting the trigger conditions: widen the date range, drop demographic dimensions, simplify the audience, or read the data from the BigQuery export, which is not thresholded.

Does turning off Google signals remove thresholding?

It removes the most common cause, because demographics and interests come from Google signals. It also removes those reports entirely, along with cross-device and remarketing features that depend on the signal. Search query data and other small-count cases can still be thresholded afterward.

Is thresholded data lost?

No. Hidden rows are still collected, still stored, and still included in aggregate totals such as total users and sessions. Thresholding is a display restriction on a specific report view, not deletion, which is why the same data can reappear over a longer date range.

For campaign numbers that do not depend on GA4’s privacy thresholds, linkutm’s analytics dashboard reports clicks at the link level alongside your GA4 reports.