GA4 Data Thresholding: Why Rows Disappear From Your Reports

You open a GA4 report for a quiet week, filter down to one campaign, and rows that were there yesterday are gone. The total at the top says visitors came, but the table below cannot account for them. This is data thresholding, and it confuses almost everyone the first time.

What thresholding actually is

Thresholding is a privacy protection. When Google signals or user-provided data is part of your reporting identity, GA4 can associate visitors with demographic and interest information from signed-in Google accounts. To prevent anyone from using a report to identify a specific person, for example by filtering down to one visitor from one small town, GA4 withholds rows whose user counts fall below an internal minimum.

Google does not publish the exact threshold values, and they can vary by report. What is consistent is the behavior: small rows vanish, totals stay, and a warning indicator appears next to the report title telling you thresholding has been applied.

The important thing to understand is that nothing was deleted. The events are all in your property. GA4 is choosing not to show them to you in this particular view.

When it applies

Thresholding kicks in when a report could expose information about individuals. In practice that means:

Reports that include demographics or interest dimensions are the most reliable trigger, since that data comes from Google signals. But thresholding is not limited to those reports. Once Google signals is included in your reporting identity, ordinary reports with low user counts can be thresholded too, which is why small sites and tightly filtered views are hit hardest.

Low-traffic properties feel this constantly. A niche B2B site with a few hundred visitors a month can find entire landing pages and campaigns invisible in its own reports, not because tracking failed but because every row is small.

Explorations are affected as well as standard reports. The BigQuery export is the exception: it contains the raw events and is never thresholded.

How to see more of your data

There are a few practical levers, in rough order of how often they help.

First, change the reporting identity. Under Admin, the reporting identity setting controls how GA4 identifies users across the data. Choosing the device-based option removes Google signals data from reporting, which removes the reason to threshold. This is a display setting, not a collection setting, so you can switch it, read the numbers you need, and switch back without losing anything.

Second, review whether you need Google signals at all. Its main benefits are demographics reporting and remarketing audiences. If you use neither, turning it off, or turning off its inclusion in reporting where that control is available, stops thresholding at the source.

Third, zoom out. Thresholding depends on user counts within the current query, so a 90-day range often shows rows that a 7-day range hides. The same goes for removing filters and comparisons that slice the audience thinly.

Finally, for analysis that must be complete at any granularity, use the BigQuery export. It is free to link on standard properties within BigQuery's own limits, and the event-level data there is exactly what was collected.

The trade-off to be honest about

Thresholding exists because GA4 mixes analytics with data from signed-in Google accounts. That mix is what makes demographics reports possible, and it is also what makes them risky enough to need suppression. You cannot have the ad-platform integration without the privacy machinery it drags along. If you rely on remarketing audiences, thresholding is part of the price, and the levers above are about choosing when to pay it rather than escaping it.

How Analyse sidesteps this entirely

Analyse never collects data that could identify an individual visitor. There are no cookies, no signed-in identity graph, and no demographics pulled from an ad network, which is the design described in the cookieless analytics guide. Because no row in a report can be traced back to a person, there is nothing to suppress: a page with three visitors shows three visitors. Small sites in particular get complete tables at every date range. What you give up is demographics and remarketing, which is the same trade covered honestly in the GA4 comparison.