Thresholding is when Google Analytics 4 withholds some rows of a report because showing them could reveal something about an individual visitor. The report still loads, but certain figures are missing or lower than the true count, and a notice tells you data has been held back.
How thresholding works
GA4 applies thresholds automatically when a report could single out a small number of people. It is most common in three situations:
- reports using demographic or interest dimensions such as age, gender or interest categories, which GA4 draws from Google signals;
- short date ranges or narrow segments where only a handful of users are present;
- reports combining potentially identifying dimensions, such as search terms, with low user counts.
Google does not publish the exact threshold and you cannot change it. At the time of writing (October 2026), GA4 shows a data quality icon in the report header, and hovering over it confirms whether thresholding was applied. Rows below the threshold simply drop out, so the totals in that report no longer match totals elsewhere in GA4.
Thresholding is different from sampling. Sampling estimates figures from a subset of data to answer heavy queries faster; thresholding removes figures to protect privacy. It is also different from the “(other)” row, which bundles values together when a report has too many unique rows, a problem caused by high cardinality.
Your reporting identity setting affects how often you see it. Device-based reporting uses only the browser’s cookie identifier and tends to reduce thresholding, while the blended and observed options can bring more of it into play. Google has changed how signals data feeds reporting more than once, so check the current help page before relying on any one setting.
Why it matters
Smaller UK businesses run into it most. A specialist firm with a few hundred visitors a week, looking at a single day or a single campaign, can find whole rows missing. If you report enquiries by age group to a board or a client from a thresholded report, the figures will be understated and will not reconcile with your ad platforms or your CRM.
It also explains many “missing data” scares. When someone compares an exploration with a standard report and gets different numbers, thresholding is one of the first things to rule out, along with sampling and consent.
Common mistakes
- Treating a thresholded total as the real figure and reporting it upwards.
- Switching on Google signals for demographic insights without realising it can trigger thresholding across many reports.
- Analysing a single day or a tiny segment, then concluding a campaign produced nothing.
- Blaming the cookie banner or a broken tag for gaps that are really thresholding.
How to act on it
First check whether the data quality icon is showing. If it is, widen the date range, remove demographic dimensions or report at a broader level, such as channel rather than campaign. In the property’s reporting identity settings, test whether device-based reporting removes the issue for the reports you rely on. The setting applies to all reports retrospectively and changes how users are counted, so note the date you change it.
For figures that must be exact, such as orders or leads, count them from the source system: your shop platform, booking system or CRM. If you need event-level detail without thresholds, GA4’s BigQuery export provides the raw events, although it leaves out the demographic data that triggers most thresholding anyway. Setting up reporting that holds together when GA4 trims its figures is part of the digital marketing strategy and measurement work I do.