Sampling

Sampling is when GA4 calculates a report from a subset of your events rather than every one, then scales the result up to estimate the whole. It's a speed trade-off: complex or wide-ranging queries return fast because GA4 didn't count everything. When a report is sampled, GA4 shows an icon in the header telling you the percentage of data used.

Why it matters

Most small businesses rarely hit sampling in GA4's standard reports — it tends to appear in Explorations over long date ranges or with many dimensions. But when it does happen, the numbers become estimates, not exact counts. That's fine for spotting trends and bad for any decision that hinges on a precise figure, like reconciling revenue or counting conversions.

A concrete example

Say you build an Exploration covering a full year, segmented by city and device, and the header shows "based on 42% of sessions." Your reported figure of 1,000 conversions is GA4's estimate from roughly four months of data scaled to twelve. The true number could be a few percent higher or lower. The shape of the trend is reliable; the exact total is not.

The common misreading

The mistake is quoting a sampled number as if it were exact, then panicking when it doesn't match another tool — a classic data discrepancy. The fixes are simple: shorten the date range, simplify the report, or for precise work send your data to BigQuery, which never samples. Always check the header icon before trusting a figure.