GA4 Thresholds, Sampling and Data Quality for Marketers
Why GA4 numbers look off: data thresholds, sampling in explorations, consent gaps and (other) rows, and what you can do about each.
Most "wrong" GA4 numbers come from four causes: data thresholds hiding small rows, sampling in large explorations, consent and blockers removing visits, and (other) rows from high-cardinality reports. Learn to recognise each, then use the fix that fits.
- Thresholds hide rows that could identify users, often when Google signals or demographics are involved.
- Sampling starts when a query exceeds 10 million events on standard properties.
- The data quality icon at the top of a report tells you which issue applies.
- BigQuery export has raw, unsampled, unthresholded events, but no Google signals data.
- Consent and blockers mean GA4 undercounts; treat it as a trend tool, not a ledger.
Checked 4 November 2026 (CET) against Google Analytics Help and Google developer documentation. GA4 menus and limits change, so confirm in your own property. "Our view" marks MentionX opinion. Ledgerline is a fictional company used for illustration.
01Which problem do you have?
| Symptom | Likely cause | Fix |
|---|---|---|
| Rows missing, totals don't add up, warning icon mentions thresholding | Data thresholds | Remove demographic or interest dimensions, or switch to device-based reporting identity |
| Shield icon turns yellow in an exploration | Sampling | Shorten the date range, or use BigQuery (lesson 17) |
| A row called (other) | High-cardinality limits in standard reports | Use an exploration or BigQuery |
| GA4 lower than server logs or CRM | Consent, blockers, referrer loss | Expected; compare trends, not totals |
02Reporting identity and thresholds
Thresholds often appear when Google signals is on. Switching Admin → Data display → Reporting identity to Device-based can reduce thresholding in reports. It is reversible and changes reporting only, not collection. Weigh it against losing cross-device joins in reports.
03What the numbers can and can't support
GA4 is good for direction: which channels and pages are growing, and which funnels leak. It is not good for exact counts that must reconcile with finance. For AI traffic, the gap is bigger than average, because apps strip referrers; see why AI referral traffic is undercounted and server-side tracking of AI agents.
Our view: if a decision depends on a gap smaller than about the size of your consent opt-out rate, GA4 can't settle it. Run an experiment instead.
Next: Linking Search Console and Google Ads. Or return to the GA4 for Marketers track.
04Common questions
Can I turn off data thresholds in GA4? No. Google says thresholds are system defined and can't be adjusted, but you can reduce how often they appear, for example by changing reporting identity or date ranges.
When does GA4 sample data? When a query exceeds the quota: 10 million events for event-level queries on standard properties, up to 1 billion on 360. This mostly affects explorations.
Is BigQuery data thresholded? The export holds raw events, so report thresholds don't apply, and Google says Google signals data is not exported. Counts can differ from the GA4 interface.
Questions people ask
3 answersCan I turn off data thresholds in GA4?
+
No. Google says thresholds are system defined and can't be adjusted, but you can reduce how often they appear, for example by changing reporting identity or date ranges.
When does GA4 sample data?
+
When a query exceeds the quota: 10 million events for event-level queries on standard properties, up to 1 billion on 360. This mostly affects explorations.
Is BigQuery data thresholded?
+
The export holds raw events, so report thresholds don't apply, and Google says Google signals data is not exported. Counts can differ from the GA4 interface.
Sources
- Google Analytics Help, About data thresholdssupport.google.com
- Google Analytics Help, About data samplingsupport.google.com
- Google Analytics Help, BigQuery Exportsupport.google.com
- All accessed 4 Nov 2026.