GA4 Thresholding & Sampling Explainer
Explain missing rows, withheld data, and "why numbers don't match"
GA4 can apply thresholding to protect user privacy and sampling when report/exploration volume exceeds limits. This recipe produces a plain-English explanation, detection checklist, and mitigation playbook (including how to change the question—not just the dashboard).
INGREDIENTS
PROMPT
Diagnose whether we're seeing GA4 thresholding or sampling, then propose mitigations. Output: - Diagnosis with reasoning (thresholding vs sampling vs something else) - Mitigation plan (what to change in queries/reports) - One-paragraph explanation for non-technical stakeholders - A reusable "data limitations" note for recurring reports Inputs: - Report/exploration URL or description: - Dimensions/segments used: - Date range: - What looks wrong (missing rows, low numbers, warnings):
How It Works
This recipe helps teams stop "debugging ghosts" when data is withheld or sampled by design.
Triggers
- Demographics/interest reports hide rows or show warning indicators
- Looker Studio/GA4 numbers don't match in explorations
- Analysts suspect sampling or thresholding but can't explain it to stakeholders
Inputs
- Which reports/explorations show limited data
- Whether demographics/signals are involved
- Typical query depth (dimensions, filters, segments)
Outputs
- Root cause: thresholding vs sampling vs other
- Stakeholder-safe explanation
- Mitigation checklist (what you can change)
Actions / Steps
- Identify whether the question uses demographic/signals dimensions (thresholding risk).
- Identify whether the query exceeds event/report quotas (sampling risk).
- Mitigate by changing the question: aggregate more, remove sensitive dims, adjust date ranges, move analysis downstream.
- Create a standard "data limitations" note for recurring reports.
Parameters
- "Allowed granularity" policy per audience/report
- Date range defaults (exclude "today" where appropriate)
- Escalation path (analytics lead review)