Study first
Review the ideas behind the questions
Review the GA4 privacy controls that shape whether data should be collected, modeled, or removed. Focus on PII risk, data redaction limits, consent signals, and how privacy settings affect reporting decisions.
Keep personal data out of collection
GA4 privacy checks start before a report opens. The safest workflow is to prevent personal or sensitive data from being sent in URLs, titles, event parameters, campaign fields, or location details.
- Do not send data that Google could use or recognize as personally identifiable information into Analytics.
- Page URLs, URL parameters, and page titles should be checked because the basic page tag can collect them.
- Fine-grained location data, such as GPS or latitude and longitude, should not be passed to Analytics.
In Practice
Prevention beats cleanup
A privacy issue should be fixed before collection wherever possible. Redaction and deletion workflows help, but they do not make a risky implementation safe by themselves.
Campaign fields can carry risk
UTM and campaign values are reporting fields, but they still need privacy review. Names, emails, or other identifiers should not be placed in campaign parameters.
Common mistakes
Treating hashed, hidden, or report-only identifiers as safe to send without review.
Review whether the value could identify a person before it is collected, especially in event parameters, user properties, campaign values, URLs, and titles.
Q&A
Where should a team look first when email addresses appear in page URLs?
Review the URL path, query parameters, and page title before those values are sent to Analytics.
Can a marketer put a customer email in a UTM campaign value if it helps segmentation?
No. Campaign parameters sent to Analytics should not include personally identifiable information.
Use redaction and consent settings deliberately
GA4 has controls that help reduce privacy risk and verify consent signal health. Those controls need to be configured against the actual collection path instead of treated as a universal safety net.
- Data redaction can remove likely email addresses and selected URL query parameters for web data streams.
- Data redaction should be tested because email redaction works on a best-effort basis.
- Consent settings can show whether GA4 is receiving consent signals for ads measurement and ads personalization.
In Practice
Redaction is stream-specific
A web-stream redaction setting does not cover every way data can enter GA4. Server-side and upload paths still need their own privacy checks before data is sent.
Consent signals have different jobs
Consent checks should separate behavior analytics signals from advertising-related signals. A linked ads use case may need ads measurement and ads personalization consent signals, not only a generic banner check.
Common mistakes
Turning on web data redaction and ignoring Measurement Protocol or Data Import.
Review the server-side or upload path separately because GA4 data redaction does not prevent PII through Measurement Protocol or Data Import.
Q&A
What should a redaction test prove before launch?
It should show whether the configured email and query-parameter patterns are redacted as expected for the sample values.
Where should a team check whether GA4 is receiving ads consent signals?
Open Consent settings for the relevant data stream and review the advertising-related consent signal details.
Read privacy-affected reports with the right limits
Consent and deletion controls can change what GA4 reports can show. Analysts should know when data is modeled, where modeling is unsupported, and how deletion requests affect historical values.
- Behavioral modeling depends on consent mode setup, tag loading behavior, and enough consented and unconsented traffic.
- Modeled behavioral data is not supported in every GA4 surface, including audiences and BigQuery export.
- Data-deletion requests remove selected text values for a specified date range, but events can still count in overall metrics.
In Practice
Modeling is conditional
A cookie banner alone does not prove modeled data will appear. Check the Consent Mode implementation, traffic thresholds, reporting identity, and data-quality messages before explaining report changes.
Deletion windows need planning
A deletion request should be scoped to the data that needs removal. If Consent Mode is involved, the end date may need extra days so modeled data is covered.
Common mistakes
Expecting BigQuery export to include modeled behavioral data from unconsented users.
Use GA4 reporting surfaces that support modeled behavioral data; BigQuery export is listed as unsupported for behavioral modeling.
Q&A
Why might modeled data be unavailable even after Consent Mode is installed?
The property may not meet the setup, traffic, reporting identity, or model-quality conditions required for behavioral modeling.
What happens to an event when a data-deletion request removes selected parameter text?
The selected text is replaced with a deleted-data value, while the event can still count in overall report metrics.