Local SEO

How to configure and preview GA4 web-stream data redaction

A URL can carry an email address or a sensitive query value into Analytics without anyone intending to send it. Use an existing GA4 web stream's Redact data settings and a synthetic preview to check the specific patterns you plan to remove before future web events are collected.

By LocalCMO Editorial TeamReviewed by LocalCMO Editorial Team9-minute guideUpdated October 29, 2026

Expected outcome

What you will have when you finish

A record of the approved web stream, enabled redaction options, named query keys, synthetic preview inputs and outputs, unresolved exposure paths, and the owner responsible for follow-up. This is not a historical data-deletion request or a privacy certification.

Before you start

Gather these details before you begin. It keeps the task focused and prevents guesswork later.

  • An existing GA4 property and its intended web data stream; installation, tagging changes, and app-stream configuration are outside this guide
  • A property-level Editor or Administrator authorized to change this stream's data-redaction settings
  • A list of actual query parameter names used by your website, checked with the website or analytics owner; never paste real customer values into a test
  • A private note for the property and stream, approved key names, synthetic samples, observed preview results, remaining risks, and review owner
Access and approval

Google requires Editor access or above at the property level to configure data redaction. Agree the stream, email option, and exact query keys with the site's analytics and privacy owners before changing them. Do not change another stream or run a deletion request as part of this procedure.

Redaction is a narrow collection safeguard, not a privacy sign-off

A synthetic preview is useful because it separates two questions: did an email-shaped string match, and did the named URL query key match even when its value was not an email? A replacement in one sample does not answer whether the website is sending personal information somewhere else. The safest change is to avoid putting it into URLs, titles, and event data in the first place.

Google’s GA4 data-redaction instructions give the current Admin path, preview procedure, and collection boundaries. Its PII prevention guidance explains why URL paths, parameters, titles, and user-entered data need attention at the source. If data has already reached Analytics, the separate data-deletion request documentation describes a different process; this guide does not perform it.

Keep the observation and approval record with the responsible analytics team. LocalCMO’s marketing audit trail and human approval workflow are background on accountable marketing decisions, not tools for editing GA4 data streams or declaring this property compliant.

Set up and preview redaction on one GA4 web stream

Step 1

Identify the web stream and the values to avoid

Set the collection boundary before editing the stream.

  • Record the intended property, website domain, web stream, review date, and approving owner. Confirm this is a web stream rather than an app stream; Google's data-redaction control is currently web-only.
  • Ask the website owner which URL query keys might contain personal or otherwise sensitive values. For example, a booking page might use contact_email; record the key name, not a visitor's value. Also inspect whether sensitive information appears in the URL path, page title, form inputs, event parameters, or other destinations.
  • Choose harmless synthetic test data only. Use example.com and invented addresses under that domain. Do not copy a customer URL, email, token, or screenshot containing personal information into a shared test record.
  • Agree whether switching on either option or adding keys could affect intended measurement. If you cannot establish the correct stream or permitted keys, stop and obtain an owner decision before changing settings.

Step 2

Enable the agreed redaction options

Use the stream's Redact data settings, not an event-exclusion filter.

  • In Google Analytics, choose the correct property. Go to Admin > Data collection and modification > Data streams, select the web stream, then in Events open Redact data. A property-level Editor or above is required.
  • Inspect the current switches instead of assuming email redaction is enabled. Google says it is on by default for newly created properties, but older properties may require it to be enabled. Turn on the email-address option if approved.
  • Turn on URL query parameter redaction if approved, and enter each agreed query key by name, pressing Enter after each. For the example booking URL, enter contact_email, not an example address or an entire URL. Check that the intended keys actually appear in the configured list.
  • Record the resulting settings and check that they remain present when you revisit the stream. If you cannot see or retain the agreed changes, ask the property owner to resolve access or configuration; a preview alone does not establish a persistent setting.
Practical tip: Email recognition scans text patterns in event parameters on a best-effort basis. Named query-key redaction targets URL query values within Google's specified URL-bearing event parameters; the two options are not interchangeable.

Step 3

Preview synthetic email and query values

Compare the redacted preview to an explicit expected result without sending test traffic.

  • In Test data redaction, build a synthetic URL from https://example.com/ by appending ?contact_email=alex@example.com and then &campaign=autumn, with no spaces between the parts. Select Preview redacted data. With email and contact_email redaction enabled, expect the contact_email value to be replaced with (redacted); check that the harmless campaign value is still visible. Do not visit the sample URL.
  • Test the named key independently: keep the same URL but replace alex@example.com with sample-code. Because sample-code is not an email address, its replacement tests the configured query key rather than merely the email pattern.
  • Enter a separate invented email in ordinary sample text, such as Reach tester@example.com, to check the email option without relying on a URL key. Note the exact input, selected options, and displayed Redacted version for each preview.
  • If the expected value remains in a preview, recheck the selected web stream, each switch, and the entered key name. Do not substitute live customer data to force a match or describe an inconclusive preview as a privacy pass.

Step 4

Review what the preview cannot establish

Keep uncovered collection paths separate from a successful sample.

  • Google describes email detection as best effort: some addresses may escape and email-like harmless text may be removed. Query-key replacement applies only to the keys you specify and to URL query parameters in page_location, page_referrer, page_path, link_url, video_url, and form_destination. It is not a blanket scrub of every field or URL path.
  • The web-stream control does not evaluate HTTP header values, including the referer header, and does not prevent PII collection through Measurement Protocol or Data Import. A synthetic preview cannot verify those other collection paths.
  • If an actual website URL, title, user-entered value, or custom event may contain personal information, ask its implementer to stop sending that information at the source. Google requires page URL paths, parameters, and titles to be free of PII; redaction does not transfer that responsibility to Google.
  • Do not treat this test as deletion of previously collected data. Google provides a separate data-deletion request process for data on Analytics servers; escalation and authorization for any historical incident belong to the property and privacy owners.

Step 5

Hand off the result and follow-up

Make the observed scope and unresolved risks reviewable.

  • Save a private record of the property and stream, authorized editor, approval, switch states, query keys, synthetic inputs, Redacted version outputs, and observation date. Mark each case Expected replacement, Unexpected result, or Not tested.
  • List anything this preview did not cover: other query keys, URL paths, headers, app streams, Measurement Protocol, Data Import, and historical events. Assign an owner to investigate any sensitive-data path at its source.
  • Ask the analytics or privacy owner to review the record and decide whether a separate implementation check or incident response is needed. Do not claim all site data is safe just because the example values were replaced in the preview.

Read the synthetic preview conservatively

These observations concern only the current sample and selected web-stream settings.

Preview observationWhat it supportsNext action
contact_email value is replaced for both the invented email and sample-codeThe preview demonstrates a match for this named query key in the two samples.Record settings and outputs; still review other keys and collection paths.
Invented email text is replaced outside a URLThe email text pattern matched this sample; it does not establish perfect recognition.Keep email redaction enabled if approved and audit what the website actually sends.
sample-code remains in the contact_email query valueThe preview has not demonstrated that the named key is being redacted.Check the query switch, exact entered key, and selected web stream; keep the result unresolved.
A harmless campaign value is also replaced unexpectedlyThe configuration or pattern matched more text than anticipated in this sample.Inspect the configured keys and text; discuss measurement impact with the owner before accepting the configuration.

Common mistakes to avoid

Testing only an email inside the named query key

The email pattern could hide a missing query-key configuration. A non-email synthetic value in the same key separates the two checks.

Pasting real customer data into the preview

Use invented values and example.com. A test record or screenshot should not become another place where personal information is exposed.

Assuming the preview cleans every collection channel

Redaction is limited to a GA4 web-stream client-side step; it does not evaluate headers or protect Measurement Protocol, Data Import, or old data.

Confusing stream redaction with GA4 traffic filters

An internal-IP exclusion's Testing label evaluates whether traffic would be discarded. Redact data previews replacement of matching text before web collection; it is a different control and task.

How to know the task is complete

Success checks

  • The approved GA4 property and web stream, current switch states, and exact query keys are recorded.
  • The preview shows observed output for a synthetic email, an email-bearing URL, and a non-email value under an agreed query key.
  • Unexpected results remain unresolved rather than being labeled compliant; the record names an owner for each uncovered path.
  • No real personal information was used, no historical deletion was requested, and no app or other stream was changed.

If something does not work

  • Redact data is not available: verify that you selected a web stream and have property-level Editor access; ask an authorized owner rather than borrowing credentials.
  • The invented email remains visible: verify the email switch and sample text; report a best-effort detection limit rather than assuming all addresses are covered.
  • The synthetic non-email query value remains visible: verify the URL-query switch and entered key, then retry a harmless sample. Escalate if it still fails.
  • Historical data may already contain personal information: stop treating this preview as remediation and consult the privacy owner about the separate Analytics data-deletion process.

GA4 email and query parameter redaction FAQ

Does GA4 data redaction work for app streams?

No. Google's data-redaction setting is currently available only for web data streams. Do not infer that an app stream or server-side Measurement Protocol is protected by this preview.

Is an email address inside a URL enough to test the query-key option?

No. Email recognition might replace that value even if the key is missing. Use an invented non-email value under the same named key and inspect its Redacted version separately.

Will enabling this setting remove data GA4 collected before today?

No. Google evaluates events before collection for this web-stream feature. Historical removal is a distinct data-deletion request process requiring owner review; the synthetic preview neither starts nor validates one.

Does a matching preview prove our website complies with privacy rules?

No. Email recognition is best effort, query redaction covers named keys in specified URL-bearing event parameters, and other paths remain outside this test. Google assigns responsibility for regulatory compliance to the entity collecting the data.

Can LocalCMO change our GA4 stream or certify its data redaction?

This guide documents a task to perform in Google Analytics with an authorized property Editor. LocalCMO's linked audit-trail and approval pages describe marketing evidence and human decisions, not a GA4 stream connection, redaction service, or compliance certification.

Put the guide into practice

Use LocalCMO for the next step

Keep the source evidence, proposed action, human approval, and final result connected instead of working from separate screenshots and notes.