GA4 BigQuery PII Auditor

GA4 BigQuery PII Auditor

A metadata-only scan of a GA4 BigQuery export for fields that might carry personal data — it never writes to BigQuery.

Why this matters

If you're not the one who built the tracking setup, it's genuinely hard to know what's actually being collected. A parameter gets added by a teammate, an agency, or a tool update months ago — and it just stays there, invisible unless someone goes looking. For a non-technical person especially, there's often no practical way to tell whether something like an email address, a phone number, or another personal identifier ended up sent to GA4 and stored in BigQuery, without anyone realizing it.

This tool exists so that check doesn't require being a BigQuery expert: sign in, pick a dataset, and see whether any field name looks like it might be carrying personal data. It's just as useful run the other way, too — after a cleanup effort, re-running the same audit is the simplest way to confirm the PII actually stopped showing up, rather than just hoping it did.

What this tool does

This tool connects to a GA4 BigQuery export dataset (with the owner's permission) and runs two checks: a scan of event-parameter, user-property, and item-parameter field names against a list of terms that commonly indicate personal data (email, phone, address, and similar), and a draft inventory of every event/parameter combination actually observed in the data. It never looks at field values, and its own code never creates, edits, or deletes anything in BigQuery.

It's built for anyone maintaining a GA4 property who wants a quick, repeatable way to catch accidental PII collection in analytics events, rather than hand-reviewing every event parameter.

What access it asks for

The tool requests two Google permissions: the BigQuery scope (bigquery — to list datasets and tables, and run SELECT/dry-run queries; Google's API requires this full scope even for a read-only dry run, a narrower bigquery.readonly scope isn't accepted for that) and a narrow Resource Manager scope solely to list the GCP projects the signed-in user can see, so they can pick one from a list instead of typing a project id by hand. The tool's own code never writes to, or deletes anything in, Google Cloud — and on top of that, whoever signs in should only be granted "BigQuery Data Viewer" + "BigQuery Job User" (not Editor/Owner) on the projects this tool is used against, which blocks writes at the Google Cloud permissions level too.

What happens to the data

Each audit is run on demand against the project/dataset you select: the tool queries BigQuery, summarizes the results in memory, and renders the report. See the privacy policy for exactly what is and isn't kept between visits.

Who built this

This tool was built for personal/freelance analytics review work, and is shared here for evaluation by other teams and by Google as part of the app verification process.

Read the privacy policy.