What is BigQuery Export?

The GA4 interface aggregates everything it shows you, applies sampling above certain thresholds in Explorations, hides low-count rows under data thresholding when Google Signals is on, and caps retention at 14 months maximum (2 months by default). BigQuery Export sidesteps all four limits at once. Once linked, GA4 writes a fresh, unsampled, event-level table into a Google Cloud project you own, kept for as long as you're willing to pay to store it.

Where to find it:

Admin > BigQuery Links (under Property Settings) > Link. Choose your Google Cloud project (billing must already be enabled on it), choose Daily export, Streaming export, or both, pick a data location, and submit. The export is not retroactive. It only starts collecting from the day you link it forward, so historical data that already fell out of GA4's retention window is gone from both places.

Why it matters:

A NiceLookingData audit checks every property for a live BigQuery link and flags it as a warning when there isn't one, because the fix has to happen before you need it: you cannot backfill history after the fact. Three pitfalls show up repeatedly:

1. It's opt-in and easy to forget. It isn't part of any default setup flow, so most properties simply never enable it. The gap is usually discovered only when someone asks a "how did we perform last spring" question that GA4's UI can no longer answer.
2. Daily and Streaming aren't the same toggle. Daily export lands as one finalized table per day, typically within a few hours after midnight, which is fine for most reporting. Same-day SQL access needs Streaming export specifically switched on as well. Teams that assume one implies the other find their intraday dashboards empty.
3. It isn't free, even though linking is. Google doesn't charge to create the link, but BigQuery storage and query costs are billed through your own Cloud project. Usually low for small-to-mid traffic sites, but not zero, and an unfiltered dashboard querying the full events table repeatedly can run costs up unexpectedly.

Example: A property runs on GA4's default UI for two years. The team finally links BigQuery Export in year three, expecting to backfill and analyze the prior two years in SQL, but the export only starts from the day it's linked. The two years of history already outside GA4's 14-month retention window are unrecoverable in either place.

BigQuery is also the standard fix for Looker Studio's "Quota Errors": the GA4 Data API enforces strict per-property hourly limits that the BigQuery connector doesn't share. Full setup and starter SQL queries: GA4 BigQuery Export: Setup, Schema, and 5 Essential SQL Queries and How to Set Up GA4 BigQuery Export (Step-by-Step). See also GA4 Data Retention: Why You're Losing Historical Data.