BigQuery + GA4: The Setup Guide Nobody Writes Simply

The GA4 interface is fine for high-level reporting. But the moment you need to answer a question like “what’s the average number of sessions before a user converts, broken down by traffic source and device type”, you’ve hit a wall. BigQuery removes that wall.

Why Connect GA4 to BigQuery?

The GA4 UI applies sampling to many reports. BigQuery gives you raw, unsampled event-level data. Every event, every user, every session — available for SQL queries.

You also get:

• Data retention beyond 14 months (GA4 limits raw data retention; BigQuery keeps it as long as you want)
• Joins with your own data — combine GA4 events with your CRM, ad spend, or revenue data
• Custom attribution modelling — build your own models rather than relying on GA4’s
• Bulk exports for Looker Studio — much faster dashboards when you’re querying BigQuery rather than GA4 directly

Prerequisites

• A Google Cloud project with billing enabled (you’ll stay within the free tier for most small-to-medium sites)
• Editor or Owner access to your GA4 property
• The BigQuery API enabled in your Cloud project

The Setup (Shorter Than You Think)

1. In GA4, go to Admin → Product Links → BigQuery Links → Link
2. Select your Google Cloud project
3. Choose your data location (pick the region closest to you)
4. Select Daily and/or Streaming export
5. Click Save

That’s it. GA4 will start exporting to BigQuery within 24 hours. You’ll find your data in a dataset named after your GA4 property ID.

Understanding the Schema

Each day’s data lives in a table named events_YYYYMMDD. The core fields you’ll use constantly:

SELECT
  event_name,
  event_date,
  user_pseudo_id,
  (SELECT value.string_value FROM UNNEST(event_params) WHERE key = 'page_location') AS page_url,
  (SELECT value.string_value FROM UNNEST(event_params) WHERE key = 'session_id') AS session_id
FROM `your-project.analytics_XXXXXXXX.events_*`
WHERE _TABLE_SUFFIX BETWEEN '20250101' AND '20251231'
  AND event_name = 'page_view'

The UNNEST(event_params) pattern trips people up at first but becomes second nature quickly.

Your First Useful Query

Session count by channel, last 30 days:

WITH sessions AS (
  SELECT
    user_pseudo_id,
    (SELECT value.int_value FROM UNNEST(event_params) WHERE key = 'ga_session_id') AS session_id,
    (SELECT value.string_value FROM UNNEST(event_params) WHERE key = 'medium') AS medium,
    (SELECT value.string_value FROM UNNEST(event_params) WHERE key = 'source') AS source
  FROM `your-project.analytics_XXXXXXXX.events_*`
  WHERE _TABLE_SUFFIX >= FORMAT_DATE('%Y%m%d', DATE_SUB(CURRENT_DATE(), INTERVAL 30 DAY))
    AND event_name = 'session_start'
)
SELECT
  CONCAT(source, ' / ', medium) AS channel,
  COUNT(*) AS sessions
FROM sessions
GROUP BY channel
ORDER BY sessions DESC

Costs

For most sites under a few million monthly sessions, you’ll comfortably stay within BigQuery’s free tier: 10 GB of storage free, 1 TB of query processing per month free. The GA4 export itself is free.

If you want help setting this up or writing your first set of queries, book an hour with me — it’s a good use of the session.