Understanding the GA4 BigQuery Export Schema

The GA4 Export Table Structure

When GA4 data is exported to BigQuery, it lands in date-sharded tables named events_YYYYMMDD (one table per day). Each row represents one event fired by one user. The table has approximately 30 columns, but the most important for analytics are:

• event_date: date of the event (string, YYYYMMDD format)
• event_timestamp: Unix microseconds when the event fired
• event_name: the name of the event (page_view, purchase, custom_event_name)
• event_params: REPEATED RECORD — array of key-value pairs with all event parameters
• user_pseudo_id: GA4's anonymous user identifier (equivalent to client_id)
• user_id: the user_id you set explicitly (if implemented)
• user_properties: REPEATED RECORD — user-scoped properties
• items: REPEATED RECORD — ecommerce item data (for purchase/add_to_cart events)
• geo.country, geo.city: geographic data
• device.category, device.operating_system: device data
• traffic_source.source, traffic_source.medium: acquisition data

Accessing Nested event_params

The most commonly needed values (page URL, revenue, session ID) are in the event_params RECORD array. To access them, you need to UNNEST or use a subquery:

-- Get page URL from event_params:
SELECT
  event_name,
  (SELECT value.string_value FROM UNNEST(event_params)
   WHERE key = 'page_location') AS page_url
FROM `project.analytics_PROPERTY.events_*`
WHERE _TABLE_SUFFIX = '20260115'
  AND event_name = 'page_view'

Common event_params Keys

• page_location: full URL of the page (string_value)
• page_title: page title (string_value)
• session_id: identifies a session (int_value)
• engagement_time_msec: time engaged on page (int_value)
• value: revenue for purchase events (float_value)
• currency: currency code (string_value)
• transaction_id: order ID for purchase events (string_value)

Accessing Items Array

For ecommerce analysis, the items RECORD contains one row per product per event:

SELECT
  event_name,
  item.item_id,
  item.item_name,
  item.quantity,
  item.price
FROM `project.analytics_PROPERTY.events_*`,
UNNEST(items) AS item
WHERE _TABLE_SUFFIX = '20260115'
  AND event_name = 'purchase'

Summary

The GA4 BigQuery schema uses nested RECORD types for event parameters, user properties, and ecommerce items. Each event is one row; nested values (page URL, revenue, product data) are accessed via UNNEST or subqueries on the REPEATED RECORDs. The key columns for most analyses: event_name, user_pseudo_id, event_params (nested), items (nested for ecommerce), traffic_source, and device. Master the UNNEST pattern for event_params — it is used in nearly every GA4 BigQuery query.

See our BigQuery Setup service for schema guidance and query development.

Need help querying GA4 data in BigQuery? Contact Adslytics.