What are _ga4 cookies? _ga versus _gid and its bigquery export mapping; ga4 cookies explained

GA4 uses first-party cookies set by Google Analytics, and the raw event data is exported into Google BigQuery.

Below is the practical mapping you need for analysis.

1. _ga Cookie → user_pseudo_id

Example: _ga=GA1.1.123456789.1699999999

The important part is the Client ID:

123456789.1699999999

In BigQuery

It maps to: user_pseudo_id

This is GA4’s device-level identifier

Equivalent to the Client ID

Used to identify unique users (per browser/device)

Example Query

SELECT
  user_pseudo_id,
  COUNT(*) as events
FROM `project.dataset.events_*`
GROUP BY user_pseudo_id

That’s essentially grouping by _ga client ID.

2. _ga_<measurement-id> Cookie → Session Info

GA4 also sets cookies like:

_ga_ABC123XYZ

This stores session-related info including:

• session_id
• session_number

In BigQuery, these map to:

Extracting Session ID

SELECT
  user_pseudo_id,
  (SELECT value.int_value
   FROM UNNEST(event_params)
   WHERE key = 'ga_session_id') AS session_id
FROM `project.dataset.events_*`

This equals the session stored in the _ga_<measurement-id> cookie.

3. _gid Cookie → NOT Directly Exported in GA4

_gid is used for short-term user differentiation (24 hours).

Important:

It does NOT have a dedicated field in BigQuery
It is not exported directly

Instead:

• Its logic is absorbed into sessionization
• GA4 event processing handles session grouping internally

Full Identity Mapping in GA4 + BigQuery

Important Real-World Notes

1. Consent Mode Impact

If consent is denied:

• _ga may not persist
• user_pseudo_id may reset
• BigQuery will show more fragmented users

2. Cross-Device Reality

user_pseudo_id ≠ real user
It’s device/browser specific.

If you implement:

user_id

Then you can stitch users across devices.

3. Rebuilding Sessions in BigQuery

To truly match GA4 UI sessions:

You need:

CONCAT(user_pseudo_id, ga_session_id)

That equals one session.

Exact Matching Formula

Unique Users (GA4 UI equivalent)

COUNT(DISTINCT user_pseudo_id)

Sessions (GA4 UI equivalent)

COUNT(DISTINCT CONCAT(
  user_pseudo_id,
  (SELECT value.int_value
   FROM UNNEST(event_params)
   WHERE key = 'ga_session_id')
))

Architecture Flow

Browser
→ _ga cookie (client id)
→ GA4 collects event
→ user_pseudo_id generated
→ Event exported
→ BigQuery stores raw event rows