BigQuery GA4 Export — From Linking to First Query

Admin → BigQuery Links. Pick daily + streaming. Streaming costs basically nothing at most B2B scales and saves you a debugging headache later.

Each row is one event. event_params is a repeated record — you'll UNNEST it constantly. user_properties is structured the same way.

SELECT DATE(TIMESTAMP_MICROS(event_timestamp)) AS d,
       COUNT(DISTINCT user_pseudo_id) AS users
FROM `project.dataset.events_*`
WHERE event_name = 'session_start'
  AND _TABLE_SUFFIX BETWEEN '20250101' AND '20250131'
GROUP BY d ORDER BY d;

SELECT
  event_date,
  (SELECT value.string_value FROM UNNEST(event_params)
   WHERE key = 'page_location') AS page_location,
  COUNT(*) AS views
FROM `project.dataset.events_*`
WHERE event_name = 'page_view'
GROUP BY 1, 2;

WITH steps AS (
  SELECT user_pseudo_id,
    MAX(IF(event_name = 'sign_up', 1, 0)) AS s1,
    MAX(IF(event_name = 'wallet_funded', 1, 0)) AS s2,
    MAX(IF(event_name = 'p2p_send_completed', 1, 0)) AS s3
  FROM `project.dataset.events_*`
  WHERE _TABLE_SUFFIX BETWEEN '20250101' AND '20250131'
  GROUP BY user_pseudo_id
)
SELECT SUM(s1) AS signed_up,
       SUM(s2) AS funded,
       SUM(s3) AS sent
FROM steps;

Always filter by _TABLE_SUFFIX — wildcard scans of events_* cost real money. Materialize daily aggregates into a separate table for dashboards.