Google Analytics monitoring with Topologix

Analytics & Apps

GA4 web analytics on the canvas through a service account — 16 presets: 12 straight off the GA4 Data API (realtime active users, users/sessions/pageviews/events over 1-to-28-day windows, engagement and bounce rates) plus 4 near-real-time BigQuery presets that query the GA4 streaming export over 1-minute to 30-minute windows.

Requirements

• A GA4 property and a GCP service account JSON key, with the service-account email added as Viewer under GA4 → Admin → Property Access Management.
• No OAuth dance: the server signs a JWT with the key and exchanges it for a token scoped analytics.readonly + bigquery.readonly, refreshed automatically every hour.
• For the “(BigQuery)” presets only: GA4 → Admin → BigQuery Links with Streaming enabled, and the service account granted BigQuery Data Viewer on the GA4 dataset plus BigQuery Job User on the project.

1 · Set up the integration

1. Open the Integrations pageClick Integrations in the toolbar and pick Google Analytics from the catalogue.
2. Create the service accountGCP Console → IAM → Service Accounts → create an account and add a JSON key, then grant that email Viewer on the GA4 property.
3. Paste the JSONDrop the whole key file into Service Account JSON — the drawer echoes the detected client_email. The optional BigQuery Project ID and BigQuery Dataset fields default to the JSON’s project_id and analytics_<propertyId>.
4. Create, then verifyClick Create, then reopen the integration and click Test — it lists account summaries and reports how many GA4 properties are accessible; with a BigQuery project set it also runs a trivial SELECT 1 to validate the Job User role and appends “BigQuery OK”.

2 · Add it to a flow

1. Select the nodeClick the node, open Metrics, then + Add and choose the Google Analytics integration.
2. Pick property and metricThe GA4 Property dropdown is discovered from the service account (it’s the numeric Property ID — not the “G-” Measurement ID), and the Metric list shows every preset tagged with its path: realtime, report, or bigquery (~1m lag).
3. SaveOptionally set a label, pick the interval, and click Add Source.

Good to know

• Report presets use GA4’s own relative date ranges (yesterday, 7daysAgo, 28daysAgo), and the rate presets (engagement, bounce) are converted to percentages for the card.
• The BigQuery presets read events_intraday_* — active users via APPROX_COUNT_DISTINCT(user_pseudo_id), events via COUNT(*) — so they see activity within about a minute, far ahead of the Data API’s processing lag.
• Misconfiguration comes back as the exact fix, not a stack trace: a missing streaming export reads “turn on Streaming in GA4 → Admin → BigQuery Links”, a 403 names the two BigQuery roles to grant.