Documentation · Updated July 2026
Connect Search Console
Metricsjar reads Google Search Console data from the bulk export in BigQuery. The export is more useful than repeatedly calling the Search Console API, but Google only starts writing data after you enable it. It does not backfill earlier days.
Before you start
You need:
- Owner access to the Search Console property.
- A Google Cloud project with billing enabled.
- Permission to create a BigQuery dataset and grant access in that project.
The export itself does not require you to give Metricsjar a Google OAuth token.
Your local agent runs the BigQuery queries with your existing gcloud
authentication.
Let your agent guide the setup
The Google console step needs a browser, so your agent may stop and give you a direct instruction for the Search Console settings page. After the export is enabled, the agent can verify the dataset and register it locally.
Paste this into your coding agent:
Set up the Google Search Console bulk data export to BigQuery for my site, then
register the dataset with Metricsjar. Enable the export before doing any
optional sources because Search Console does not backfill.
The agent should:
- Confirm the exact Search Console property.
- Choose or create the Google Cloud project and BigQuery dataset.
- Verify that the Search Console export service account can write to the project.
- Register the project, dataset, property, and Google account in the local Metricsjar site configuration.
- Run the doctor with a clean scheduled-job environment.
When data appears
The first export is not immediate. A correct setup can remain empty while Google prepares the first tables. Metricsjar should report this as waiting for data, not as a zero-traffic site.
Once rows arrive, the Acquisition tab can show queries, pages, clicks, impressions, movers, and tracked keyword positions. DataForSEO is optional and adds independent rank verification and backlink data.
Common problems
The dataset exists but the doctor cannot query it
Listing a BigQuery dataset does not prove the active account can run jobs. Metricsjar checks with a dry-run query. Confirm that the selected Google account has permission to create BigQuery jobs in the project.
It works now but the scheduled refresh fails
Credentials exported in one terminal session disappear when the scheduled job runs. Store the selected account in the Metricsjar configuration and let the doctor test with the temporary shell variables removed.
There is no earlier history
That is expected. Search Console starts the bulk export after setup and does not backfill. Keep the export enabled even if you are not ready to configure the rest of Metricsjar.