Refresh error with BigQuery credentials

Please try the solution that has been marked accepted in below link: 

https://community.fabric.microsoft.com/t5/Service/The-credentials-provided-for-the-GoogleBigQuery-source-are/m-p/4390694#M263869

It is similar to what you are experiencing.

Hi laracitrinelaw,

Thank you for reaching out to the Fabric Community. We are happy to assist you with your issue!

This is a very common issue you're hitting an OAuth token expiry problem. OAuth tokens used by Power BI for BigQuery don't stay valid indefinitely and require periodic re-authentication. Here's how to fix it permanently:

Root Cause

OAuth tokens expire quickly and often fail on scheduled refresh which is exactly why your first few refreshes work, then it breaks again daily.

Immediate Fix

• Go to Power BI Service → Semantic Model → Settings → Data source credentials
• Delete the existing BigQuery credentials
• Re-authenticate fresh using OAuth2
• Try the refresh again it should work right away

Permanent Fix - Switch to a Service Account

If you're using OAuth, consider switching to a service account with a private key service accounts are much more reliable for automation and scheduled refresh.

Steps:

1. In Google Cloud Console → IAM → create/use a Service Account
2. Assign roles: BigQuery Data Viewer + BigQuery Job User
3. Generate a JSON key file
4. In Power BI Service → Data source credentials → select Service Account → upload the JSON key
5. This avoids token expiry entirely

Extra Checks

• Make sure to re-enter credentials in both Power BI Desktop AND Power BI Service remove the old ones first, then add updated credentials in both places to keep everything in sync.
• Verify IAM permissions include BigQuery Data Viewer and BigQuery Job User roles missing permissions can cause intermittent failures.
• Use a dedicated service account to avoid expiry issues in the future.

Hope this resolves it! Let us know which fix worked for you.

Please consider marking this as the Accepted Solution to help other community members find this fix more easily. If this helped you, Kudos are always appreciated!