Loading tutorials…
Loading tutorials…
Your dashboard worked yesterday. Today three charts show 'Configuration incomplete' and one shows numbers that look wrong. Here's the diagnostic sequence specialists run before changing anything.
Who this is forAnyone whose Looker Studio dashboard suddenly stopped working, started showing wrong numbers, or has charts breaking randomly. The fix typically takes 30-90 minutes if you know the diagnostic order.
What you'll need
Step 1
Looker Studio errors fall into 4 buckets: configuration (chart-level), data source (connection-level), permissions, and quota. Each has a different fix order.
Look at the error message on the failing chart. Hover the red error indicator for the full message.
Configuration errors: "Configuration incomplete," "This combination of fields isn't supported." Chart-level — fix the chart's field selection.
Data source errors: "Failed to fetch data," "Data source connection error." Source-level — fix the underlying source connection.
Permission errors: "You don't have permission to view this data." Either you lost access to the underlying source, or the data source uses Viewer's credentials and the current viewer lacks access.
Quota errors: "Quota exceeded," "Rate limit reached." Source-side quota was hit (GA4 API limit, Google Ads limit). Wait or reduce refresh frequency.
Step 2
Configuration errors mean the chart's field selection is incompatible with the data source. Open the chart and inspect Dimension/Metric assignments.
Click the broken chart. Right panel → Data tab. Look at Dimensions, Metrics, Filters.
Most common cause: a field was renamed or removed at the data source level. The chart still references the old field name. Re-pick the field from the current schema.
Second most common: a metric is set to an aggregation the field doesn't support (e.g., AVG on a text field). Change the aggregation.
Third: a calculated field returns null for all rows in the filtered range. Remove filters one at a time to isolate.
If the chart used a calculated field that no longer exists, you'll need to either recreate the field or rebuild the chart with current fields.
Step 3
If every chart from one source fails, the source connection is broken. Open Resource → Manage added data sources → click the source → check status.
Resource → Manage added data sources. Look for any source with a red error indicator.
Click the source → Edit Connection. You'll see the connector configuration.
For Google connectors (GA4, Google Ads, Sheets), the most common cause is revoked OAuth access. Reauthorize by clicking Reconnect.
For partner connectors (paid), check the partner's credentials — API keys often expire or get rotated. Update in the connector config.
After reconnecting, click Refresh fields (top of the data source). Verify the field schema looks correct. Save.
Step 4
Permission errors fall into two camps: report sharing (who can view the report) and data source credentials (who can read the underlying data).
If the error appears for one viewer but not others: the issue is on the viewer's side. Either they lost access to the report, or the data source uses Viewer's credentials and they lost access to the underlying source.
If the error appears for everyone including the owner: the data source's OAuth has expired or been revoked. Open the source → Edit Connection → Reconnect.
For Owner's credentials sources: the owner's Google account must continue to have access to the underlying source. If the owner left the company or had access revoked, every viewer sees errors. Re-assign ownership to a person with active access.
Document data source ownership in your team docs: 'GA4 source owned by [name]@everestx.com. If they leave, transfer to [backup name].'
Step 5
Quotas are usage limits per data source per time window. Hit them with too-frequent refreshes, too many concurrent users, or oversized queries.
GA4: free API has 10K requests/property/day. Heavy dashboards with multiple users can hit this on busy days.
Google Ads: 15K operations/day on standard accounts.
Sheets: 100 read requests / 100 seconds / user (different model).
BigQuery: 1 TB of free query data/month, then $5/TB.
Fixes: reduce data freshness (12h → 24h cuts refresh load in half), reduce concurrent viewers (some dashboards refresh per-view), or upgrade source quota.
For repeated quota hits, switch to BigQuery export of the source data. Removes per-source quota limits at the cost of BigQuery query fees.
Step 6
File → Version history → See version history. Looker Studio retains 30 days of versions. Restore from a known-good version when the current state is unfixable.
File → Version history → See version history.
A panel opens with the version timeline. Look for the last version before the issue started — usually the date of the last successful PDF export or known-good screenshot.
Preview the version (click it). If it looks correct, click Restore this version.
WARNING: Restoring a version overwrites the current state. Any changes since the restored version are lost. If unsure, make a copy of the current state before restoring (File → Make a copy → "Pre-restore backup").
After restoring, the issue may reappear if the upstream cause (e.g., revoked permissions) is still present. Restore is a recovery tool, not a fix.
Step 7
After any fix, re-validate against the source-of-truth UI. Don't trust 'it works again' — verify the numbers match.
Open the underlying source (GA4, Google Ads, Sheets) in a separate tab. Set the same date range as your Looker Studio dashboard.
Compare 3 metrics: a high-value one (Revenue or Conversions), a high-volume one (Sessions or Clicks), and one breakdown row (e.g., google / cpc Sessions).
All three should match within 1-2%. Anything beyond suggests the fix introduced a new issue.
Document the fix in a Change Log on the dashboard: date, what broke, what fixed it. Builds institutional knowledge so the same issue is faster to fix next time.
Common mistakes
Changing multiple things at once
What goes wrong: You reconnect the data source AND rebuild the calculated field AND change the filter. The chart works again — but you have no idea which change fixed it. Next time the issue recurs, you re-do all three changes from scratch.
How to avoid: Change one thing. Wait 2 minutes. Test. If broken, revert and try the next change. Painful but the only way to actually learn the failure mode.
Skipping the version history check
What goes wrong: You spend 2 hours debugging a chart that broke after an unrelated edit. Then realize you could have reverted to yesterday's working version in 30 seconds.
How to avoid: When an issue is mysterious and you have no obvious cause, check Version history first. If a recent version works, restore and avoid the debug rabbit hole.
Reauthorizing the data source without checking field changes
What goes wrong: You reconnect a Google Ads source and it works. But behind the scenes, the field schema changed — a metric you used was renamed. Charts continue to fail until you Refresh fields and re-pick the renamed metric.
How to avoid: After every reauthorization, click Refresh fields at the source level. Check if any field names changed. Update affected charts.
Not reconciling numbers after a fix
What goes wrong: The chart renders again, you assume it's fixed, but the underlying join logic is now wrong. The dashboard reports inflated revenue for 3 weeks before someone catches it. Decisions made on the wrong number.
How to avoid: Always reconcile 3 metrics against the source UI after any fix. Document the reconciliation. Takes 10 minutes; prevents 3 weeks of bad data.
Ignoring the cause and patching the symptom
What goes wrong: A chart breaks weekly because a teammate keeps editing the underlying Sheet structure. You patch the chart each week. The root cause (no Sheet-structure protection) keeps the issue recurring.
How to avoid: After every fix, ask: "What upstream change would prevent this from recurring?" Protect the Sheet, document the structure, set up alerts. Patch the cause, not the symptom.
Recap
Done — what's next
How to connect Looker Studio to GA4 without the rookie mistakes
Read the next tutorial
Hand it off
Troubleshooting a dashboard is a project. Keeping dashboards healthy week over week is a job. A specialist who maintains the data sources, monitors quotas, and audits the numbers monthly typically runs $300-500/mo at $14-16/hr. Most fixes happen before you ever notice them.
See ongoing maintenance rates
The chart's selected fields are missing, renamed, or incompatible with the current data source schema. Click the chart → Data tab → re-pick valid fields. If you recently refreshed the data source, fields may have been renamed upstream.
30 days of named versions. Looker Studio also auto-saves every change — recent edits (within hours) can be reverted via Undo (Ctrl+Z). For longer-term backup, File → Make a copy and store dated copies.
Almost always a permission issue specific to that viewer. Either: (1) they were removed from the report's share list, (2) the data source uses Viewer's credentials and they don't have access to the underlying source, or (3) their Google account's session expired and they need to refresh.
The error message will say 'Quota exceeded' or 'Rate limit reached.' For Google connectors, this often resolves itself in 1-24 hours. For repeated hits, reduce data freshness (12h → 24h), or migrate the data source to BigQuery to remove per-source quotas.
Looker Studio
The native GA4 connector looks like three clicks — but the choices you make in those clicks lock in sampling, scope, and refresh patterns for the life of the report. Here's the right way to wire it up.
Looker Studio
Connecting Looker Studio to a Sheet is the most flexible reporting workflow in the platform — and the easiest to break. This tutorial covers the structure rules that keep the connection stable for years.
Google Search Console
Most owners pick the wrong property type on day one and spend the next year staring at split data. Here's the setup the way SEO specialists actually do it — including the domain-property workflow that captures every subdomain and protocol.
Looker Studio
DIY Looker Studio is a great idea — until it isn't. This is the honest framework: when the cost of self-managing exceeds the cost of hiring, and how to tell which side you're on.