Define every metric before designing the chart
For each metric, document its business meaning, formula, unit, source system, aggregation level, date field, time zone, filters, attribution settings, owner and refresh expectations. Distinguish platform conversions from CRM-qualified leads, customers and revenue; they may describe different stages.
Give metrics names that reveal their definitions. If one team uses 'lead' for a form submission and another for a sales-qualified record, do not blend them under one label. Create separate metrics or document a governed transformation.
Reconcile the underlying data
Choose known test records or dates and trace them from the source platform into the dashboard. Compare totals at the same aggregation level and confirm joins, blends, filters and deduplication do not multiply or discard records.
Check edge cases: zero denominators, missing values, refunds, late CRM updates, cross-domain sessions and events arriving after the reporting date. Show unavailable or incomplete data honestly rather than silently converting every missing value to zero.
Document freshness and reporting windows
Record when each source last refreshed and how late data can arrive. Looker Studio freshness settings balance current results against performance and quotas, and connector behaviour differs. A dashboard that refreshes visually more often than its source does not necessarily contain newer data. citeturn3search0
Use consistent comparison periods, account for time zones and note whether the latest day is incomplete. For near-real-time decisions, show the last successful refresh and a warning when the source exceeds the agreed freshness threshold.
Test the controls users can change
Create a QA matrix for date ranges, campaign filters, channel groupings, device, geography and blended sources. Verify scorecards, tables and charts respond consistently and totals reconcile when filters are cleared. Check report-level and chart-level filters separately.
Review permissions, hidden fields and exported data as well as the on-screen view. Dashboards often travel into presentations or spreadsheets, so definitions and caveats should remain visible when charts are copied or shared.
Create a change and incident log
For each material change, record the date, reason, affected metric, expected impact, validation result and owner. If a chart changes unexpectedly, the log helps distinguish a business shift from a tracking release, source schema change or revised definition.
Assign an owner for both the technical pipeline and business interpretation. When a source is unavailable or a metric is under investigation, label the issue and provide a next-update time instead of letting stakeholders assume the dashboard is complete.
Make QA part of the reporting cadence
Run lightweight checks before recurring reports and a deeper review after website releases, tracking changes, campaign restructures or CRM workflow updates. Automate obvious checks where possible, but retain human review of whether numbers make sense operationally.
Use the UTM Validator to reduce campaign-tagging errors and the Audit Checklist to review the wider measurement system. Continue with the reporting framework and GA4 landing-page guide to connect dashboard reliability with decisions.