THE SHORT ANSWER
A decision dashboard starts with a business question and presents BUSINESS OUTCOME → PRIMARY KPIs → DIAGNOSTIC METRICS → SEGMENTS → TRENDS. It adds context, comparisons, targets, annotations and action thresholds while keeping definitions visible.
Design from the decision backwards
‘Show digital marketing performance’ is too broad. ‘Should we continue increasing acquisition spend while new-customer contribution remains within range?’ identifies an outcome, decision and constraint. It tells the designer what must be visible and what can remain in a diagnostic view.
Write the question at the top of the dashboard specification. If a chart cannot help answer it, explain its guardrail role or remove it.
Use a five-level hierarchy
| Level | Purpose | Example |
|---|---|---|
| Business outcome | State whether value changed | New-customer contribution |
| Primary KPI | Represent the objective | Verified customers at allowable CAC |
| Diagnostic metric | Explain movement | Qualified conversion rate |
| Segment | Locate the change | Channel, product, device or cohort |
| Trend | Show timing and persistence | Comparable weekly series with annotations |
Give every number context
- Definition and source
- Current value and comparable prior period
- Target or operating range
- Material segments
- Volume and data latency
- Annotations for launches, outages and tracking changes
- Owner and response threshold
Different report surfaces can legitimately differ because of filters, retention, modeling and processing. Select an authoritative source for each decision and link to the definition rather than hiding disagreement.
Evidence & context: Google Analytics Help
Make alerts diagnostic, not theatrical
An alert needs a minimum volume, magnitude, persistence and owner. ‘Conversion fell’ creates noise. ‘Qualified checkout completion fell outside its operating range for two comparable periods on mobile web’ creates an investigation.
For commerce application, use Building an E-commerce Growth Dashboard. For the operating loop after the signal appears, continue to turning data into business decisions.
Sources & further reading
- Data differences between reports and explorations
Google Analytics Help. Official explanation of differences caused by supported fields, filtering, retention, thresholds, modeling and processing. It covers GA4 surfaces, not every cross-platform discrepancy.
- Understand user metrics
Google Analytics Help. Official definitions for total, active, new and returning users. Identity limits and configuration can affect interpretation.
- Ecommerce in Google Analytics
Google Analytics Help. Official documentation for ecommerce events and reports. A measurement implementation does not by itself establish causality or profitability.
Examples and exercises are illustrative unless attributed to a source. No independent expert review is claimed.
A correction, a counterexample or an experience worth sharing?
Join the conversation ↗