PM Dashboard Design Guide
(2026 Edition)
The 5-layer metric hierarchy, 7 design rules for dashboards that get checked, and 4 tool categories used by PMs at top companies.
Build Metric Fluency Daily — Free →5-Layer Dashboard Hierarchy
1. North Star (Top of dashboard)
The one metric that captures value delivered. Trend visible at a glance.
💡 Example: Daily Active Learners completing a lesson
2. Input Metrics (Second row)
3–5 metrics that cause the north star to move. PM levers.
💡 Example: New signups, activation rate, D7 retention, repeat session rate
3. Health / Guardrail Metrics
Things that must not break. Visible but not central.
💡 Example: D30 retention, NPS, churn rate, error rate
4. Segment Views
Same metrics cut by cohort, channel, geo, device — accessible via filter, not always on.
💡 Example: Same funnel, segmented by acquisition channel and device
5. Diagnostic (Linked out)
Detail views for when a top metric moves unexpectedly. Not always visible.
💡 Example: Event-level funnel, cohort retention curves, experiment results
7 Design Rules
One screen, no scrolling. If it doesn't fit, you're showing too much.
Absolute numbers AND trend. '28% retention' is incomplete without 'was 25% last week.'
Comparison to target. If a metric has a goal, show the gap.
Consistent colour language. Green = up and good; red = down and bad. Never swap.
Timestamp everything. 'As of X date' prevents stale-data confusion.
No vanity metrics. Page views, raw signups, total downloads — these rarely drive decisions.
Link to action, not just numbers. Each metric should answer 'if this changes, what do we do?'
4 Dashboard Tool Categories
Amplitude / Mixpanel
Event-based product analytics. Strong for funnels, retention, cohorts.
Looker / Metabase / Superset
SQL-based BI dashboards. Flexible, slower to set up, deeper for custom analysis.
Notion / Coda embeds
Embed live metric snapshots into PM docs. Great for weekly updates.
Tableau / Power BI
Enterprise BI tools. Common at larger companies with dedicated analytics teams.
FAQ
What's the most common PM dashboard mistake?
Showing too much. PMs often try to include every metric that might be relevant, which produces dashboards nobody reads. Great dashboards have 5–8 core metrics visible without scrolling, organised by hierarchy (north star → inputs → guardrails → diagnostic). Everything else lives in linked detail views.
How often should PMs check their dashboards?
Daily glance (30 seconds) to spot anomalies. Weekly deep review (15 min) to make decisions. Monthly reflection (30 min) to spot trends. Dashboards that aren't part of a reviewing rhythm go stale; dashboards that are checked religiously drive faster decisions.
Who should have access to a PM dashboard?
Default to more visibility. Engineering, design, and leadership should see what you see. Transparent metrics build team alignment and reduce the need for you to be the metrics spokesperson. The only exception: highly sensitive business metrics (revenue per customer, individual user data) that have legal or privacy constraints.
Practice Metric Thinking Daily
Daily scenarios on dashboard design, diagnosis, and metric trade-offs.
Start Free Trial →