Build a product analytics dashboard with useful denominators
A useful dashboard answers recurring questions with consistent dates, filters, and metric definitions. Start with traffic, activation, and revenue, then add charts only when they change a decision.
Start with three questions
Where does activity come from? Which actions are completed? What value follows those actions? A web-traffic dashboard can use visitors, pageviews, sessions, sources, and top pages. A product dashboard can use custom-event trends and explicitly submitted revenue.
Pick an owner and a review cadence. A chart that nobody reads can make the page slower and the question harder to see. State the intended action for a meaningful change: investigate a route, inspect a funnel step, or validate a campaign.
Choose the chart for the data
Number cards summarize totals or rates. Line and area charts show trends. Bars compare categories; pies show a small category composition; tables preserve exact values. Not every metric supports every chart type.
The current builder offers six chart types and constrains them by result shape. Category breakdowns support bars, pies, and tables; time-based metrics support trends and totals. Compare percentages with the relevant denominator rather than plotting unrelated units together.
| Question | Useful card |
|---|---|
| How much activity? | Number total |
| Did activity change? | Line or area trend |
| Which pages dominate? | Bar or table breakdown |
| Where does traffic come from? | Source bar, pie, or table |
Keep dates and identity consistent
Use the top-bar date range for all saved metric cards. A previous-period comparison is only useful when the windows and filters match. Label daily anonymous visitors distinctly from explicit accounts; neither is a count of lifetime unique people.
A bounce rate or pages-per-session value is a ratio. Do not sum daily percentages to produce the period metric. Aggregate the numerator and denominator, then compute the ratio. Read the session definition for one-event durations and midnight behavior.
Use templates, then prune
Start with web traffic, product activity, campaign performance, or site performance. Remove cards that repeat the same fact. Add Web Vitals when performance affects the question, and use the SQL Explorer for an investigation that preset metrics cannot express.
Saved cards use defined metric queries; arbitrary SQL charting is separate. Payment reconciliation and weekly retention have their own reports. A growth template does not automatically add either report to a dashboard.
Frequently asked questions
Can every metric use a pie chart?
No. Chart options depend on whether a metric returns a total, time series, or category breakdown.
Should every card have its own datepicker?
Saved dashboard cards follow the top-bar date range so a report uses consistent windows.
Can I start from a template?
Yes. Templates cover web traffic, product activity, campaigns, and site performance.