A good dashboard answers the questions you would otherwise ask someone to look up. A bad one is a wall of charts nobody opens after the first week. The difference is mostly in what you leave off.
Start with decisions, not data
Before choosing charts, list the decisions the dashboard should support:
- "Do we need to hire, or is the team keeping up?"
- "Which lead sources are worth more spend?"
- "Are we getting paid on time?"
- "Which jobs or sites need attention today?"
Each decision suggests one or two numbers. If a number does not support a decision, it probably does not belong on the main screen.
A useful structure
| Area | Example metrics |
|---|
| Demand | New inquiries, by source; quote requests; conversion to customer |
| Speed | Time to first response; time to quote; time to complete |
| Delivery | Jobs in progress; overdue items; on-time rate |
| Money | Revenue this month vs last; outstanding invoices; days to payment |
| Health | Data freshness; failed syncs; open exceptions |
Pick a few from each area that matter to your business. Six to ten numbers on the main view is usually enough.
Agree on definitions first
The most common dashboard argument is not about the chart; it is about the number. "What counts as a lead?" "Is a job complete when the work is done or when it is invoiced?" Write a one-line definition for every metric and get agreement before building. Show the definition on the dashboard itself.
Connect live data instead of exporting
Dashboards built from manual exports go stale the week someone is busy. Connect the dashboard to the systems that hold the data, through their APIs or a small data store fed by integrations, so numbers update on their own. See CRM integration: start by choosing a source of truth.
Always show when the data last updated. A dashboard that is silently out of date is worse than none.
Different views for different people
Owners, managers and teams need different levels of detail:
- Owners: a handful of trend numbers and anything that needs their decision.
- Managers: their area's queue, bottlenecks and exceptions.
- Teams: today's work and what is waiting on them.
One data foundation can power all three views.
What to leave off
- Numbers nobody acts on.
- Metrics without agreed definitions.
- Charts that look impressive but hide the answer.
- Data that cannot be kept current.
Check that it is used
A few weeks after launch, ask who opens it and what they look at. Remove what is ignored. Add what people still ask for by email.
We design and build dashboards on top of connected data as part of our integrations and data work.