A Dashboard People Actually Read
Last updated:
Why dashboards get abandoned
The typical dashboard shows thirty metrics because thirty were available. Nobody can hold thirty numbers in mind, so nobody looks, and within a month it is a link nobody clicks.
The fix is subtraction. A dashboard should answer one question for one role, and the discipline is in what you leave out.
One screen per role, five numbers
- Owner: revenue this month vs last, cash position, pipeline, one operational risk
- Operations: today's workload, exceptions needing a decision, anything overdue
- Sales: my pipeline, my activity, what needs chasing today
- Support: open by age, breaching soon, today's volume against normal
Five numbers each, chosen because someone would do something differently if the number changed. If nobody would act on it, it belongs in a report, not on a dashboard.
The test for every tile: what would you do differently if this number doubled? No answer means it should not be there.
Every number needs a comparison
A number alone is meaningless. “47 orders today” means nothing without knowing whether that is good.
- Against the same day last week, which handles weekly patterns
- Against the recent average
- Against a target, where one genuinely exists
- With a small trend line, which conveys direction faster than any figure
Make every number clickable
“12 overdue jobs” should be a link to the twelve jobs. A dashboard that reports a problem but not which records are affected sends the reader to another system to find out, and after three of those they stop looking.
This single decision does more for adoption than any visual improvement. The dashboard becomes a way into the work rather than a report about it.
Put it where people already are
- As the landing page of a system they already open
- A daily digest by email or chat, at a fixed time
- A screen on the wall for shared operational numbers
- Never a separate tool with a separate login — the login is where adoption dies
The morning email version is consistently the highest-engagement format we build. It arrives without anyone deciding to look, and it is read on a phone before people reach a desk.
Getting the data right first
A dashboard is only as trustworthy as what feeds it. One occasion where the numbers were wrong and nobody said so costs more credibility than months of accuracy earn.
- A visible last-updated timestamp, always
- An explicit message when data is stale, rather than showing old numbers as current
- One agreed definition per metric, documented and owned
- A reconciliation against the source system, daily
Frequently asked questions
Should we buy a BI tool or build one?
How often should it refresh?
How many dashboards should we have?
What does one cost?
Got a process somebody still does by hand?
Show us how it works today and we will tell you what it would cost to automate — or when a cheaper change would do the same job.
Related services
What we build for problems like this one