Field Note · Management Systems
A dashboard is not a management information system.
A dashboard shows things. A management system helps somebody decide what to do about them. The distinction sounds obvious. In practice, it is where a surprising amount of money disappears.
Management Systems · Applied interpretation
I have seen reporting environments with enough charts to decorate an airport control room. Everything had a colour, a trend and an owner. Then one number moved in the wrong direction and the meeting immediately returned to the oldest management technology available: people asking one another what was actually going on.
That is not a criticism of dashboards. I like good dashboards. The problem starts when the screen is asked to substitute for the management system behind it.
The useful unit is not the chart. It is the management question the chart helps answer.
Start with the awkward question
Suppose the schedule forecast slips. Is that simply something to display? Or does the change cross a threshold, threaten a commitment, require another function to act, or need a decision from someone with more authority?
Once those questions are clear, the information begins to have a job. A schedule forecast has meaning because people know when confidence is no longer good enough. A risk indicator matters because something changes when the threshold is crossed. An action is not “closed” because somebody changed its colour to green; it is closed when there is enough evidence that the problem has actually changed.
The dashboard is then what it should have been all along: a useful window into a larger management system. Different people can look through different windows without inventing different versions of reality.
Follow one signal
A milestone turns amber on Tuesday. By Thursday the meeting pack has a new colour, but nobody has established whether the supplier date moved, whether commissioning is exposed, or who can trade scope for time. The dashboard has been updated. The management problem has not.
The useful question is not “is it amber?” It is “what decision became necessary when it turned amber?”
Follow the signal past the screen
What changed?
Not just the current number. What moved, and is the movement real?
Why do we care?
What objective, commitment, exposure or forecast does it affect?
Who needs to decide?
Not who owns the PowerPoint. Who actually has the authority to change the situation?
What happens now?
A decision, an escalation, a commitment, a review — or perhaps nothing. “Nothing” is allowed if it is a conscious decision.
If the reporting environment cannot help answer those questions, the organization probably has reporting. That may be perfectly useful. It just should not be confused with decision support.
The technology comes later
This is also why I am cautious when MIS discussions begin with the tool. Power BI, a data platform and AI can make a coherent system faster, richer and easier to interrogate. They cannot decide what the organization needs to know, which signal deserves attention, or who has the right to act. Those are management design choices.
AI makes this distinction more important, not less. Give it a coherent information structure and it can help detect changes, connect evidence, surface contradictions and challenge assumptions. Give it a confused reporting process and it can produce a remarkably articulate version of the confusion.
A practical test
Take one important red or amber signal from the current dashboard and follow it past the screen. Can you identify the evidence behind it, the management question it raises, the person with authority to decide, and what changes after that decision? If the trail disappears between the chart and the meeting, do not start by redesigning the dashboard. Design that missing piece of the management system.
Follow the signal
Inspect a worked control problem
See what happens when reported status stays green while the delivery forecast deteriorates.
Inspect the decision architecture
Go behind the dashboard into evidence, interpretation, authority and action.