What to Measure on an Operations Dashboard — and What to Ignore
By Kiran · Founder, Combifer Technologies
Most operations dashboards are built the wrong way round. Someone asks what the system can measure, and everything measurable goes on the screen. The result is a wall of numbers that gets checked enthusiastically for two weeks and then never again.
A dashboard earns its place by answering questions someone actually has. Start there instead.
The four questions worth answering
For an operations lead running a team of twenty to fifty, four questions come up constantly.
What is at risk right now? Not what is late — late is already a failure. What is trending towards late, while there is still time to intervene. Work approaching its due date with no recent movement, work blocked on someone unavailable, work assigned to someone already over capacity.
Where is throughput being lost? If work is moving slower than it should, where does it stall? Almost always at a handoff or an approval, and almost always the same one repeatedly.
Who is overloaded? Distribution of open work across the team. This is not a productivity ranking; it is a capacity check, and it is usually the most immediately actionable thing on the screen.
Is this month different from last? Direction of travel matters more than a snapshot. A completion rate of eighty per cent means nothing on its own and everything next to last month's seventy.
If a number on your dashboard does not help answer one of those, ask what it is doing there.
Metrics that look useful and are not
Total tasks completed. Rewards splitting work into smaller pieces and says nothing about whether the right things got done.
Individual completion counts. Encourages people to pick easy work and avoid the hard, ambiguous tasks that matter most. It also quietly turns your dashboard into a surveillance tool, which changes behavior in ways you will not like.
Activity volume — logins, messages, comments. Measures presence, not progress. Rewards performative busyness.
Average time to completion across everything. Averages across mixed work are meaningless. A two-minute approval and a three-week project in the same average produce a number that describes nothing.
Percentage complete on long-running work. Self-reported and consistently optimistic. Ninety per cent complete is the most durable state in project management.
The one that is worth more than the rest
Time spent waiting, broken down by where the waiting happened.
Most delay in an operations team is not work taking long. It is work sitting still — waiting for an approval, a decision, an input from another team, a reply. If you can see where the queues form, you can usually fix them, and the fix is often as small as changing who is allowed to approve something.
This is the metric that leads directly to an action. Most of the others lead to a discussion.
Design so the exception is visible
A good operations dashboard is mostly empty. It shows what needs attention, not everything that exists.
If eighteen of twenty projects are fine, the screen should be about the two that are not. A dashboard that renders all twenty equally makes the reader do the filtering, and the reader will stop opening it.
Encode state in form as well as number — a colour, a chip, a severity marker — so what matters reads at a glance rather than after study.
Different questions for different roles
A team lead needs today: who is available, what is blocked, what is due.
A department head needs this month: throughput, where queues form, whether load is distributed sensibly.
A founder or director needs the trend: is this improving, which areas are consistently slower, where does risk concentrate.
The same numbers at different resolutions. Trying to serve all three on one screen serves none of them.
A test for anything on the screen
For each number, ask: if this went the wrong way, what would I do differently tomorrow?
If there is a clear answer, it belongs. If the answer is "look into it" or "nothing, but it is good to know", it is context, not a metric — and context belongs in a report someone reads monthly, not on a screen meant to prompt action.
Applied honestly, this test usually removes about half of what is currently displayed. The remaining half gets looked at, which is the entire point.