A management dashboard is one self-refreshing screen holding the five to eight kpis your board steers on: revenue and margin against budget, cash position, your most important operational indicator, customer retention and staff turnover. What separates it from a departmental dashboard is not the level of detail, but the kind of decision hanging off it.
In most organizations those figures already exist. Finance delivers them after every month-end close, sales keeps its own list, and operations has a planning board updated each morning. What is missing is rarely a figure. What is missing is one place where they sit side by side, over the same period, at the same moment.
And that is rarely where technology fails you. It is where ownership does. When someone in the board meeting says the margin cannot be right, that question travels to whoever administers the system. They can tell you exactly which table the figure came out of, but not whether it is the right figure. That second one is a decision, and decisions belong at the table.
So what follows is not only which kpis belong on a screen like that and what one looks like in practice, but also why it must not become a stack of departmental screens, who is accountable for each figure, and how often you should actually be looking at it.
What is a management dashboard?
It is one screen that refreshes itself, holding the figures your leadership team needs in order to decide. Broad in subject, narrow in count: each domain contributes one or two figures, and together they answer the question of whether the organization is on course.
The difference with a management report is who does the work. A report like that gets made: someone pulls exports out of three systems, stitches them together, writes a commentary and delivers the whole thing when the month has long since closed. A dashboard is simply there. That looks like a difference in convenience, but it is a difference in timing. What you read two weeks after the fact, you can only explain.
None of which makes the report redundant. You need it for the commentary, for your bank and for the annual accounts. It just isn’t the moment you adjust course, and in a lot of organizations that is quietly what it has become.
The difference with a departmental dashboard is worth keeping sharp too. A dashboard for sales or finance goes deep inside a single domain. A management dashboard does the opposite: one figure per domain, where the combination is what produces the insight. Revenue up twelve percent is news. That same twelve percent next to a margin down two points and absence creeping up is a decision.
The phrase management information gets used as a synonym, and it isn’t quite one. Management information is everything your leadership team needs in order to steer, including the story around it, the assumptions and the controller’s commentary. The dashboard is the measurable half of that: the part you can capture in a figure and that can update itself. The rest of your management information stays human work, and that is not a shortcoming of the screen.
Textbooks usually split dashboards into three layers: strategic, tactical and operational. In a mid-sized company that split is tidier than reality. There you have one leadership team discussing both the annual figures and next week’s planning on a Thursday morning, and that needs one screen with two speeds in it rather than three screens.
Which kpis belong on a management dashboard?
Don’t start from the figures you happen to have. Start from the four questions every leadership team asks anyway: are we making money, do customers come back, can we deliver what we sell, and do we keep the people who do that. One or two figures per question, and you land on six by yourself.
- Revenue against budget. Cumulative and per month, with the gap alongside. Without that gap it is a reading, not a steering figure.
- Gross margin as a percentage. Revenue without margin tells you nothing about whether the growth is earning you anything. This is the figure that most often triggers a different decision than the revenue next to it.
- Cash position and working capital. What is there, what is coming, and how many days you can run. For most mid-sized organizations this is the figure that costs the most sleep.
- One operational indicator. Utilization, lead time, delivery reliability or daily output. Which one it is differs per company. That exactly one of them sits on the board screen does not.
- Customer retention or recurring revenue. Winning new customers is visible work; customers quietly falling away is not. This figure makes the second one visible before it shows up in revenue.
- Staff turnover and absence. Two figures that tell you months in advance what your operational indicator is about to do.
Six is a comfortable starting point, eight is the ceiling. That five to eight isn’t an arbitrary range has been known since 1956: George Miller described in what is now a classic psychology paper that a person can hold roughly seven separate items at once, give or take two. Dashboard vendors like quoting that number without naming the source, but the point behind it holds up. Above eight tiles they compete for the same attention, and then the tile with the brightest colour wins instead of the tile with the biggest problem.
What separates a kpi from an ordinary figure, incidentally, is the target beside it. With no target attached, you are reading off a value. Eight values side by side is a reference work, not a dashboard.
The hardest choice on that list is the operational indicator, because there is no standard answer for it. The question that points to it is this one: which figure runs ahead of our revenue? In an installation or consulting business that is utilization, because an empty week now is an empty invoice two months out. In manufacturing it is delivery reliability or lead time. In wholesale it is stock turnover or the share of orders that ship complete. If the leadership team cannot agree, that is information in itself: apparently nobody knows which operational figure drives your revenue, and that is a more useful conversation than arguing over which chart goes on top.
Example: what a screen like that looks like in practice
Take an installation company with sixty employees and a leadership team of five. Their screen has three layers, and that is not a design choice but a reading order.
Six tiles at the top. Per tile the figure, the target and the gap: revenue year to date against budget, gross margin as a percentage, cash position in days, engineer utilization, the number of projects past their planned end date, and turnover over twelve months. Six figures, ten seconds.
Below that, only the charts that ask for an explanation. Not six of them, but the two or three sitting outside their target this month, with thirteen months of history alongside so you can tell season from trend.
An exception list at the bottom. The five projects with the largest variance, with name, owner and amount. This is the part the meeting is actually about, and it is no accident that it is the only part with names in it.
What is not on there matters just as much. No revenue per salesperson, no time tracking per engineer, no drill-down to individual invoice lines. Those figures do exist, and whoever needs them clicks through. On the board page they would only take attention away from the six figures that do matter.
Those three layers also decide how the screen gets used in the meeting, and that is the whole point of the arrangement. The first layer answers, in ten seconds, whether something is going on. The second answers the follow-up: is this a one-off or a line. The third points at who owns it. Without that order every meeting starts at the detail, and you are half an hour in before anyone asks whether that detail actually matters.
Not the sum of your departmental dashboards
Once finance, sales and operations each have their own screen, the conclusion looks obvious: put them side by side and you have a management dashboard. It doesn’t work that way, for a simple reason. Every departmental screen is built to answer one question deeply, while at board level you need the opposite: shallow, but across every question at once.
A board screen glued together from five departmental dashboards isn’t an overview, then. It is a meeting with a scrollbar.
The right relationship is a layer above, not a row alongside. The financial dashboard stays, with margin per client, payment terms and working capital in detail. The sales dashboard stays, with pipeline by stage and win rate. From each of those, one figure travels up to the board screen, with a click back down for anyone who wants to know how that figure came about. In Power BI that click is called drillthrough: you click a tile and land on the detail page with that tile’s filters already applied.
That only works if the same data model sits under both screens. If the board screen calculates revenue differently from the financial one, clicking through is worse than not clicking through. You then watch two figures fail to add up with your own eyes, and the conversation is over before it started.
If you have no departmental screens at all, by the way, there is no need to wait for them. Start at the top instead. Six figures at board level means six sources that need to be right, and that is a tidier assignment than three departments each building their own dashboard and discovering afterwards that they calculate revenue differently. The detail screens grow underneath it later, in the order questions come up about them. The other way round, from the bottom, you more often end up with four screens that each add up on their own and don’t together.
One owner per kpi, and they don’t sit in IT
Every figure on the screen needs one person who can say something about it in the meeting. Not whoever wrote the query, but whoever can change it.
In practice that divides itself up, as long as you say it out loud. The finance lead explains why margin is slipping and what is being done about it. The operations lead explains why utilization is behind and when that gets fixed. The managing director watches whether the growth coming out of it is growth anyone is earning anything on. Three roles, three figures, three people who cannot point at IT on a Thursday morning.
Because that is the reflex that kills a management dashboard most often. A figure sits outside its target, someone doubts the measurement, and the question moves to whoever administers the system. They can answer it, but only technically: they can show where the figure came from, not whether the definition underneath still matches how you run the business. A figure with a decision hanging off it is not an IT question but a board question. That is not a matter of the org chart, it is a matter of who is at the table when the number has to carry something.
So put the name on it. Power BI has a feature for exactly this: with goals in Power BI you attach a target and an owner to each figure, and the system keeps the progress and the notes alongside it. Whether you use that feature is secondary. That the name is written down somewhere is not: a kpi without an owner belongs to everyone in the meeting, and therefore to no one.
That is exactly where data-driven working trips up in practice. Not on the model and not on the technology, but on the question of who has to explain something on Thursday morning.
How often should it refresh, and how often should you discuss it?
Two questions that get tangled together, while the answer to the second one determines the answer to the first.
Refreshing is the easy one. Once a night is almost always enough for a management dashboard. The figures a board steers on do not move by the hour, and a cash position running live produces no better decision, only more moments of looking.
Discussing is where the gain sits, and where most meeting time disappears. The classic mistake is picking one frequency for all the tiles at once. You want utilization weekly, because you can change something about it within a week. Turnover over twelve months belongs in the quarterly meeting; look at it every week and you are looking at noise, and you will start having opinions about it.
The rule of thumb that follows: discuss a figure as often as you can influence it, not as often as it refreshes.
In practice that means two rhythms on one screen. Weekly, the three or four figures that need something doing this month, each with a name and an agreement behind it. Monthly or quarterly, the rest, worked through in one go. That saves meeting time, and it heads off the pattern where everyone looks at the same eight figures every week while there is nothing to say about six of them.
How do you build a management dashboard?
Five steps, in this order. The order is half the work: nearly every pitfall you meet in a project like this is one of these steps that got moved forward or back.
- Step 1: decide which decisions need to get better. Not which figures you have, but which three to five decisions your leadership team makes each quarter and currently makes too late or too vaguely. Skip this and you build a screen that looks good and changes nothing. That is not the best-known pitfall, but it is the most expensive.
- Step 2: pick the figure that drives each decision. That is how you land on six to eight instead of twenty-four. The pitfall here is that nobody dares turn a request down: every extra tile feels free, while it costs attention from the tiles already there.
- Step 3: pin down definition, source and target. Per figure: where does it come from, how exactly is it calculated, and what is the target beside it. This is the dullest step and the only one that decides whether the screen is still in use a year from now. As long as sales, finance and operations each hold their own definition of revenue or lead time, the meeting is about the source instead of the outcome.
- Step 4: build the model, then the screen. Connecting sources, cleaning data, setting relationships. Most of the work and least of the applause sits here. The pitfall is doing it backwards: a good-looking prototype on an export first, then discovering that the export has to be made by hand every month.
- Step 5: attach a moment to it. Who discusses it, when, and what happens when a figure falls outside its target. A dashboard with no meeting around it gets looked at for two weeks and then not again.
Only at that fourth step does the question of what you build it with come up. For a first version Excel is fine, and it is often the most honest test: if your leadership team doesn’t look at the manual overview for three months, it won’t look at the automated one either. Excel breaks down as soon as the figures come from more than two sources or several people update them. The time then goes into merging instead of into the conversation, and versions start to drift apart. That is the moment for a dashboard tool, whether that is Power BI, Qlik or something else.
What building a model like that costs and how long it takes is set out in our article on getting a Power BI dashboard built.
When a management dashboard is not your first step
In three situations we would suggest waiting a little.
- Your leadership team has no fixed meeting. With no recurring moment there is nowhere for the screen to change anything. Set that meeting up first, on paper if need be.
- The departments don’t have their own figures straight yet. A board screen draws its numbers from below. If the margin is wrong in finance, it is wrong on the board screen too, only now with a bigger audience.
- A system change is coming. If you are moving to a different ERP or accounting package within six months, you build every connection twice. Wait until the new source is in place.
In all three cases the answer is not never, it is something else first. And you can start in the meantime: put six figures into one document by hand for a few months and discuss them at a fixed moment. If that changes the conversation, you know the screen will pay for itself. If it changes nothing, you are a few hours down instead of a whole project.
Does your leadership team discuss the figures each month, or the decisions underneath them? Want to know which six to eight figures belong on that one screen for your organization? Get in touch, and we’ll walk through them together.
Frequently asked questions
What is a management dashboard?
A management dashboard is one self-refreshing screen holding the five to eight figures the board steers on, spread across finance, customers, operations and people. The difference with a departmental dashboard is not the level of detail but the reach: one figure per domain, where the combination is the insight.
Which kpis belong on a management dashboard?
Start from four questions: are we making money, do customers come back, can we deliver what we sell, and do we keep the people who do that. One or two figures per question gives you revenue against budget, gross margin, cash position, one operational indicator, customer retention and staff turnover.
How many kpis should a management dashboard show?
Five to eight, and that is not an arbitrary range: a person can hold roughly seven separate items at once. Above eight tiles they compete for the same attention, and the tile with the brightest colour wins instead of the tile with the biggest problem.
What is the difference between a management dashboard and a management report?
A management report gets made: someone pulls exports, stitches them together and delivers the result after the month-end close. A dashboard is simply there. That looks like a difference in convenience, but it is a difference in timing: what you read two weeks after the fact, you can only explain.
Who is responsible for a management dashboard?
One person per kpi who can say something about it in the meeting, and that is not whoever administers the system. Running the technology and owning the figure are two different roles. Without that second one, a kpi belongs to everyone in the meeting, and therefore to no one.
Can you build a management dashboard in Excel?
For a first version, easily. It breaks down as soon as the figures come from more than two sources or several people update them: the time goes into merging instead of into the conversation, and versions start to drift apart. That is usually the moment to move to a dashboard tool.
How often should a management dashboard refresh?
Once a night is almost always enough. The figures a board steers on do not move by the hour, and a cash position that runs live produces no better decision, only more moments of looking. How often you discuss a figure matters more than how often it refreshes.
