The Best KPI Dashboards Boards Actually Read Have Five Metrics, Not Fifteen

In 2016 I was advising a divisional board ahead of its quarterly review, and I sat in on the meeting to watch how the pack was actually used. The operations director presented a dashboard with nineteen KPIs on a single screen — a traffic-light grid, green and amber and one red, arranged in four neat rows. The board spent close to forty minutes on it. What I watched was the chair’s finger moving across the screen, settling on the metrics that were green and asking comfortable questions about them, while a single amber metric in the third row — days sales outstanding, creeping up for two quarters — was glanced at and never discussed. The cash-collection problem that amber number was signalling surfaced as a genuine crisis the following quarter. The dashboard had not hidden it. The dashboard had shown it, in amber, in row three. But nineteen metrics on one screen meant the board discussed the numbers that invited easy conversation, not the one that needed a decision, and the one that needed a decision got lost in the grid.

(This article was created with AI assistance; all stories and insights are based on 35 years of real client work.)

That meeting is the clearest example I have ever seen of the central problem with KPI dashboards at board level: the more metrics you show, the less reliably the board discusses the right one. A board does not have the time, and is not structured, to work through nineteen numbers and rank them by importance live in the room. It needs the dashboard to do that ranking for it — to put the exception in front of it and let the rest recede. This piece walks through the method: the exception filter that cuts your KPI list to the five that belong at board level, what each of those five must carry to be decision-ready, and the thirty-second scan test that tells you whether your dashboard surfaces the exception or buries it.

Why a fifteen-metric dashboard fails a board

The fifteen-metric dashboard usually comes from a good intention and a category error. The good intention is completeness: the presenter wants the board to have the full picture, so they include every KPI the function tracks. The category error is treating the board like a management team. A management team runs the business and needs the full operational picture because it acts on all of it. A board governs the business and needs the exceptions because it acts only on the things that are off-track or off-plan. Showing a board every metric the management team watches is showing the board a tool built for a different job, and the board responds by using it badly — not because the board is weak, but because the tool is wrong for the room.

There is a specific failure mode that follows. When a board is shown more metrics than it can rank, it does not rank them by importance. It ranks them by salience — by which number is most colourful, most familiar, or easiest to ask a question about. A red metric draws comment because red invites it. A green metric the board is fond of draws comment because it is reassuring. The amber metric in row three that actually needs a decision draws nothing, because amber is ambiguous and row three is far from the eye’s natural starting point. The dashboard’s information is all present, but the board’s attention is allocated by visual accident rather than by importance. The way executives actually read a dashboard is not row by row; it is by jumping to whatever catches the eye, which is why a dashboard that does not engineer the eye toward the exception leaves attention to chance.

The reframe is to treat the board dashboard as an exception report, not a status report. A status report says “here is everything, and here is how it is all doing.” An exception report says “here are the few things you need to look at, and here is the one that needs a decision.” A board wants the second. Designing for the second is what gets you from fifteen metrics to five, because once you ask “which of these would actually change a board decision if it moved?” most of the fifteen turn out to be confirmation, not exception — reassuring to include, but not board business.

The exception filter: how to choose the five

The exception filter is a single question applied to every candidate metric: would a director do anything differently if this number moved ten percent against plan? If the answer is yes — the board would ask for action, change a decision, escalate, or intervene — the metric is a board metric and it passes the filter. If the answer is no — the board would note it and move on, because acting on it is management’s job not the board’s — the metric fails the filter and belongs in the management pack. The ten-percent figure is a calibration device, not a rule; the point is to force a judgement about decision-sensitivity rather than relevance. Almost every KPI is relevant. Only a handful are decision-sensitive at board level.

When you run the filter honestly across a typical nineteen-metric dashboard, you land near five, and the five are usually predictable in shape: one on financial performance against plan, one on cash or liquidity, one on the single biggest strategic bet currently in flight, one on a material risk the board is accountable for, and one on whatever is the live issue this particular quarter. The exact five vary by organisation and by period — that last slot is deliberately movable — but the discipline is constant: each of the five earns its place by being a number on which the board would act, not merely a number the board might like to see. The presentation choices that hold an executive audience all start from this kind of ruthless selection, because attention is the scarcest resource in the room and every metric you show spends some of it.

The filter also resolves the political objection that always comes up: “but the head of [function] will want their metric on the board dashboard.” The answer is that the board dashboard is not a status board for the executive team’s pride; it is a decision tool for the directors. A metric that fails the exception filter can live, prominently, in the management pack that the executive team works from — it is not being deleted, it is being placed where it is actually used. Framing the cut this way — placement, not demotion — usually defuses the objection, because the function’s metric is still tracked and still visible, just not occupying scarce board attention it would not change a decision on.

What each of the five metrics must carry

Cutting to five metrics is only half the work; each of the five then has to be presented so a non-specialist director can read it in seconds. A bare number is not decision information. Each board metric needs five things on it, and the discipline of carrying all five is what separates a board dashboard from a spreadsheet export. The number itself, large enough to read across a room. The trend against the prior period, shown as a direction not just a second number — an arrow, a sparkline, a sign. The target or threshold, so the board can see whether the number is where it should be. A one-line reason for any movement that matters, in plain language. And a status — on track, watch, or act — that states the presenter’s own read of whether this metric needs board attention.

The status line is the part most dashboards omit and the part that does the most work. A traffic light alone does not say what the board should do; amber especially is ambiguous, which is exactly why the amber metric in 2016 got ignored. Replacing the colour with a verb — “watch” or “act” — removes the ambiguity and pre-commits the presenter to a read. “Days sales outstanding: 58 days, up from 47, threshold 45, rising because two large accounts are slow-paying, status: ACT” is a board metric. “DSO: 58 (amber)” is a number the board will glance past. The first asks for a decision; the second asks the board to work out whether a decision is needed, and a board that has to work that out usually defers it.

The one-line reason is the second piece dashboards skip, and it is what stops the board asking the obvious question and stalling the meeting. If a metric has moved and there is no explanation on the slide, the first thing a director does is ask why — and now the meeting is in an unscripted detour while the presenter explains. Putting the reason on the slide pre-empts the question, keeps the board moving, and demonstrates that the presenter understands the number rather than merely reporting it. Five metrics, each carrying these five things, is a board dashboard that a director can act on without a single clarifying question — which is the standard to aim for.

The thirty-second scan test

The test that tells you whether your dashboard is board-ready is the thirty-second scan. Show the dashboard to someone who is not in your function for thirty seconds, then take it away and ask one question: which metric needs discussion this period, and what is the decision? A board-ready dashboard produces an immediate, confident answer — the design has surfaced the exception and the reader has caught it. A dashboard that fails produces hesitation, or worse, the wrong metric: the reader names the red one or the familiar one rather than the one that actually needs a decision, which is precisely the failure that played out across forty minutes in 2016.

Thirty seconds is longer than the eight seconds a single decision slide gets, because a dashboard legitimately carries more and the board does spend longer on it. But it is still short, and it is still a simulation of real conditions: the reader has not studied your metrics, does not know the back-story, and is allocating attention the way a director does — by jumping to whatever the design pushes them toward. If the design pushes them toward the exception, they catch it. If the design treats all five metrics as visually equal, they default to salience, and you learn that your dashboard is a status board wearing the clothes of a decision tool.

The scan test is also diagnostic about your five-metric selection, not just your layout. If the reader catches an exception but it is not the one you intended, sometimes the layout is fine and the selection is wrong — you have included a metric that draws attention without deserving it. If the reader catches nothing, either there genuinely is no exception this period (in which case say so explicitly: “all five on track, no decision required”) or the exception is buried by equal visual weight. Run the test before every board meeting and you will know, in advance, what the board is going to discuss — which means you can decide whether that is what you want them to discuss while there is still time to change it. The practices that distinguish a board-ready pack are mostly tests like this one: cheap to run beforehand, expensive to skip.

Designing the dashboard so the exception surfaces itself

A five-metric dashboard that gives all five metrics equal visual weight can still bury the exception, because equal weight forces the reader to do the ranking. The design job is to make the exception heavier — to let the layout itself point at the metric that needs a decision. This is not decoration; it is the difference between a dashboard that does the board’s ranking for it and one that leaves the ranking to chance. The mechanics are simple. The metric with status “act” goes top-left, where the eye starts, and gets more space and stronger contrast. The metrics on track recede — smaller, quieter, confirmatory. The dashboard is not a grid of equals; it is a hierarchy with the exception at the top.

In 2022 I worked with a chief operating officer who had inherited exactly the nineteen-metric grid from the earlier story — different organisation, same disease. We ran the exception filter and landed on six metrics, then cut to five. We rebuilt the slide so the one metric on “act” status occupied the top-left quadrant at twice the size of the others, with the four on-track metrics in a quiet row beneath. At the next board meeting, the chair opened the dashboard discussion by pointing straight at the top-left metric and asking what the board needed to decide about it. The conversation went where it needed to go in the first sixty seconds, not the last five minutes. The COO told me afterwards the meeting had felt “steered” for the first time — not manipulated, but steered toward the decision that mattered, by a layout that did the steering silently.

The principle generalises beyond the single “act” metric. When all five are genuinely on track, the design should say so loudly — a clear “no board decision required this period” line at the top — rather than presenting five green metrics and inviting the board to manufacture a discussion out of them. When two metrics need attention, rank them: primary exception top-left, secondary exception beside it, the rest beneath. The board’s attention is a budget, and the dashboard’s layout is how you spend it. Spend it on the exceptions, deliberately, and the meeting follows the spend.

Where the other ten metrics live

The metrics that fail the exception filter do not disappear; they move to the management pack, and the management pack is a different document with a different purpose. The board pack carries the five exceptions and exists to drive board decisions. The management pack carries the full operational picture and exists for the executive team to run the business. Keeping them as two documents, with two audiences and two purposes, is what allows the board dashboard to stay at five without anyone feeling their metric has been deleted. The completeness instinct is satisfied — everything is still tracked — it is just satisfied in the right document.

This separation also protects you when a director asks about a metric that is not on the board dashboard. “That sits in the management pack — it’s tracked weekly by the exec team, currently on plan, happy to pull it up if useful” is a confident answer that demonstrates the metric is governed without it having to occupy board attention by default. The two-document structure means you are never caught out by an off-dashboard question; you have simply chosen, deliberately, which metrics earn the board’s scarce attention and which are available on request. That deliberate choice is the senior skill, and it is visible to the board as competence rather than as omission.

One thing to do before your next board dashboard

Take your current board dashboard and list every metric on it. Run the exception filter down the list: for each one, ask whether a director would change a decision if the number moved ten percent against plan. Keep only the metrics that pass — you will land near five — and move the rest into a separate management pack. Give each surviving metric the five things it needs: the number, the trend direction, the threshold, a one-line reason, and a status verb of on track, watch, or act. Put the “act” metric top-left and let the on-track ones recede. Then hand it to someone outside your function for thirty seconds and ask which metric needs discussion and what the decision is. When they answer correctly and instantly, your board will too.

Frequently asked questions

Is five a hard rule, or can a board dashboard have six or seven?

Five is a target, not a law. The real rule is the exception filter: show the metrics on which the board would act, and no more. For most organisations that lands between four and six. The number matters less than the discipline behind it — if you have seven metrics and all seven genuinely pass the exception filter, seven is defensible, though it is worth pressure-testing whether the sixth and seventh are exceptions or whether they crept in because a function wanted visibility. The failure is not “six instead of five.” The failure is fifteen, where the filter has not been run at all and the dashboard is a status export rather than a decision tool. Aim for five, accept four to six, and treat anything in double digits as a sign the filter has been skipped.

My board has always seen the full dashboard — won’t cutting it look like I’m hiding performance?

This is the most common worry and the two-document structure is the answer. You are not cutting the metrics; you are moving the operational ones to a management pack that stays available. When you introduce the change, say so explicitly: “The board dashboard now shows the five metrics that may need a board decision this period; the full operational set is in the management pack behind it, and I’m happy to pull up any of it.” Framed that way, the board reads the change as sharper governance, not concealment — you are bringing them the exceptions and keeping the rest one page away. In practice boards welcome it, because the forty-minute slog through nineteen metrics was never enjoyable for them either. What looks like hiding is hiding the detail entirely; what you are doing is positioning it.

What if every metric on my dashboard genuinely is on track — what do I present?

You present that, clearly and briefly, and you do not manufacture a discussion. A board dashboard with all five metrics on track should carry an explicit line — “all five board metrics on or ahead of plan; no board decision required this period” — and then the board can spend its time on the forward-looking items that actually need it, like strategy or risk. The mistake is presenting five green metrics with no framing and letting the board feel obliged to interrogate them, which generates a discussion with no decision at the end of it. Naming the “no decision required” state is a sign of confidence and it respects the board’s time. A period with no exceptions is good news; present it as good news and move on.

How is a board KPI dashboard different from the one I show my own management team?

The audiences do different jobs, so the dashboards do too. Your management team runs the business and acts on the full operational picture — it needs the complete dashboard, updated frequently, with every metric the function tracks. The board governs the business and acts only on exceptions — it needs the five metrics where a board decision might be required, with the rest available but not on screen. Showing the board the management dashboard is the original error: it hands a governance audience a management tool and then watches them use it for the wrong job. Keep two documents with two purposes. The management pack can be as detailed as your team needs; the board pack should be as sparse as the board’s decision rights require. The skill is knowing which metrics belong in which, and the exception filter is how you decide.

The Winning Edge — weekly newsletter

The Winning Edge is a weekly (Thursday) newsletter for senior professionals who report to boards and executive committees. One short email a week on the structural moves that turn a wall of numbers into a board decision. Subscribe to The Winning Edge →

For the full set of skills behind board-level reporting — slide structure, storytelling, confidence, and delivery — the Complete Presenter bundle of seven products brings them together for £99, with lifetime access.

About the author

Mary Beth Hazeldine is Owner & Managing Director of Winning Presentations Ltd. With 24 years of corporate banking experience at JPMorgan Chase, PwC, Royal Bank of Scotland, and Commerzbank, she advises senior professionals across financial services, insurance, consulting, and technology on the structural moves that turn operational data into board-level decisions.