Your dashboard is not your business

Here is a test worth running at the next management meeting. For each chart on the dashboard, ask the people in the room: name one decision made in the last twelve months because of how that number moved. Not a discussion about the number. An actual decision.

The answers are usually uncomfortable. In a typical Mittelstand controlling pack, somewhere between half and two-thirds of the charts cannot pass this test. The numbers are correct, the visualizations are clean, and nothing in particular happens after the meeting because of any of it. Microsoft’s own adoption guidance for Power BI and Fabric draws the line in the same place. It calls it a common misconception that adoption relates primarily to usage or the number of users: “Adoption isn’t just about using the technology regularly; it’s about using it effectively.” Effectiveness, it adds, is much more difficult to define and measure.¹

This is the BI failure mode that does not get talked about much in mid-sized companies, because nothing is obviously broken. The data is fine. The visualizations are fine. The dashboard gets opened. It just does not change what the company does. Below are the three categories of chart that most often fail the test.

1. Charts that show what happened but not what to do about it

A chart shows that revenue in the southern region is down 14% over four months. The trend is clear. What happens next?

In most companies, what happens next is a conversation. Someone asks why. Someone else promises to look into it. The chart appears again in the next monthly review, still red, and the same conversation happens again. The dashboard has surfaced a problem and stopped there, because the work of figuring out what to do about it was never built into the process.

A useful discipline: for any chart on a management dashboard, the team should be able to say in advance what an unfavorable reading means and what the first action will be. Not a generic action. A specific one, owned by a specific person, on a defined timeline. Without that, the chart is an alarm with nobody assigned to answer it.

2. Charts that show problems nobody owns

The second category is structurally different. The chart works, the action would be obvious, and still nothing happens, because no single person is responsible for acting on what it shows.

This is the orphaned-signal problem, and it is common in matrix-organized mid-sized companies. The sales director sees the regional drop and assumes the regional manager will act. The regional manager sees it and assumes the sales director will raise it. The CFO sees it in the board pack and assumes someone closer to the business is already on it.

The shift that separates BI organizations producing decisions from those producing reports is straightforward: instead of asking “what reports do we need,” teams ask “what decisions are we enabling.”² The second question forces the ownership question. The first lets it stay invisible.

BI investments routinely produce dashboards full of orphaned signals. The fix is not technical. Adding a single column to the metric inventory does most of the work: for each KPI on the dashboard, the name of the person expected to act when the number indicates action is needed.

3. Charts that exist because nobody removed them

The quietest category. Many of the metrics on a mid-sized company’s management dashboard were chosen years ago, for reasons nobody remembers, in a business that has since changed shape. They continue to be reported, monthly, in the board pack, where they get reviewed out of habit and produce no decision.

Removing them feels risky. Adding new ones feels productive. Over time, the dashboard accumulates charts the way a desk accumulates papers, and the volume of information rises without the volume of decisions rising with it. Practitioner audits of failed BI implementations have repeatedly found that the dashboards that actually get used are not the most comprehensive ones, but the most heavily edited.³

The test surfaces this directly. A chart that cannot point to a decision it has produced in the last twelve months is not informing the business, even if everyone in the room agrees the number is interesting.

What the test is for

The test is not a tool for criticizing the BI team. It is something the management team should be running on itself. The question is not whether the dashboard is good, but whether the company is actually using it to run the business.

The answer determines whether the next BI investment is worth making. A company that cannot name decisions produced by its current dashboard does not need a bigger one. What it needs is the slower work of designing each chart backward from the decision it is supposed to inform, with named owners and agreed actions, before any new tool is bought. A dashboard that produces no decisions is not a partial success that better technology will complete. It is the central problem the technology was supposed to solve.