When a leadership team does not trust its numbers, the request that lands on my desk is almost always the same. Build a better dashboard. Cleaner design, more drill-downs, a single screen everyone can look at. The assumption is that the reporting is the problem.
It usually is not.
A dashboard is a display. It shows whatever the underlying model produces. If three teams calculate the same metric three ways, a dashboard does not resolve the disagreement. It renders it in higher resolution. Now everyone can see, precisely, that the numbers do not agree. The meeting that follows is not about the data. It is about who is right, and that is a question a dashboard cannot answer.
The real problem is ownership. Not ownership of the dashboard. Ownership of the metric. Someone has to own the definition: what counts, what is excluded, over what period, at what grain. Someone has to own the data that feeds it. And someone has to own the decision the metric is supposed to drive. When those three owners are clear, the reporting tool almost stops mattering. When they are not, no tool will save you.
I see this most clearly in functions that have grown by accretion. The metric existed before anyone wrote down what it meant. Over time, different teams needed a version of it, so they each built one. Each version was reasonable in its local context. None of them was wrong, exactly. But there was no single owner with the authority to say which definition the organization would use, so all of them survived. The dashboard request is the symptom. The missing owner is the disease.
This is why I diagnose before I prescribe. Before touching a reporting layer, I want to know one thing: who decides what this number means? If the answer is “it depends who you ask,” the work is not a dashboard. The work is assigning ownership and writing the definition down somewhere that has authority. That is uncomfortable, because it forces a decision the organization has been avoiding, often for years. The ambiguity was load-bearing. Someone benefited from the number staying flexible.
There is a tell. When you propose a single owned definition, the people who object most are usually the ones whose local version made their results look best. That is not a data conversation. It is an accountability conversation wearing a data costume. A dashboard lets everyone keep their own version. An owned definition does not. That is exactly why the owned definition is the harder sell, and exactly why it is the one that fixes the problem.
None of this means dashboards are useless. A good dashboard built on an owned model is genuinely valuable. The sequence is what matters. Ownership first, definition second, model third, and only then the display. Teams reverse this constantly. They buy the tool, build the screens, and discover six months later that adoption is low because no one trusts the numbers, because no one owns them. The tool was never the constraint.
So when someone asks me for a better dashboard, my first question is not about the tool. It is: when this number is wrong, whose job is it to fix it? If there is a clear answer, we can talk about dashboards. If there is not, we have found the actual project. It is less glamorous than a redesign and worth far more than one.
Reporting problems are operating-model problems in disguise. Treat the disguise and you get a prettier version of the same distrust. Treat the operating model, assign the owner, write the definition, and the reporting problem quietly goes away. The dashboard was never going to fix ownership. Ownership fixes the dashboard.