An executive meeting can spend twenty minutes debating whether revenue is up before anyone asks what to do about it. Sales has one figure, finance has another, and the dashboard has a third. Each may be correct within its own rules. The problem is that the organization has never agreed which rule governs the decision in front of it.
I would resist the familiar response: commissioning one more dashboard. A new display may make the disagreement more attractive, but it will not settle it.
Numbers carry decisions inside them
Take a deceptively simple measure such as “active customer.” Does it mean someone who purchased this month, holds an open account, used a service recently, or is eligible to be contacted? Each definition supports a different decision. A marketing team may need a recent-engagement measure; finance may need a recognized-revenue view. Forcing one definition on every context is as unhelpful as allowing every department to invent its own.
The leadership task is to name the decision first, then agree on the measure that serves it. Who will use the number? What action might change if it moves? Which exclusions are legitimate? What period and source of truth are appropriate? If those questions are unanswered, a precise-looking decimal can create false confidence.
This is why I think metric ownership belongs in the operating rhythm of a business, not in an appendix to a data platform project. An owner needs the authority to approve a definition, hear objections and decide when a change is warranted. Our data engineering teams can make the measure consistent and traceable; they cannot decide on behalf of a client's commercial and operational leaders what the business means by success.
Make disagreement visible before it becomes political
When two teams bring different numbers to a review, the instinct is often to choose the larger one, or the one in the newest system. I would do the opposite. Put both calculations side by side. Show their sources, filters, update times and intended uses. Then ask where the difference originates.
Sometimes the gap is a quality defect: missing orders, duplicated accounts, late records. Sometimes it is a legitimate distinction: booked revenue versus collected cash, or placed orders versus fulfilled orders. These situations need different responses. A defect needs correction and monitoring. A legitimate distinction needs clear labels and a rule for which number belongs in which meeting.
There should also be a route for escalation. If an agreed definition no longer reflects how the business works, a frontline analyst should not quietly change a report and hope nobody notices. Equally, a governance committee should not make every useful change wait for a quarterly meeting. Give the metric owner a way to approve small clarifications promptly, with a record of what changed and why.
Build a small number of trusted measures first
I would rather have ten measures that leaders can challenge and act on than a hundred that merely fill a screen. Start with decisions that recur: where to allocate capacity, which customer segment needs attention, whether a process is improving, and where risk is rising. For each, record an owner, a definition, a source, a refresh expectation and a known limitation.
Then test the measure in a real meeting. When it changes unexpectedly, can the team explain the movement without exporting several spreadsheets? Can a manager trace it back to the relevant process? Does the discussion end in an action with a named owner? These are better tests of a data programme than counting dashboards delivered.
The technology matters. Reusable data models, lineage and quality checks reduce the effort required to maintain agreed measures. But the platform should serve the agreement; it cannot substitute for one. If a metric is contested every month, the architecture may be excellent while the decision system is still broken.
What I'd do differently
At the start of a transformation, I would ask the client's executive team to bring one number they rely on and one decision it informs. We would compare definitions before discussing tools. Where the numbers conflict, I would ask the business to assign an owner and agree a short deadline to settle the meaning, with the data team documenting the calculation and exceptions.
Only then would I ask for a dashboard. The purpose of a trusted KPI is not to end debate. It is to move the debate from “whose number is right?” to “what should we do now?”
Put this insight to work.
If you are weighing a similar decision, our team can help you clarify the problem, test your options and choose a practical next step.