The meeting that starts with reconciliation
A leadership team was assembling board reporting by hand from three tools. The visible symptom was effort — days of it, every month. The real cost was that every meeting opened by arguing about which number was correct instead of what to do about it.
When two people can each produce a defensible pipeline figure and the figures differ, the disagreement is not about data quality. It is that "pipeline" was never defined precisely enough for two systems to compute it the same way.
Designing the argument out
The work was less about building a dashboard than about writing down definitions and getting them signed off: what counts as qualified, when a stage advances, which date governs a forecast, how a deal in one currency appears in a consolidated view.
Each of those had been decided implicitly and differently in three places. Making them explicit was the entire intervention; the dashboard afterwards was comparatively simple engineering.
This is why buying a reporting tool rarely fixes reporting. The tool computes faithfully whatever definitions it is given, and if those were never agreed it produces a fourth authoritative-looking number.
The hard part
Someone had to be able to decide. Several definitions had genuine trade-offs — a stricter "qualified" makes the pipeline smaller and the forecast more accurate, and different people in the business are measured on those two things.
Those trade-offs are business decisions, not technical ones, and a project without a named decision-maker stalls at exactly this point. Naming that person early is the single highest-leverage move in this kind of work.
What transfers to other leadership teams
- If two people can produce different pipeline numbers, you have a definitions problem, not a tooling problem.
- Write definitions down and have one named person approve them before building anything.
- Board-ready reporting on demand is the test — if it still takes a week of preparation, the definitions did not hold.
- Forecast conversations should be about deals. Any time spent on whether the data is right is time the system should have saved.