The Monday meeting number fight
We once walked into a company running 14 dashboards that all disagreed on revenue. Fourteen. The CFO had one number, the VP of Sales had a bigger one, and the product team had a third that matched neither. Every Monday meeting opened with ten minutes of arguing about whose number was right before anyone could discuss what to do about it.
Nobody in that room was lying. Every dashboard was correctly reporting the metric it was built to report. The numbers disagreed because the metrics were never the same thing to begin with.
Three departments, three honest numbers
Finance counts recognized revenue. A 12-month contract signed in January shows up as one twelfth per month, because that's what accounting standards require.
Sales counts bookings. That same contract shows up in January, all of it, because that's when the deal closed and that's what commission is paid on.
Product counts subscription events. The contract shows up when the account activates, which might be March if onboarding drags, and it might be missing entirely if the data pipeline only tracks self-serve signups.
Same contract, three numbers, three dashboards. Multiply by every discount rule, refund policy, and currency conversion nobody documented, and you get 14 dashboards that will never agree. Not because the tools are broken. Because each one answers a slightly different question, and the questions were never written down.
Why a new BI tool won't save you
The instinct at this point is to buy something. A new BI platform, a "single source of truth" project, maybe a semantic layer. We've watched companies spend six figures on replatforming only to rebuild the same disagreements in nicer charts, because migrating a dashboard migrates its definition, and the definitions were the problem.
The fix is organizational before it is technical. One person, usually whoever owns data, sometimes the CFO, has to be able to say: this is the company's revenue number, this is how it's calculated, and every other revenue-shaped number is either derived from it or clearly labeled as something else.
The fix, in the order that works
First, pick the number that goes to the board. Not the best number or the most sophisticated one. The one your leadership already treats as real. That's your anchor, and it usually belongs to finance.
Second, trace it backwards. Which tables, which filters, which currency logic, which refund handling. Write it down where the whole company can see it. This takes days, not months, and it's the highest-value documentation your data team will ever produce.
Third, deal with the other 13 dashboards. Some get deleted, and deleting a dashboard nobody trusts costs you nothing. Some get renamed to what they actually measure: "bookings" is a fine metric with a fine name, it just isn't revenue. The few that survive get rebuilt on top of the anchor definition so they can't drift again.
Fourth, put a name next to the definition. Metrics drift when they're everyone's job. A definition with an owner gets defended; a definition without one gets quietly forked the next time a team wants their number to look better.
How you know it worked
The Monday meeting stops opening with an argument. That's the whole test. When people disagree about what to do, that's strategy and it's healthy. When they disagree about what happened, that's a data problem, and it's one of the cheapest ones to fix relative to what it costs to leave alone.
If this sounds familiar and you want a second pair of eyes on your revenue definitions, that's a conversation we have often. It's usually the first thing we untangle in a data strategy engagement, and it's diagnosable in two or three working sessions.