Every department reports a different revenue figure: why, and what to do

Sales reports 8% growth. Finance sees a 2% decline. Both have a dashboard, both have an export, and the first twenty minutes of the monthly meeting once again go to the question of whose number is right — instead of what should happen next.

If this sounds familiar: it's almost never a calculation error. Everyone is calculating correctly. That's exactly the problem.

The problem is not that someone is calculating wrong. The problem is that everyone is calculating right — with a different definition.

Where the gap really comes from

In the engagements I do, the discrepancy almost always comes from the same five corners. Walk through them and you'll probably recognise your own organisation in three or four.

1. Different definitions. "Revenue" sounds like a word with one meaning, until you push. Including or excluding VAT? Booked at order, at invoice or at payment? Do credit notes count, and in which month? Sales counts at signature, finance at invoice date — and then a structural gap isn't an anomaly but a mathematical certainty. The same goes for "customer" (active? unique per branch?), "employee" (FTE or headcount?) and virtually every other core figure.

2. Different moments. Finance looks at a closed month; sales looks live in the CRM. A report that runs on Monday and an export from Wednesday give different answers to the same question — and nobody remembers which number came from which moment.

3. Different sources. The CRM, the accounting system and the POS each hold their own truth, and they diverge: an order that exists in the CRM but hasn't been invoiced yet, a correction applied only in the books. Every department faithfully reports from the system it works in.

4. Manual steps in between. Somewhere between source and report there is almost always an Excel: an export with a filter ("just removing the internal orders"), a formula someone once adjusted, a tab that has been copied forward for years. Each of those steps is a silent change of definition — invisible and undocumented.

5. History shifts. A customer moves region, a team is split, a product group is reclassified. One report applies the new structure retroactively, the other doesn't — and "compared to last year" means something different in every report.

Why another meeting won't fix it

The reflex is understandable: get everyone in a room and agree which figure is leading. That works for exactly one quarter. Then there's a new colleague, an adjusted formula or a new source, and the gap is back.

The cause is structural: the definitions live in heads and in Excel formulas, not in a place where they are enforced. As long as every department has its own route from source to report, every route produces its own truth — no matter how good the agreements are.

What you can do yourself this month

Without hiring anyone, and with immediate payoff:

• Pick the five figures the argument is actually about and write one definition for each, including the edge cases (VAT, credit notes, snapshot moment, internal orders). One page, visible to everyone. That single document takes the sting out of half the discussions.
• Appoint one owner per figure. Not a department — a person. When things change, the owner decides, and the definition gets updated instead of worked around.
• Trace one figure from source to report, once. Put the route on paper: which system, which export, which Excel, which filters. Almost always you find a step along the way that nobody knew existed. That walk costs an afternoon and is the best diagnosis there is.
• Cut parallel exports where you can. Every copy of the data is a future discrepancy. One report as the agreed source per topic — even if that report isn't perfect yet.

The structural fix

The agreements above make things better; they don't make them hold. It only holds once the definition isn't written in a document but enforced in the technology: one place where the data from all sources comes together, where exactly one calculation exists per figure, and where every change is tested before it reaches a report.

That is what a data warehouse fundamentally is — not a bigger dashboard, but the end of parallel routes. The argument doesn't stop because people honour agreements better; it stops because only one number exists.

This is how I build that as a package: Your data warehouse — three to four weeks, fixed approach, with the definitions as week one. What it costs and where that money actually goes, I've written down without detours in What does a data warehouse cost for an SME? And if you'd rather first know where your own landscape stands, start smaller: the Data scan maps sources, routes and definition gaps in two weeks.

Curious where the routes diverge in your organisation? Book an intro call — the lineage walk above is something I'll gladly do with you during a first conversation, free.