Three people, three answers, one undefined word
In many companies the same question produces different answers. How to settle your definitions before you start reporting.
Ask three departments how many active customers you have and you get three numbers. That is not a data quality problem. That is a language problem.
Three people use the same word for three different things, and as long as that is true, every report you build is a subject of debate instead of a basis for a decision.
An example that actually matters
At one client, "active customer" was defined as someone who had ordered in the past twelve months. That sounds watertight. It is written down somewhere, anyone can recalculate it, there is no room for discussion.
And still it was not enough, because frequency says much more than the fact that something was ordered at some point.
Someone who buys every month and someone who buys once a year both land in the active category. While those are two completely different customers, with a different value, a different risk of leaving and a need for a different approach.
By putting them in one bucket, you lose exactly the difference you want to steer on. Your definition is not wrong, it is too coarse. And that is the kind of mistake you cannot see in the number itself.
The four questions every definition has to answer
Whatever you are defining, these are the questions it usually gets stuck on.
Which period. Twelve months rolling, or calendar year? In January that gives a completely different number.
Which event counts. An order, a delivery, an invoice, a payment? With a customer who orders in December and pays in January, you are looking at four different answers.
What does not count. Credit notes, internal sales, samples, trial orders, cancellations. Any of those can shift your total.
At what level. Per location or consolidated? Per legal entity or per brand? A customer who buys at two locations, is that one customer or two?
When those four are settled, you have a definition. If not, you have a word.
Where you start
Not with everything. If you try to define your whole company before you build anything, you will never get past the meeting room.
Start with the numbers decisions are made on. In practice there are five to ten of them. Revenue. Margin. Active customer. Outstanding balance. Inventory value. Absence. For your business maybe a few others, but it is always a short list.
Put the four questions above next to each one and let the people who work with that number answer them together. Not the IT side, not the consultant. The people who use it.
Where you write it down
In a document that is accessible and that carries a date. It does not have to be more than that.
Per number: the name, the definition in plain language, which source delivers it, who owns it, and when it was last changed. That last one matters more than it seems, because if a definition changes, you need to know from which point on your series is no longer comparable.
Attach that document to your dashboard, literally. A link in a corner is enough. Then someone who does not trust a number can go and look for themselves instead of raising it in a meeting.
Why this saves so much time
The honest answer is that this conversation usually takes an afternoon and that it gets postponed because it is boring.
But the bill arrives either way. Either you pay that afternoon up front, or you pay it in the form of a year of reports nobody dares to steer on because there is always someone who disputes the number.
I have seen projects that were technically perfect and died anyway, purely because it was never decided what revenue means. The dashboard ran, the numbers matched the source, and nothing was done with it.
The sign that it is not settled
You recognize it by one sentence. When someone in a meeting says "yes, but that number does not take into account", that is not a remark about the report. That is proof that there are two definitions in the room.
Write that sentence down when it falls. Usually that is the fastest way to build your list of terms that need defining.
The claim
A report that gets disputed is not a bad report. It is a report about a word that was never defined, and that is a conversation you have to have, not your supplier.