One number, twenty two definitions

If you want to understand how an organization really works, do not read the strategy deck. Ask twenty people to define the same metric.

I have done a version of this exercise more than once, and it goes the same way every time. Everyone is confident. Everyone is slightly different. And when you line the definitions up next to each other, you realise the differences are not sloppiness. Each team had a reason.

One team counted a deal from the day it was qualified, because that is when their work started. Another counted from first contact, because they owned that part. One excluded renewals, because renewals were not their target. Another included them, because their number would have looked absurd without them. Nobody was being careless. Each definition was a reasonable answer to a different question.

Then someone asks for one number across all of them, and you discover you do not have one.

Why this is not a data problem

The instinct is to treat this as cleanup. Pick the correct definition, write it down, enforce it, move on. I have watched that approach fail, and I have been the person applying it.

It fails because you are not standardizing a column. You are overriding how a group of people understand their own work. A manager whose number drops because you changed the start date does not experience that as improved rigour. They experience it as their performance getting worse for reasons they did not cause and cannot explain to their team.

So they do the rational thing. They keep their own spreadsheet. Now you have the standard number, which is used for reporting, and the real number, which is used for managing. Two sources of truth is strictly worse than the twenty two you started with, because at least those were honest about being local.

What actually worked

When I built a common measurement layer across twenty four manager pods, the thing that made the difference was accepting that one definition was not the goal.

Separate the spine from the slack. There was a small set of things that genuinely had to be identical everywhere, because leadership was going to add them up. Everything else was allowed to vary by pod, as long as the variation was declared. The spine was deliberately thin. Most of the arguments I expected never happened, because most of the contested measures turned out not to need to be comparable at all.

Write down what each team currently means, before proposing anything. This took longer than building the thing and was the highest leverage week of the project. Half the apparent disagreements dissolved once both teams could see the other's definition in plain language. They had been arguing about the number when they were actually using different words.

Show the definition next to the number. Not in a data dictionary that nobody opens. Right there, where someone is looking at the figure that affects them. The test is whether a manager who thinks the number is wrong can find out why in under a minute without asking anyone. If they cannot, they will assume it is wrong and stop using it.

Let the first version be slightly worse than the spreadsheets. This was uncomfortable. The spreadsheets were tuned over years and the new layer was not going to beat them on day one. What it could beat them on was being the same everywhere, being explainable, and not needing someone to maintain it by hand. I said that out loud rather than pretending the numbers were better. Saying it out loud is, I think, why people stayed.

The part nobody tells you

A standard definition is a political object. It decides who looks like they are doing well. That is not a cynical observation, it is just true, and it means the work is not finished when the query is correct.

The people whose numbers get worse under the new definition need to hear it from you before they see it in a report. Not as a negotiation over the definition, which you should not reopen once it is agreed, but as a straight explanation of what changed and why. I have skipped that step before. It cost more time to repair than it would have taken to do.

Fifteen measures, one agreed spine, declared local variation, every number traceable in a click. That is the shape that held. Not because it was elegant, but because the people being measured by it could see themselves in it and could check it when they doubted it.

Which is really the only test that matters. A definition is not standard because you published it. It is standard when the spreadsheets stop.