One Sentence Per Number

The meeting has an agenda and the agenda doesn't survive the first slide.

Somebody says the number looks wrong. Somebody else says it depends what you're counting. Then twenty minutes go on reconciling two figures that are both, in their own terms, correct.

It resolves eventually — usually by somebody deferring rather than by anyone establishing what the number should be. And it recurs next month, in the same shape, occasionally with the same two people.

The most reliable tell that this is what's happening: nobody ever writes down the resolution. The meeting reaches agreement and produces no artifact, which guarantees the next one starts from the same place.

One word doing two jobs

Take revenue. To whoever runs sales, it's what was signed this month. To whoever handles the money, it's what was collected. To you, it might be what was delivered.

Three people, three numbers, one word — and every one of those definitions is standard somewhere.

The same fracture runs through every operational term you use.

An active client. Currently in a project, or currently paying, or has paid within twelve months?

A lead. Anyone who filled the form, anyone who replied, or anyone who qualified?

A completed project. Delivered, accepted, or invoiced?

Each has two or three defensible readings, and each of your people has silently picked one.

The argument isn't about the data. It's about the language — which is why consolidating your tools doesn't fix it.

Why it's so durable

Because nobody experiences themselves as having chosen.

Each person's definition feels like the plain meaning of the word. So the disagreement doesn't present as a definitional problem — it presents as somebody's data being wrong. That's why these arguments have a faintly adversarial edge, and why they never resolve: both parties are defending accuracy, and both are accurate.

It's also why the instinctive fixes don't work.

A new dashboard produces one place where two people still mean different things by the same word, and now they argue in front of a single screen. A data warehouse standardizes the pipes and leaves the semantics untouched. Even hiring an analyst mostly makes somebody a referee — the disputes resume the moment they stop refereeing.

The test that ends it

Ten minutes, at the next meeting, the moment the argument starts.

Ask both people to write down — independently, before either speaks — how the disputed metric is calculated. Three things: what's included, what's excluded, and at what moment it counts.

The two sentences will differ. Both people will be astonished, because each assumed their definition was the definition.

That third element is where most of the disagreement actually lives, and it's the one everyone forgets to specify. At what moment does this count is the difference between signed and collected, between delivered and accepted, between a lead and a qualified one.

The rule

Every recurring metric needs one written definition, owned by one named person, stored where the number is displayed.

Not a data governance process. A sentence next to the figure.

One sentence. Included, excluded, when it counts. If it takes a paragraph, the metric is probably two metrics.

One owner. Not a committee. One person who decides what the word means and is the escalation point when a genuine ambiguity appears. The point isn't authority for its own sake — it's that a definitional argument becomes a two-minute question rather than a twenty-minute debate.

Stored at the point of display. This is the part that gets skipped and it's the part that makes it work. A definitions page nobody opens is functionally identical to no definitions. The sentence has to be visible at the moment of dispute, which means next to the figure, in the dashboard, in the report.

Two things that make this fail

Optimizing instead of deciding. Founders derail this by trying to settle on the best definition of revenue, which turns a ten-minute clarification into a philosophical project. Almost any consistent definition beats two inconsistent good ones. Pick, write, move on — and change it later if it turns out to be wrong, which is cheap once it's written down.

Trying to collapse a genuine distinction. Sometimes two definitions are both needed, because signed and collected are genuinely different real things. Don't resolve it — name it. Two metrics, two words, both defined, both displayed. Most of these arguments are one word doing two jobs, and the fix is a second word rather than a winner.

The bill for twenty minutes a month

The meeting time is the smallest part. Twenty minutes a month is annoying; the real damage is that the reconciliation consumes the slot where the decision was supposed to happen.

The numbers stop being used. People who've watched three arguments about a figure learn to distrust it, and distrusted numbers get quietly worked around — a private spreadsheet, a personal count, a gut read. You end up with a reporting system nobody disputes because nobody consults it.

Your team gets disqualified from decisions. If a number can be contested at any moment, nobody below the founder can act on it with confidence. So every decision touching that metric routes upward for adjudication, and you become the definition of record — which is a genuinely terrible thing to be.

It eventually costs you externally. A business that can't state its own numbers consistently will state them inconsistently to someone who matters — a lender, a buyer, a partner, a client. The damage there is to credibility rather than to a meeting.

The dozen words worth doing first

You don't need a glossary. A dozen terms cover almost every argument a founder-led business has, and they're the same dozen everywhere.

Revenue. Active client. Lead. Qualified. Closed. Delivered. Utilization. Margin. Churn. Pipeline value. Capacity. Complete.

Take that list, cross off the ones nobody has ever disagreed about in your business, and write the sentence for what's left. Most founders end up with five or six, and they can do the whole thing in an hour.

Three of them are worth flagging as unusually prone to trouble.

Pipeline value is almost always wrong somewhere, because it depends on whether unqualified opportunities count, whether verbal agreements count, and whether the number is weighted. Three people will have three answers and all three will be defensible.

Utilization is the most reliably contested number in a service business — billable hours over available hours, except that both halves of that fraction have four plausible definitions each.

Complete breaks quietly rather than loudly. Delivery says the work is done, finance says it isn't invoiced, the client says they haven't accepted it, and everybody is right. This one usually needs two words rather than one definition.

When two people are both right and both stuck

There's a failure mode worth naming because it looks like stubbornness and isn't.

Sometimes a definitional argument persists after everyone agrees on the definitions, because the two people need genuinely different numbers for genuinely different jobs. Sales needs signed revenue to manage a team. Finance needs collected revenue to manage cash. Neither is being difficult.

The tell is that the argument recurs after being resolved. A one-off disagreement is a definition problem; a recurring one after agreement is a missing metric problem, and adding the second number ends it permanently in a way that any amount of alignment work won't.

Why it's a Telemetry problem, not an admin one

Transparency is a leadership strategy — but transparency requires that the thing being shared means the same thing to everyone reading it. Shared numbers with private definitions produce the appearance of transparency and the experience of confusion, which is worse than not sharing at all.

A trend line assembled from two definitions isn't a trend. It's an artifact.

And that goes directly at what the Telemetry pillar is for. The point is to stop reacting and start recognizing — and recognition is pattern matching over time, which is impossible if the definition of the thing being tracked drifts between whoever happened to report it.

Where the sentence should live

One placement detail decides whether any of this survives, and it's the thing most likely to be skipped because it feels cosmetic.

The definition has to be visible at the moment of dispute. Not in a glossary, not in a shared drive, not in the appendix of a reporting document — next to the number, wherever the number is displayed.

A definitions page nobody opens is functionally identical to no definitions. The whole failure mode being fixed here is that people don't realize they disagree, which means the correction has to arrive unprompted, in the place where the disagreement would otherwise start.

In practice: a note field on the dashboard tile, a footnote under the figure in the weekly pack, a comment on the spreadsheet cell. The medium doesn't matter and the proximity does.

There's a second-order benefit that shows up within a couple of months. Once definitions sit next to numbers, people start noticing when a definition is wrong — because they read it in the context of a specific figure that looks odd, which is exactly when a definitional error is most detectable. Definitions in a document get written once and never revisited. Definitions next to numbers get corrected.

Ten minutes at the next meeting

You don't need a project for this. Next time the argument starts, stop and have both people write their definition down.

That ends that specific dispute permanently. Do it three times and you'll have most of a definitions list — and you'll have it in priority order, because the metrics that get argued about are the ones that matter.

Then put each sentence next to its number, and give it a name.