The Decimal Point Is the Tell

A strategic-foresight brief crossed my desk recently that did something I have learned to watch for. It took a quality that no one can measure — call it organisational drift, the slow divergence between what a system is asked to do and what it can actually absorb — and gave it a number. Not a range. Not a direction. A rate, expressed to two decimal places, a fraction of a percent per day, compounding tidily to a double-digit annual figure that landed on the slide with the quiet authority of a fact.

The underlying observation was sound, and I want to be fair about that: the gap between what institutions are told to do and what they can carry is real, and most commentary misses it. But the number attached to it was not a measurement. It was a decision dressed as one. And the ability to tell those two things apart — to test a figure before you build a plan on it — is one of the most valuable and least practised skills in this profession. So let me teach it, the same way I teach everything: by trying to break the thing.

Why the number is seductive, and why that is the danger

Precision reads as credibility. "Things are diverging" is an opinion anyone can dismiss. "Divergence is running at 0.06% a day" sounds like someone has been measuring — and once a figure has a decimal point, it stops being argued with and starts being cited. It migrates into the risk register, the board deck, the funding case. Nobody re-derives it. The decimal point did its work: it converted a judgement into an input.

That is precisely why it deserves the most scrutiny, not the least. A vague claim advertises its own uncertainty; you know to weigh it. A precise one hides its uncertainty behind the digits. The more exact a number about an inherently fuzzy thing, the harder I look — because exactness on something unmeasurable is not a sign of rigour. It is usually a sign that rigour has been skipped and dressed up.

The three questions that break a number

This is the same discipline as the Alarm Test — you state the claim at its strongest and then genuinely try to falsify it — narrowed to a figure. I ask three questions, in order, and a manufactured metric rarely survives the first two.

What is being measured? A number needs a unit, and a unit needs a definition. "Divergence per day" invites the obvious question: divergence of what, from what, on what scale? If the thing has no agreed unit — if you could not, even in principle, describe the instrument that would read it — then the figure is not quantifying anything. It is assigning a mood a coordinate.

Where did the input come from? Trace the number to its source and you will find one of two things. Either a measurement — a dataset, a survey, a defined proxy someone can inspect and argue with — or an assertion, a starting value chosen because it felt right and then run through arithmetic that is impeccable and irrelevant. Compounding a made-up daily rate into an annual one is real mathematics performed on an imaginary input. The tidiness of the output tells you nothing about the honesty of the input.