Somewhere right now a capable team is spending forty minutes deciding whether a risk scores a six or a seven. They will not resolve it, because there is nothing to resolve. The distinction does not exist outside the worksheet, and while they argue it, the hazard they agree is serious sits unaddressed. This is scoring theatre, and it is one of the most reliable ways to waste a good analysis.
The classic scoring method multiplies three ratings, severity by occurrence by detectability, each on a scale of one to ten, into a single number between one and a thousand. It has real virtues. It is granular, it is familiar, and it is what most of the published literature uses. It also has vices that are better documented than most teams realise. Multiplying two subjective judgements does not produce an objective one; it launders opinion into false precision, so that a score of two hundred and eighty-eight looks meaningfully different from two hundred and eighty when no clinical team alive could tell them apart. Worse, wildly different risks collide on the same number. A rare, poorly caught catastrophe and a daily, poorly caught nuisance can produce an identical score, and the ranking will not know the difference.
The deepest problem is quieter. Those one-to-ten scales are ordinal. The gap between an eight and a nine is not the same size as the gap between a two and a three, and multiplication assumes it is. The arithmetic rests on a fiction, harmless if you hold the numbers lightly and dangerous the moment you let a comfortable product bury a severe hazard beneath two small factors.
So hold the numbers lightly, and protect one rule above all the others: severity trumps arithmetic. Any failure mode that could kill or permanently harm a patient goes forward for action regardless of how the product works out, because no multiplication is allowed to demote a catastrophe on the strength of it being rare and, on paper, detectable. The rarest failure in the room can be the most important one in the building.
The healthcare adaptation many teams now use replaces the ten-point scales with four-point ones, anchored to concrete definitions, and adds a short decision tree after scoring: is this a single point of failure, does an effective control genuinely already exist, would the failure be obvious enough to be caught anyway. The anchors do the work the ten unanchored points never did, and the tree quietly dissolves most of the six-versus-seven fights, because whether a hazard scores an eight or a nine rarely changes which branch it takes.
Which system you use matters less than how you hold it. The scores are a device for forcing a shared conversation and a defensible ranking. They are structured judgement, not measurement, and a team that treats the number as physics will end up polishing decimals while the process stays exactly as dangerous as it was.
This article accompanies the book Before Harm Happens.