Being terrible at something worth 2.6% of the exam is cheaper than being mediocre at something worth 33%.
Ask a candidate what they are going to study next and they will name the topic that worries them most. It is the natural answer and it is usually the wrong one, because "what am I worst at" and "what should I study next" are different questions with different answers whenever an exam is weighted.
What an objective is worth to you is not your score on it. It is how much the exam cares about it, multiplied by how far below standard you currently sit:
points on the table = blueprint share × (target − your mastery)
Both terms matter and people only ever look at the second. Consider two domains on a 53-question exam:
| Domain | Share | Your mastery | On the table |
|---|---|---|---|
| A — the light one, and you are terrible at it | 2.6% | 0.20 | 15 pts |
| B — the heavy one, and you are merely so-so | 33.1% | 0.65 | 60 pts |
Domain A is the one that feels urgent. You are scoring 20% on it; it is embarrassing; it nags. Fixing it completely is worth about fifteen scaled points, because the exam will ask you between one and two questions about it.
Domain B does not nag at all. You are getting two-thirds of it right and it feels handled. Closing that smaller gap is worth four times as much, because it is a third of the paper.
A week spent on A instead of B is not a small misallocation. It is most of your remaining study time spent on 2.6% of the exam.
This only bites when the weights are lopsided, and on these exams they are. CCDV-F publishes domain shares from 33.1% down to 2.6% across 8 domains — a factor of more than ten between the heaviest and the lightest:
The two rows in the table above are those extremes: the lightest domain and the heaviest, which is where the fourfold difference in what they are worth comes from.
CCAR-F is the interesting exception: it weights every objective equally, so the ordering there really is just your own weakness, and the trap inverts. There is no light objective to skip, and the one you find dullest costs exactly as much as the one you enjoy.
It does not mean abandoning the light objectives. An objective worth 2.6% is still worth more than the zero you get by never sampling it again, and an estimate built on objectives you stopped measuring months ago slowly stops being an estimate. Something has to keep the mastered and the trivial ones ticking over, or the number goes stale without ever looking wrong.
And it does not mean chasing perfect scores on the heavy ones. The target is clearing the cut, not perfect recall of a question bank — past a certain point an objective stops being where your marks are, and the arithmetic above says so on its own.
The ordering that falls out of this is not one you can produce by reading the exam guide, because half of it is a fact about you. It needs your measured mastery per objective, which is the thing a lifetime accuracy percentage cannot give you.
ExamGauge is built on the argument above: 1523 original practice items across four Claude certification exams, scored on the real 100–1000 scale against the 720 cut. The diagnostic is free and needs no card.