Apply the "so what?" test to every supporting point

Each supporting point should directly and uniquely answer the question above it in the pyramid.

Why it works

The "so what?" test checks vertical integrity of the pyramid: the parent claim must follow from the children, and each child must be necessary for the parent to hold. Points that survive "so what?" are genuinely load-bearing. Points that don’t ("this is interesting context") are decoration that imposes reading cost without argumentative contribution. Removing them tightens the structure.

How to do it

1. Write your main recommendation, then list each supporting point below it.
2. For each supporting point, ask: "If this were false, would it weaken the recommendation?"
3. If the answer is no, the point is decorative — cut it or find the level where it is genuinely relevant.
4. If removing a point would hurt the argument, ensure it is stated as a direct reason, not as context.

Evidence

The "so what?" test is a practitioner editing heuristic, not a studied technique. Its effectiveness derives from the logical property it checks: whether a proposition is necessary for the conclusion it is claimed to support. (anecdotal)

The test is harder to apply to inductive groupings (where supporting points together suggest a conclusion) than to deductive ones (where the conclusion follows necessarily). Most real arguments are inductive; apply the test with that in mind.

Common mistake

Keeping a point because it is true and interesting rather than because it is load-bearing — truth and relevance are not the same thing in a structured argument.