Assumption Audit Worksheet: Finding the Load-Bearing Premise

How to use this

Bring a real conclusion your team or organization has reached recently. Working from a real case is important, because the difficulty of this worksheet is entirely in applying it to things you already believe.

Allow half an hour. Twenty minutes is enough for steps one to four; step five is where the useful conversation happens.

The five steps

Step 1: State the conclusion in one sentence

Precisely enough that it could be wrong. “The campaign worked” is not yet a conclusion. “The March email campaign caused the increase in first-time donations” is.

Step 2: List the evidence, separately

One item per line, observations only. No interpretation. Include items that seem unimportant, because those are frequently the ones that discriminate.

Step 3: Write every assumption you can find

Aim for at least five. Include the boring ones: that the data is accurate, that the categories mean what you think, that the comparison periods are comparable.

Step 4: Find the load-bearing one

For each assumption, ask: if this were false, would the conclusion fail, or would it merely be less precise?

Assumptions that only affect precision are not load bearing. Usually one or two of your list will fail the conclusion outright. Those are the ones the argument actually rests on.

Two patterns account for most cases. The first is “nothing else relevant changed”, which almost every before-and-after comparison depends on and almost nobody states. The second is a population shift, where evidence about one group is used to support a claim about a wider one.

Step 5: Name the test

For the load-bearing assumption, write down one observation that would tell you whether it holds, and whether that observation is obtainable this week.

If it is obtainable, the output of this worksheet is a task rather than a debate. That is the best possible result.

Common failure modes

Listing assumptions nobody disputes. Ten obvious assumptions can crowd out the one that matters. Step four exists to force the sort.

Stopping at step four. Identifying the assumption without naming the test leaves the group with a shared doubt and no way to resolve it, which is worse than where they started.

Auditing only conclusions you dislike. Run this on a conclusion the group is pleased with at least once.

For facilitators

Do this with a real internal decision, not a constructed case, and be prepared for it to be uncomfortable. Set the expectation at the start that the output is a list of things to check, not a verdict on anyone’s judgment.

Version and reuse

Version 1.0, reviewed August 2026. Reuse terms are described on the Terms of Use page.

Next step

Practice in the Test Your Assumptions lab

Topics covered: Critical ThinkingInformation Resilience