Six-Step Causal Analysis
What is Six-Step Causal Analysis?
Six steps. (1) Define the outcome precisely — object, period, magnitude: “sales of this product in the Southeast fell 20% for three consecutive months”, not “performance is poor”. (2) Force out at least three candidate causes, spanning market, product, channel, and team. (3) Draw the causal chain so every intermediate link can be explained. (4) Find evidence for and against each hypothesis — if it holds, what should you be able to see (price-sensitive customers leaving)? If you can’t see it, lower your confidence. (5) Verify in a small area — one region or one team, changing exactly one factor. (6) Keep reviewing: an improvement doesn’t confirm the cause (something else may have moved), and no improvement doesn’t mean the measure was useless (were the other conditions steady?).
Six-Step Causal Analysis: common mistakes and how to handle them
Four questions for verification: what else should accompany this cause; is there a case with the cause but not the effect; a case with the effect but not the cause; and can you change the factor cheaply and watch what happens? Outcomes that matter usually have several causes at once, so start with the factor that has the largest effect, is actually changeable, costs least, and carries least risk. A worked case: rather than cutting price when volume fell, the analysis found long delivery times and slow support; a new delivery arrangement raised the close rate without touching margin.
What the Six-Step Causal Analysis questions test
identifying which of the six steps an analysis skipped; spotting a vaguely defined outcome; choosing the observation that would falsify a hypothesis; designing a small test that changes only one factor; catching the bad inference at review time — it improved so we were right, or it didn’t so the measure was useless.