Why Did It Move?

By the end of this lesson, you should be able to: run a metric investigation as a fixed procedure rather than a hunt, explain why the order of the steps matters, split a single decline into several independent causes that sum correctly, and finish with a recommendation instead of a description.

One question, one procedure

Everything in this track has been a piece of this. Here it is assembled, run against the question you'll actually be asked.

Basket's checkout conversion fell from 20.7% to 16.2% over the quarter. Why?

Most candidates start guessing. The move that separates a good answer is to run the same six steps every time, in the same order, and say so out loud.

Loading diagram…

The order isn't a preference. Each step removes an explanation that would otherwise contaminate the next one.

Step 0: is it real?

Before anything, a baseline with an interval. Lesson 14 put Basket's daily noise at ±4.8%.

A 4.5-point fall on a 20.7% base is a 22% relative decline, four times the noise band. It clears comfortably. There's something to explain.

This step ends more investigations than it starts, and that's its value. Most "why did X drop" questions are answered by "it didn't".