Build an issue tree that tells you what to analyze first
An issue tree is finished when it tells you which analysis to do first and what result would prove you wrong. Anything less is a table of contents for work nobody finishes. Here is a worked tree, from the question to the kill result.
Transcript
A project kickoff. On the whiteboard: why are margins down? Ten minutes later, eight branches, each with an owner. Six weeks later, eight decks and no answer.
The branches overlapped, and nobody had said what they expected to find. An issue tree is finished when it tells you which analysis to do first, and what result would prove you wrong. Anything less is a table of contents.
Pick the tree that fits the question. Why did it happen? Split by the arithmetic of what moved. How could we do it? Split by the levers. Is our answer right? Split into the conditions that must all hold.
A worked tree, for an invented bakery chain. Why did gross margin fall from 38% to 35% in 2025, and what should change before the spring price list? The CFO needs the answer by March 15.
The whole is the three-point fall. The cut is margin per unit sold, giving four branches: price, mix, ingredient cost, and labor and overhead. A shift to wholesale moves both mix and price, so it counts under mix only.
No branch for competitors or the economy. Those are reasons a part might have moved, so they go one level down. Put them at the top, and the same lost point gets counted twice.
The hypothesis: most of the fall came from flour and butter costs never passed through to prices. The kill result, written before any data: if ingredient costs explain less than 1 of the 3 points, drop it and test mix.
Why commit so early? Charles Conn, co-author of Bulletproof Problem Solving, puts it bluntly: Every time you see a 50-page work plan that stretches out to three months, you know it’s wrong.
Then prune. Conn asks two questions of every branch: how important is it, and how much can we move it? Large and movable gets the work. Large and immovable becomes context. Small gets dropped. In his Pacific salmon work, ocean conditions were huge, but nobody could adjust them.
The fair objection: a hypothesis biases the team toward finding it. It can. So make it easy to lose, with a threshold agreed in advance. A team with no hypothesis is not neutral. It just analyzes everything at the same depth, and runs out of time.
Four rules of thumb before anyone owns a branch. A question, not a topic. An exact whole. A hypothesis that could be wrong. A number on the kill result. A tree that could never prove you wrong was never testing anything.
