Issue Trees
An issue tree turns MECE logic into a working map of the problem. It is the single most practical tool in the consulting toolkit and the one most tested in interviews.
Anatomy of an issue tree
At the root sits the question. Each level below breaks the level above into MECE components. Branches continue until they reach a question that can be answered with a specific analysis or a specific piece of data — the point at which further splitting adds no value.
Depth is typically three to four levels. Beyond that, teams usually discover they are splitting for its own sake.
Diagnostic versus solution trees
A diagnostic tree asks why something is happening: why has margin fallen. A solution tree asks how to achieve something: how might we raise margin. Most engagements need both, in that order — diagnose the cause, then generate options against it.
Mixing them produces the classic beginner error: a tree whose branches are simultaneously causes and remedies, so nothing can be sized or tested.
Sizing and pruning
Once the tree exists, size each branch quickly and roughly. Rough sizing is enough to see which branches hold most of the value. Then prune: park small branches explicitly, note them in an appendix, and spend the engagement on the two or three that matter.
Pruning must be visible to the client. Silently dropping branches destroys trust when someone later asks whether you considered them.
Turning branches into workplan
Each surviving leaf becomes a work item with an owner, a required data source, an analysis method, and a due date. This is how an issue tree converts directly into a project plan, and why interviewers use it to assess whether a candidate can actually run work rather than just discuss it.
Example: a two-level tree in practice
Question: why has our EBITDA margin fallen from 14 to 9 percent in two years?
Level 1 — arithmetic: revenue effects and cost effects.
Level 2 — revenue: price realization (list price, discount depth, customer mix) and volume (retention, new acquisition, share of wallet). Cost: direct cost per unit (input prices, yield, labour productivity) and fixed cost base (headcount, facilities, overheads).
Rough sizing shows discount depth explains roughly 60 percent of the decline and input prices most of the rest. Six branches are parked with one line of explanation each. The engagement now has two workstreams instead of twelve, and a defensible reason for that choice.
- Each level of an issue tree must be MECE relative to the level above.
- Split until a branch can be answered with a specific analysis — usually three to four levels.
- Keep diagnostic (why) and solution (how) trees separate.
- Size roughly, prune visibly, and document what was parked.
- Surviving leaves become the workplan with owners and data sources.
Module quiz
Pass mark 70%1. When should you stop splitting a branch of an issue tree?
2. What is the main risk of pruning branches without telling the client?
3. A tree contains the branches 'high input costs' and 'renegotiate supplier contracts' at the same level. What is wrong?