# What is an issue tree? > **Issue tree:** An issue tree breaks one central question into sub-questions, level by level, until each branch can be answered with a fact, an analysis or a decision. - Source: https://ai-presentations.com/glossary/issue-tree/ - Updated: 21 October 2026 - Language: English - German version: https://ai-presentations.com/de/glossar/logikbaum/ Also called: logic tree, problem tree An **issue tree** (also called a logic tree) turns one big question into smaller ones. The key question sits at the root. Each level below splits it into sub-questions that are [MECE](https://ai-presentations.com/glossary/mece/): no overlaps and no gaps. You keep splitting until a branch can be settled with one analysis, one number or one decision. There are two common kinds. A *why* tree diagnoses ("Why did the margin fall?"). A *how* tree looks for solutions ("How can we release €4m of cash?"). ## A worked example A toolmaker wants to release €4m of cash within twelve months. Working capital is inventory plus receivables minus payables, so the first level is MECE by definition: - **Reduce inventory** - Finished goods - Work in progress - Raw materials - **Collect cash faster** - Shorten customer payment terms - Chase overdue invoices - **Pay suppliers later** - Renegotiate supplier terms Every leaf is something a team can size and test. Sizing shows where to start. On the slide below, €11.2m sits in inventory and receivables, and finished goods alone exceed the benchmark by about €1.6m. *The first two levels of the tree as numbers: inventory in three parts, then receivables and payables. Example with invented numbers.* 3. Inventory and receivables tie up €11.2m, and finished goods alone sit about €1.6m above benchmark ## How to build an issue tree 1. **Write a specific key question.** Include the metric, the scope and the deadline: "How can we release €4m of cash within twelve months?" 2. **Split by one logic per level.** Use a formula, a process or a clear set of categories, never two at once. 3. **Run the MECE test.** Could any item sit in two branches? Does anything fall outside all of them? 4. **Stop at answerable branches.** A leaf is done when one analysis or one decision settles it. 5. **Size and prune.** Estimate which branches matter most and analyze those first. ## Issue trees on slides The full tree rarely belongs in the main part of a deck. Show it once, on the approach slide or in the appendix, so readers see what you examined and what you ruled out. Then let each branch become a section and each leaf a slide with an [action title](https://ai-presentations.com/guides/action-titles/). An issue tree starts from the question; the [pyramid principle](https://ai-presentations.com/guides/pyramid-principle/) starts from the answer. Use the tree to find the answer and the pyramid to present it. ## Typical mistakes - Overlapping branches, such as "Sales" and "Marketing" next to "Customers". - Two logics on one level, for example regions next to products. - Leaves nobody can test, such as "Improve the culture". - A tree so deep that nobody reads it. Prune it. - A tree with no answer at the end. The tree is a means; the deck needs the conclusion. Next step: put numbers on the branches with a [driver tree](https://ai-presentations.com/glossary/driver-tree/). ## Frequently asked questions ### What is the difference between an issue tree and a decision tree? An issue tree structures a problem: it splits a question into sub-questions so that an analysis covers everything. A decision tree evaluates a choice: it maps options, chances and payoffs, often with probabilities. Use the first to organize the work and the second to pick between options. ### How many levels should an issue tree have? Usually two to four. Stop when each branch can be answered with one analysis, one data point or one decision. Deeper trees tend to turn into task lists. ### What is the difference between an issue tree and a driver tree? An issue tree is made of questions and stays qualitative. A driver tree is made of metrics and calculates, because each box is a formula of the boxes below it. Draw the issue tree when you do not yet know what to measure, and the driver tree once you do. ### What is a hypothesis-driven issue tree? It starts with an answer instead of a question. You state what you believe is true, then branch into the conditions that must hold for it to be true. Each branch becomes a test, and you drop the hypothesis if a test fails. ## Related pages - [MECE](https://ai-presentations.com/glossary/mece/): MECE means mutually exclusive, collectively exhaustive: a way to split a problem into parts that do not overlap and leave nothing out. With examples. - [Driver tree](https://ai-presentations.com/glossary/driver-tree/): A driver tree breaks a metric such as revenue or profit into the measurable drivers behind it, linked by simple math. See a worked example you can copy. - [Pyramid principle](https://ai-presentations.com/guides/pyramid-principle/): The pyramid principle by Barbara Minto: give the answer first, group the support and order it. With a worked example and an outline you can reuse for a deck. - [Storyline](https://ai-presentations.com/glossary/storyline/): A storyline is the logical chain of messages that carries a presentation from its opening to its conclusion. Definition, an example and how to build one. - [So what](https://ai-presentations.com/glossary/so-what/): The “so what” is the implication of a fact: why it matters to the audience and what they should do about it. Definition, examples and a test for slides. - [Consulting proposal](https://ai-presentations.com/templates/consulting-proposal/): Propose the work in 8 slides: what you heard, the evidence, the approach in phases, the timeline, fixed fees per phase, the team and next steps. --- ai-presentations.com is published by Slaide (https://www.slaide.de), a sole proprietorship run by Nathalie Scholl in Heilbronn, Germany. [How this site is written, checked and corrected](https://ai-presentations.com/editorial-policy/).