Issue Tree Generator with Driver Values

What is an issue tree?

An issue tree is a diagram that takes one problem, written as a question, and breaks it into smaller parts, level by level. The root sits on the left and each level to the right splits the node before it, so you can see every part of the problem on one page and decide which parts to examine first. It is also called a logic tree or, when the branches carry numbers, a driver tree.

The generator above draws the tree as you type. You can edit the nodes directly, paste an indented outline or a CSV from a spreadsheet, mark the end nodes as hypotheses and export the result.

How to make an issue tree

1. 1

   Write the root as a precise question

   Name the decision-maker, the metric and the time frame: "Why did profit fall 20% year over year?" is workable, "Fix profitability" is not. Type it in the first row.

2. 2

   Split it into branches that do not overlap

   Use the add-branch button to break a node into 2 to 5 parts. Good splits are mutually exclusive and together cover the whole: profit = revenue minus costs; revenue = volume times price.

3. 3

   Keep going until a branch can be tested

   Stop splitting when you can answer a node with a data pull, a calculation or an interview. Mark those end nodes as hypotheses (diagnostic trees), or leave them as Question nodes that read as actions (solution trees).

4. 4

   Reorder and tidy

   Move nodes up and down, or in and out a level. A moved node always takes its branches with it, and the picture redraws at once.

5. 5

   Add numbers if it is a driver tree

   Tick "Driver tree: add numbers", type a value on each leaf and choose sum or product for each parent. Every parent recalculates as you type.

6. 6

   Export

   Download SVG or PNG for a slide, the CSV for a spreadsheet, or the outline text to keep the tree as an editable list. Paste the outline or CSV back later to carry on editing.

Diagnostic, solution and driver trees

The layout is the same; the question you ask at each level decides which kind of tree you are building. Pick the type before you start, because mixing "why" and "how" branches on one tree is the most common reason a tree stops making sense.

A diagnostic tree works best when it is hypothesis-driven: instead of collecting every fact first, you state a likely answer at the end of each branch, such as "win rate on demos fell", and then look for the one piece of data that would prove or kill it. Mark those end nodes as Hypothesis. Branches that fail the test are crossed out of the analysis, and the ones that survive tell you where to dig next.

Driver tree: turn the branches into a formula

Tick "Driver tree: add numbers" and each leaf gets a value box. Every node with branches then shows the sum or the product of its branches, whichever you choose, and the number updates as you type. There is no separate subtraction rule: to subtract, enter the amount as a negative number, so profit is revenue plus (minus costs).

In the example below, new customers are 1,000 qualified leads times a 0.2 win rate, which is 200 customers. At an average price of $6,000 that is $1,200,000 of new customer revenue. Existing customers add 500 times a 0.9 retention rate times $6,000, which is $2,700,000. A credits line of minus $50,000 brings annual revenue to $3,850,000. Change the win rate and the whole chain recalculates, which makes it easy to see which lever moves the total most. A unit such as $ or % is only a label and never changes the maths, so enter a percentage as a decimal (0.2 for 20%), and type numbers without thousands separators.

Issue tree vs decision tree, fishbone and mind map

These four diagrams look alike but answer different questions. An issue tree splits a problem into parts; a decision tree compares choices under uncertainty; a fishbone diagram sorts the causes of one effect into fixed categories; and a mind map lets ideas spread in any direction.

A common sequence is to brainstorm on a mind map, structure the problem as an issue tree, test the hypotheses, and then score the solution options with a Pugh matrix.

Common issue tree mistakes

• A root that is a topic, not a question. "Customer retention" invites a list of ideas; "Why did 12-month retention drop from 90% to 85%?" tells you what to split.
• Branches that overlap. If "marketing" and "ads" both appear on one level, the same cause is counted twice and the tree cannot be trusted as complete.
• Branches that leave a gap. Check each level by asking whether everything that could cause the parent is somewhere below it. If not, add an "other" branch or re-cut the split.
• Splitting on a different basis at the same level, for example by region on one branch and by product on its sibling. Pick one lens per level.
• Going deep on the first branch and stopping early on the rest. Aim for similar depth, then prioritize which branches to test first.
• Leaving the ends as vague topics. A leaf should be something you can check, such as "win rate on demos fell", not "sales issues".

Using it responsibly

The generator draws and calculates what you enter. It does not check that your branches are complete or non-overlapping, and it does not know whether your numbers are right. Treat the tree as a structured way to organize an analysis, and confirm the figures against your own data before sharing a driver tree with others.