Build issue trees: diagnosis trees and solution trees

Issue trees (also called issue maps and logic trees) are at the heart of our problem-solving approach and, therefore, are a recurrent theme on this site. We use two types of trees: diagnosis trees and solution trees.

An issue tree is a graphical breakdown of your key question. Trees have four basic rules:

  • They consistently answer why or how questions (depending on your key question)
  • They progress from the key question to the analysis as they move to the right
  • Their branches are mutually exclusive and collectively exhaustive (MECE)*
  • They use an insightful breakdown (more on this here)

* The classification of items is MECE; the items themselves are usually only independent and collectively exhaustive (ICE), not entirely MECE.

Issue trees are useful to efficiently consider all the potential answers for your problem and understand their relations.

Why issue trees are for diagnosing your key question

Why issue trees—or diagnosis trees—help you search for all the possible causes of a problem. You list these in logical groups on the first column to the right of your key question, ensuring that your groupings are mutually exclusive (there are no overlaps) and collectively exhaustive (there are no gaps)—or MECE. As you progress to the right, you drill further down in the details of each grouping. In that sense, diagnosis trees are similar to Ishikawa diagrams, except that they flow in the opposite direction.

Why issue trees / issue maps look for the root causes of a problem.

Why issue trees / issue maps help you identify the root causes of a problem.

 

How issue trees / solution trees are for actively looking for ways to correct your problem

A 'how' issue tree

A how issue tree, showing how the tree helps to capture all the potential answers to the key question.

With how trees, you look for all the potential solutions to your problem. As a general rule, you want to know the why before you get to the how, so if you don’t know the root cause(s) of your problem, find these first.

Issue trees progress further into details until elements are sufficiently explicit. Then come the hypotheses, analyses and data sources (more on that in later posts). Once you have tested which solutions are viable, you are ready to select one, for instance by using a decision matrix.

A standard decision tree breaks down a key question into MECE branches and then states explicitly the expected value of each branch—by multiplying the probability of success by the payoff associated with success. Although we usually don’t use a how issue tree to spell out the expected value of each outcome, these trees are in fact more complete that standard decision trees (click here to see a formal comparison between issue trees and decision trees). Since we use how issue trees to identify potential solutions and help deciding which one(s) to implement, we refer to them as solution trees.

Diagnosis and solution trees share some properties

Irrespective of their types, issue trees are insightful: they break down the key question in a meaningful way, adding value with each column.

Issue trees are useful for several things. First, they provide a graphical breakdown of the problem, a sort of roadmap. By including all possible answers to your key question in your issue tree, you reduce the probability that you’ll overlook some. Also, they spell out the various alternatives in minute details. They are useful in managing your problem-solving effort—especially if you are leading a team—because they help you identify visually where you are dedicating resources.

In the example above, we looked at how a taylor house can sell high-price suits. Using a decision tree allows to clarify that offering a high-quality product is important but it is not the only possibility.

Indeed, associating prestige with the brand and offering an exceptional buying experience are two other ways to go about charging more for a suit.

Issue trees are standard in any strategy consultant’s toolbox. However, although these consultants insist that branches in trees are mutually exclusive, the final recommendation might be a mix of various branches (as in the example above: we could both improve the quality of the product and associate prestige to our brand as part of our effort to sell more high-priced suits); therefore, branches really are independent (more here) and not necessarily mutually exclusive.

Learn more about diagnosis trees and decision trees

In any situation, don’t exhaust yourself finding the one perfect issue tree, there usually is more than one.

Conducting a profitability analysis (i.e., looking at how revenues and costs fare up) is a common theme in business. Here is a profitability issue tree that may help you do so.

Here is a slide deck on how to build issue trees from my course, and if you are confused by the many types of trees that you hear about, here is a proposal for a naming convention.

Here is an example of a hypothesis tree and a solution tree for the same case.

Note: as of 2013, I’ve come to prefer referring to trees as maps as doing so seems to reduce confusion between issue trees/maps (especially solution ones) and decision trees. So, I’m now talking more about diagnostic maps or why maps (formally known as why trees) and solution maps or how maps (formally know as how trees).

References

Davis, Ian, David Keeling, Paul Schreier and Ashley Williams. “The Mckinsey Approach to Problem Solving.” McKinsey Staff Paper, no. 66 (2007): 27.

De Kleer, Johan and Brian C Williams. “Diagnosing Multiple Faults.” Artificial intelligence 32, no. 1 (1987): 97-130.

Duncker, Karl and Lynne S Lees. “On Problem-Solving.” Psychological monographs 58, no. 5 (1945): i.

Fischhoff, Baruch, Paul Slovic and Sarah Lichtenstein. “Fault Trees: Sensitivity of Estimated Failure Probabilities to Problem Representation.” Journal of Experimental Psychology: Human Perception and Performance 4, no. 2 (1978): 330.

Kazancioglu, Emre, Ken Platts and Pete Caldwell. “Visualization and Visual Modelling for Strategic Analysis and Problem-Solving.” In Information Visualisation, 2005. Proceedings. Ninth International Conference on, 61-69: IEEE, 2005.

Platt, John R. “Strong Inference.” Science 146, no. 3642 (1964): 347-353.

Wojick, David. Issue Analysis – an Introduction to the Use of Issue Trees and the Nature of Complex Reasoning, 1975.

Related Posts:

7 comments to Build issue trees: diagnosis trees and solution trees