How to Use Trees and Fish to Diagram Root Causes

How effective we are at solving problems and keeping them solved depends on our ability to address them at their source. When we put out fires but fail to put in measures to prevent similar fires in the future, we fight the same fires again and again. Instead of increasing our fire-fighting budget, we should increase our investment in prevention at the source.

There are a several common techniques for identifying likely root causes. Either for the sake of simplicity, or because it worked in the past, many organizations zero in only one technique. Often this becomes an exercise in compliance, filling out a template rather than thinking through how to approach the problem. No single tool is versatile or reliable enough to get us to root cause in a majority of situations. Different situations call for different tools.  When in doubt, we can go a long way with a combination of fish, trees, and go see.

Solving real problems is easiest when we have a firm grip on reality. Reality tends to exist outside of our heads. But we need our heads in order to process objective reality that we can perceive into actionable thought. In short, we need to go see. Finding the prime suspects that cause a problem by observing the process is the simplest and most direct way, but this option is not always available due to the nature of the process or problem. This requires us to use a number of techniques to visualize our thinking.

One common technique is brainstorming. A group of people knowledgeable about the problem get together and list what they think are the causes. On the positive side, this is easy to do. It is just talking. Done in a structured way with affinity diagrams to simplify and organize brainstorming and PICK charts to help to prioritize, brainstorming can support rapid idea generation. Nevertheless, there is the risk that when we brainstorm causes, our thinking can be subjective, unscientific, or not grounded in a thorough understanding of the factors affecting the process. Also, without data to weigh one idea versus the next, the most vocal or highest ranked voices can sway the outcome.

The fishbone diagram is not a form of brainstorming. We should not be generating creative ideas about what could be causing the problem, i.e. guessing. The fishbone diagram works with a predetermined set of factors known to affect the outcome of a process. In general these are known as the “5M” for manpower, material, method, machinery and measurement. These are the factors that contribute to the majority of problems in manufacturing, healthcare, construction, landscaping and so forth. For retail businesses or the sales process of an organization, we can use “5P” for price, promotion, place, people, product.

To identify factors affecting problems in knowledge work, it is a simple matter of adjusting these categories to suit. Simply select four to six plausible factors are known or agreed to affect the outcome the process. An organization that has a limited idea of what factors contribute to a good outcome for their process needs to spend time understanding their process before diving into problem solving.

The fishbone diagram is not a template for 5 why analysis. Although it is possible to show 2, 3, 4 or more links in the cause-and-effect chain, getting to root cause is not its purpose. The fishbone diagram is more of a map of possible sources of variation in the process for further investigation.

A tree diagram looks like a fishbone diagram rotated 45 degrees. However, it serves a very different function. Unlike a fishbone diagram whose purpose is to broaden the list of suspects, the tree diagram is used to narrow down and eliminate possible causes, ideally to one or more addressable root causes. The tree diagram is an appropriate structure for practicing 5 why analysis.

Used in proper combination, these tools and methods help us to see reality clearly, structure our ideas logically and guide our investigation toward finding the root cause.



    November 6, 2018 - 11:31 am

    Excellent article! I was really struck by the statement that the “fishbone diagram is not a form of brainstorming.” How fine a point would you put on that? I’ve long treated fishbone diagrams as just another form of affinity diagrams, where the categories are set up front. – So as I think through this: would you say that the distinction is that with brainstorming/affinity the goal is opening up as many possibilities for consideration of a problem as possible, where as with a Fishbone diagramming exercise the inputs should really be targeted hypotheses, that is issues you can readily confirm at the gemba?

    • Jon Miller

      November 6, 2018 - 3:40 pm

      Hello David
      Good question.
      Anything that helps generate ideas can be used for brainstorming. However, that is not the intent of the fishbone diagram.
      Brainstorming is creative, open-ended, and imaginative while the fishbone diagram is closed-end, looking at scientific factors and measurable outcomes.
      Like affinity diagrams, fishbone diagrams list ideas under a header. However, affinity diagrams headers are the result of finding natural groupings and relationships among ideas generated, while the fishbone begins with a list of common factors (the 5M, etc.).
      Also, the ideas under the affinity are not often causally linked to each other, while they are be design in the fishbone diagram as the branching bones show.
      Both are constraining / converging tools, the difference is that brainstorming starts out by looking for a large number of ideas while root cause analysis is looking for ideally just one.
      The issues on the fishbone should not be targeted hypotheses, rather, likely starting points for investigation via go see, tree diagram, 5 why analysis, design of experiments and so forth.
      Not sure that answered your question on whether it’s a fine point or otherwise.

  2. Matt Kellett

    November 9, 2018 - 7:20 am

    In the book “how to be successful with continuous improvement” the author used a fish one with problem statement then had teams think of all the causes that could create the problem. They then affinities those caused by similarity and created custome headings for the fish one legs vs using the 5m or 5p headings. Ultimately they would then zero in on the leg with the most likely drivers (Pareto) and select top 3 potential causes on the leg then begin their RCA – 5 why (drip, drip – hopefully most people get that reference. Lol.) or data pulls or gemba etc. Do you see any weaknesses to this approach of grouping vs using the 5m/5p? I’ve tried both. Teams early in their journey like the affinities approach but I’m not sure if there are trade offs that I cant see to that approach. Thoughts?

    • Jon Miller

      November 9, 2018 - 10:32 am

      Hello Matt

      The weakness with the brainstorming + affinities approach is that you are guessing. Focusing on 5M or other science-based variables points us to where we need to make observation and further study. As you mentioned. It is less a question of whether a team is early in their CI journey and more one of whether they understand their processes and the factors that contribute to good outcomes. Creating the custom headings is OK if that is a means to understanding the process thoroughly, and as long as we don’t reimagine the wheel when existing categories will do.

Leave a Reply

Your email address will not be published. Required fields are marked *