Ishikawa diagram (6M): how to find the root cause

In one line

The Ishikawa diagram (or "fishbone") sorts every hypothesis about what is causing a problem into 6 categories, so you do not stop at the first cause that "sounds logical" and end up attacking the wrong symptom.

What it is

The Ishikawa diagram, also called a cause-and-effect diagram because of its "fishbone" shape, is a simple graphical tool for organizing hypotheses about the cause of a problem before you go looking for a solution. It was developed by the Japanese engineer Kaoru Ishikawa and is one of the 7 basic quality tools.

The structure is always the same: the head of the "fish" holds the Effect, the specific problem you want to solve, worded as precisely as possible. A central line runs out from there, and 6 branches come off that line, one for each possible category of cause: the 6Ms.

1. Machine — the equipment, tool or technology used
2. Material — supplies, raw material, parts or components
3. Method — the procedure or the way the task is done
4. Manpower — the people who do the work
5. Measurement — how the process is measured, calibrated or controlled
6. Environment — the physical setting where the work takes place

Each branch is filled in with concrete hypotheses, not generic complaints, about why that category might be contributing to the effect.

What it is for

It is for a specific situation: when a problem has several possible causes and the team risks discussing only the first one someone thinks of (typically "it's the machine" or "it's the operator"). The diagram forces you to go through the 6 categories one by one before deciding where to investigate first, so no source of cause is left out just because nobody thought to mention it.

It is the natural step before a deeper analysis: first you open up the range of possible causes with Ishikawa, then you prioritize which ones to investigate (for example with a Pareto chart, if you have frequency data), and only then do you dig into the chosen cause, for example with the 5 Whys technique to reach the root cause.

How it is applied

The basic steps, to be done as a team (not alone at a desk):

1. Gather the people who know the problem — operators, supervisors, maintenance, quality. The more different points of view, the more complete the diagram.
2. Define the Effect precisely. "Low quality" is not enough; "3 out of every 10 parts come out with burrs on the late shift" is.
3. Go through the 6 categories one by one and write down, for each one, every hypothesis the team proposes. At this stage nothing is discarded and nobody debates whether it is true or not.
4. Review the whole diagram and mark which hypotheses look most likely or easiest to verify first.
5. Confirm the prioritized hypotheses with data, instead of assuming the most "obvious" one is the right one.

This order matters: step 3 is for generating ideas, not for filtering. Mixing the two makes the team self-censor, and the diagram ends up with two or three causes instead of the ten or twenty there really are to choose from.

Real example

A case study published in the academic journal CULCYT (2025) applied a 6M Ishikawa diagram to the inspection and adjustment area of a plastic thermoforming company in Mexico City (about 400 employees in total, 35 in that particular area).

Effect: average productivity below 50% across the area's three shifts, and a scrap level of 74% on the inspected parts, far below the internal target, set at a minimum of 70% productivity per part.

The analysis team (continuous improvement manager, production manager, shift leads, supervisors and operators) went through the 6 categories over 6 months of on-site work, combining interviews, direct observation and document review. Some of the concrete hypotheses recorded for each category:

Machine — of the 5 machines in the area, there was no common operating standard among them; maintenance was done "by eye," based on each supervisor's empirical knowledge, with no preventive plan.

Material — out-of-specification material arrived from other areas, with no defined times for inspecting it and no clear criteria for releasing it; rejects were not recorded anywhere.

Manpower — about half of the area's staff had physical limitations (temporarily reassigned after an injury at another workstation), which led to high turnover; new staff learned by watching another operator, with no formal training process.

Measurement — there were no measurable production targets per catalog item and rejects were not recorded; material up to a year old had piled up in the area, with no traceability.

Environment — ovens left on all day produced heat and a bad smell even when not in use; chairs and tables of the wrong height forced the operators to change posture constantly.

The most useful finding was not any single cause but the pattern: when the roughly 30 hypotheses collected were cross-checked against the three types of Lean waste (Muda, Muri, Mura), most fell under Muda, waste caused by a lack of standardization and documentation, rather than overload or variability. That showed the team where to focus the improvement effort first, instead of spreading it equally across the 6 categories.

How to build one

To fill it in on screen (what you enter is saved in your browser), use the interactive tool.

The same 6 categories, with 1-2 guiding questions for each, work as a starting point for a brainstorming session:

Effect: describe the specific problem, with a concrete data point if possible (what happens, where, how often or how much).

1. Machine
   • Is the equipment operating outside its normal parameters (wear, expired calibration, overdue maintenance)?
   • Does the same machine operate differently between shifts or operators?
2. Material
   • Does the supply or material entering the process meet the specification?
   • Is there variability between batches or suppliers that could explain the problem?
3. Method
   • Is there a written, up-to-date procedure for this task, or does everyone do it "their own way"?
   • Did anything in the process change recently without being documented?
4. Manpower
   • Did the person doing the task receive formal training, or did they learn by watching?
   • Is there high turnover or instability at this workstation?
5. Measurement
   • How do you know whether the process is going well: is there objective data, or does it rely on someone's perception?
   • Is the measurement criterion or instrument the same for everyone, and is it calibrated?
6. Environment
   • Could the conditions of the place (temperature, noise, lighting, tidiness) be affecting the result?
   • Does the ergonomics of the workstation help or hinder doing the task well?

Build your own diagram

Enter the effect (the specific problem) and add hypotheses under each of the 6 categories. It is saved in this browser, so you can come back later and keep filling it in.

Benefits

• It gives structure to a brainstorming session: nobody skips a whole category just because nobody came up with anything for that branch
• It works well in a team, since each area contributes hypotheses from its own knowledge (maintenance does not see what quality sees, and quality does not see what the operator sees)
• It needs no prior data or software: a whiteboard, a sheet of paper or this very template is enough
• It is the natural gateway to deeper tools, such as the 5 Whys or a formal RCA

Limitations to keep in mind

• It organizes hypotheses, it does not confirm them: each branch is still just an opinion until it is verified with data (which is why the CULCYT real case combined it with the 5 Whys for the subcauses, and with Lean's 3 Mu to prioritize afterward)
• With a poorly trained team, the tendency is to pile many trivial causes onto a single branch and lose focus
• On its own it does not prioritize which cause to attack first; for that you need to cross it with frequency or impact data
• If the Effect is poorly defined (too broad or ambiguous), the whole diagram loses its usefulness, no matter how many hypotheses are collected

In summary

The Ishikawa diagram does not solve a problem by itself: it structures the conversation so the team systematically reviews the 6 typical sources of cause (machine, material, method, manpower, measurement and environment) before deciding where to investigate. The real case from the plastics plant shows the value of doing it rigorously: of almost 30 hypotheses collected across the 6 categories, cross-checking them afterward against the types of Lean waste showed the team that most of the problem was a lack of standardization, not the machine or the operator, who are usually the first suspects.