# Fishbone Diagram: 5M Method, Example, and Hypotheses to Test

Fishbone diagram guide: use the 5M method, a traceable example, and a reusable template to test hypotheses without inventing a root cause.

- Canonical URL: https://www.harmate.com/en/blog/fishbone-diagram-5m-method-example-and-hypotheses-to-test
- Author: Harmate Team
- Published: 2026-09-04
- Updated: 2026-09-05T05:21:24.699812+00:00
- Language: en

## Content

# Fishbone Diagram: 5M Method, Example, and Hypotheses to Test

A fishbone diagram, also called an Ishikawa or cause-and-effect diagram, helps a team organize **possible causes** around an observable effect. It does not prove the root cause of a problem on its own.

Its value is investigative: surface competing hypotheses, keep the wording or traces that prompted them, make contradictions visible, and define the next check. This guide covers the 5M method, a complete example, and a reusable template.

## In brief: what is a fishbone diagram for?

A fishbone diagram broadens the exploration of a problem without settling on the first plausible explanation. The team starts with an observable effect and groups possible factors into families.

The [Agency for Healthcare Research and Quality](https://digital.ahrq.gov/health-it-tools-and-resources/evaluation-resources/workflow-assessment-health-it-toolkit/all-workflow-tools/cause-and-effect-diagram) describes it as a visual way to convey suspected and possible causes, then identify which ones need further analysis. The [Institute for Healthcare Improvement](https://www.ihi.org/library/tools/cause-and-effect-diagram) likewise presents it as a way to explore and display possible causes. It prepares an investigation; it does not replace observation, comparison, or a proportionate test.

## The 5M method — and when to add a sixth M

Categories are prompts, not pre-filled answers. The classic families are:

- **Manpower / people**: roles, skills, coordination, workload;
- **Methods**: instructions, processes, sequences, rules;
- **Machines / equipment**: tools, devices, interfaces;
- **Materials**: information, resources, or inputs;
- **Mother Nature / environment**: context and external constraints.

Add **Measurement** as a sixth branch when indicators, instruments, or measurement criteria are part of the problem. Do not force every factor into one box: a hypothesis may involve both methods and environment, and an experience can remain outside a category if classification would make it less precise.

## 1. Define an observable effect

A fishbone becomes unreliable when its head already contains an explanation. “Participants are not motivated” assigns an intention. “Participants leave the exercise before the group debrief” describes an outcome that can be documented.

Before drawing, specify:

- the situation in scope;
- the observed behavior or outcome;
- when it appears;
- the source that makes it observable.

For example: “During remote workshops, several participants leave the individual exercise before the group debrief, according to end-of-session responses and connection records.” It does not yet explain why. That is deliberate.

## 2. Collect experiences before filling the 5M branches

Writing People, Methods, Equipment, and Environment first encourages a team to fill familiar boxes. People closest to the process or most comfortable speaking may then define the problem for everyone else.

Collect concrete situations first:

- “Tell us about the moment when the exercise became difficult to continue.”
- “What did you need in order to move forward?”
- “When does the same exercise work better?”
- “What could contradict your first explanation?”

[Open-ended questions](https://www.harmate.com/en/blog/open-ended-questions-get-actionable-data-without-bias) preserve wording, context, and exceptions that preset choices can erase. They do not guarantee representativeness or remove bias. They provide richer material for exploring unexpected factors, which still need to be compared with other evidence.

A short individual collection before group discussion can also protect less-heard accounts. Keep the full wording; classify later.

## 3. Build the fishbone without turning branches into facts

Use this sequence:

1. **Write the observable effect** at the head of the diagram.
2. **Gather sources**: open responses, observations, incidents, documents, or available measures.
3. **Group related factors** without deleting wording that does not fit neatly.
4. **Use the 5M/6M categories as checks** after the first reading.
5. **Write each branch as a hypothesis**: “the instruction arrives too late” rather than “bad method.”
6. **Keep its source and limit visible**: what quote or fact prompted it, and what account qualifies or contradicts it?
7. **Name the next check**: observation, activity log, comparison, follow-up interview, or limited process test.

Each branch can now be reviewed by someone who was not in the workshop. They can see what was reported, what was inferred, and what remains to be established.

## 4. A traceable fishbone example

Consider a fictional case; the quotations below are entirely illustrative. After several workshops, participants report difficulty contributing to the first group exercise. Three statements stand out:

> “I did not know what the first step expected from me.”

> “I understood the instruction, but I could not find a way into the conversation.”

> “The start was clear for my group; we simply needed more preparation time.”

These accounts do not identify one cause. They open several hypotheses:

| Possible cause | Branch | Initial trace | Contradiction or limit | Next check |
|---|---|---|---|---|
| The first instruction lacks an example | Methods | One participant did not know what was expected | Others understood it | Observe how the instruction is restated in two sessions |
| Turn-taking blocks entry into the exercise | People | One participant could not enter the conversation | The issue does not appear in every group | Compare turn-taking rules and interview facilitators |
| Individual preparation is too short | Environment / Methods | One group reports missing preparation time | No timestamp confirms when disengagement begins | Review the actual agenda and test a longer preparation period |

The table does not elect a true root cause. It shows which hypotheses deserve investigation and which information is missing. A response that contradicts the majority is not noise: it may reveal a success condition or a difference between groups.

![Five-branch fishbone diagram: pictograms represent methods, equipment, people, materials, and environment. Three visual hypotheses extend the methods, people, and environment branches; the other two intentionally have no sub-cause.](https://api.harmate.com/media/blog/body/8e4a8877-0b40-40c4-a64d-9036d526d3f4.webp)

The five families structure the investigation without forcing every branch to be filled: only supported hypotheses are added.

## 5. A reusable blank template

Use this table during or after the session:

| Possible cause, written as a hypothesis | 5M/6M branch | Source wording or trace | Contradiction, exception, or missing information | Next check |
|---|---|---|---|---|
|  |  |  |  |  |
|  |  |  |  |  |
|  |  |  |  |  |

Add an owner and review point to the check, but do not treat ownership as evidence of causality. A branch may remain open, be contradicted by available information, or fall outside the decision scope.

## Fishbone vs. 5 Whys vs. Pareto: which tool should you use?

These tools can complement one another, but they answer different questions. The [NHS England improvement handbook](https://www.england.nhs.uk/commissioning/wp-content/uploads/sites/44/2017/11/the_handbook_of_quality_and_service_improvement_tools_2010-2.pdf) treats fishbone, 5 Whys, and Pareto as complementary tools for exploration, deeper questioning, and prioritization:

| Tool | Useful question | Output | Limit to keep visible |
|---|---|---|---|
| Fishbone | Which possible explanations should we examine? | A map of hypotheses | It does not prove causality |
| 5 Whys | How far should we investigate one hypothesis? | A chain of questions | Each “why” can remain speculative |
| Pareto | What deserves attention first? | Prioritization against a chosen measure | Frequency or weight does not prove mechanism |

Start with a fishbone when the problem is still open. Use 5 Whys to deepen a hypothesis that has a sufficient trace. Use Pareto to prioritize when the measure and its scope are explicit.

## Why a vote does not establish a root cause

A team may vote for the branch it considers most plausible. The vote helps allocate investigation time; it does not turn a shared impression into causality.

A methodological review in *Global Health: Science and Practice* notes that a fishbone diagram does not isolate the root cause by itself, can make causes appear equally important, and initially rests on opinion rather than evidence ([open-access article](https://pmc.ncbi.nlm.nih.gov/articles/PMC4858489/)).

Before acting, look for an independent trace suited to the hypothesis:

- direct observation of the actual process;
- a timestamped measure or working document;
- additional accounts from different situations;
- a comparison that accounts for audience, context, and protocol;
- when feasible and proportionate, a limited test of the proposed change.

Even when traces converge, keep the conclusion at the right level: “this hypothesis is supported enough to test an action” is more precise than “we proved the root cause.”

## Common mistakes

### Filling categories before listening

The 5M framework becomes a closed survey in disguise. Start with lived situations, then use categories to organize what emerged.

### Summarizing until the source disappears

“Communication problem” can collapse very different experiences into one label. Keep the quotes that led to the theme. A method for [analyzing a large volume of open responses](https://www.harmate.com/en/blog/analyzing-200-open-responses-without-bias-a-method-for-actionable-decisions) helps retain signal without manufacturing consensus.

### Confusing frequency with causality

A frequently mentioned factor is visible, not necessarily decisive. A rare incident may reveal a critical condition. Frequency directs attention; it does not replace verification.

### Looking for a culprit instead of a system

The People branch can quickly become “human error.” Ask which information, constraints, interfaces, or rules make the error possible. The goal is to understand a modifiable mechanism, not to dress blame up as analysis.

### Asking a question that contains the answer

“Does insufficient training explain the problem?” installs a cause before the investigation. [Framing bias](https://www.harmate.com/en/blog/framing-bias-why-your-questionnaires-already-contain-the-answers) can produce a coherent fishbone that discovers very little. Invite people to describe what happened, then examine competing explanations.

## Move from the diagram to a reviewable decision

At the end of the session, every branch should include:

- a possible cause, expressed without false certainty;
- the wording or trace that led the team to retain it;
- conflicting accounts and missing information;
- the next check, with an owner and a review point.

You can then distinguish hypotheses to explore now, hypotheses contradicted by available material, and hypotheses outside the current decision scope. A branch never becomes true because it was drawn well.

When the material comes from a questionnaire, a tool can help connect related wording, surface themes, and keep excerpts within reach. Harmate links readings to the responses behind them and lets teams question results in context; it does not decide causal relationships for them. Its [open-question analysis workflow](https://www.harmate.com/en/product/open-questions) can prepare a reviewable fishbone, after which field investigation takes over.

## Frequently asked questions

### When should you use a fishbone diagram?

Use one when the effect is precise enough to observe but several explanations remain plausible. If an independent trace has already established a cause, a fishbone can document contributing factors, but it does not need to manufacture a discovery.

### How many branches should a fishbone have?

There is no universal number. The 5M framework is a starting point; keep the branches that help the decision and group details only when they remain understandable. Remove branches that are decorative or irrelevant.

### Does a fishbone diagram find the root cause?

Not by itself. It maps possible hypotheses. The cause — or the level of causal support the evidence actually permits — must be examined through observation, measurement, comparison, or an appropriate test.

### Should you always start with the 5M categories?

No. Starting from experiences and an observable effect helps avoid forcing responses into preset categories. Use the 5M families afterwards to check that important factors were not overlooked.

## Key takeaway

A fishbone diagram is useful when it broadens exploration without closing the diagnosis too early. Define an observable effect, listen before classifying, keep source wording and contradictions visible, and pair every hypothesis with a check.

The fish organizes the discussion. Evidence comes from elsewhere.