Fishbone Diagram (Ishikawa): How to Use It with Examples
The fishbone (Ishikawa) diagram explained — the 6M categories, how to build one step by step, a manufacturing example, and how it pairs with 5 Whys.
A fishbone diagram is a visual cause-and-effect tool that maps every contributing factor behind a problem onto a structured skeleton, so a team can see all potential causes at once rather than debating them one by one. The "fish head" on the right is the problem statement (the effect); the "bones" branching to the left are categories of causes; and the smaller sub-branches are specific causes within each category. Because it resembles a fish skeleton and was developed by Japanese quality pioneer Kaoru Ishikawa, it is also called an Ishikawa diagram or a cause-and-effect diagram.
The tool is a staple of manufacturing quality systems — it appears in Six Sigma DMAIC, ISO 9001 corrective-action procedures, and reliability-centered maintenance programs — but it applies anywhere a team needs to structure a root cause investigation before jumping to conclusions.
What Is a Fishbone (Ishikawa) Diagram?
A fishbone diagram organizes potential causes of a problem into major categories so that nothing is overlooked during an investigation. The key characteristics:
- Visual and collaborative — the entire team sees the same picture and contributes simultaneously
- Exhaustive by design — predefined categories force the team to consider cause families they might otherwise skip
- Hypothesis-generating, not hypothesis-proving — it surfaces candidate causes; data and testing confirm or rule them out
- Non-quantitative — it does not rank causes by likelihood or impact (that step happens separately, often through FMEA or data analysis)
The diagram is almost always built in a group setting — typically a 60-to-90-minute facilitated session with the people closest to the process. The output is a populated diagram that feeds the next step of a root cause analysis.
The 6M Categories
The standard manufacturing framework groups causes into six categories, each starting with M. These are sometimes called the 6Ms of fishbone analysis:
| Category | Also Called | What It Covers |
|---|---|---|
| Machine | Equipment | PLCs, drives, motors, conveyors, tooling, fixtures, maintenance history |
| Method | Process | Work instructions, SOPs, process parameters, sequencing, changeover procedures |
| Material | Input | Raw material quality, lot variation, supplier changes, shelf life, storage conditions |
| Manpower | People | Operator skill, training, fatigue, shift handover, staffing levels |
| Measurement | Inspection | Gauge calibration, sensor accuracy, sampling plans, measurement system variation |
| Mother Nature | Environment | Temperature, humidity, vibration, dust, EMI, seasonal variation |
Why 6M? The categories were chosen because they cover the full input space of a manufacturing process. If a cause does not fit neatly into one category, it usually means the problem statement needs sharpening or the cause is actually a symptom of something deeper.
Some industries add a seventh M — Management or Maintenance — when systemic organizational factors need to be separated from day-to-day process variation. Service and healthcare contexts sometimes use the 8P variant (Product, Price, Place, Promotion, People, Process, Physical Evidence, Productivity).
How to Build a Fishbone Diagram Step by Step
Follow these seven steps to run a fishbone session from a blank page to an actionable cause list:
Step 1 — Define the Effect (Problem Statement)
Write a precise, measurable problem statement and place it in the fish head. Vague effects produce vague diagrams.
- Weak: "Machine keeps breaking down"
- Strong: "Line 3 packaging machine produced 4.2% overfill defects during shift A on 12 June 2026, up from a 0.3% baseline"
Specificity about what, where, when, and how much prevents the team from drifting into unrelated issues.
Step 2 — Draw the Spine and Category Bones
Draw a horizontal arrow pointing right to a box containing the problem statement. Add six diagonal bones branching from the spine — three up, three down — and label each with one of the 6M categories. This is the blank skeleton.
Step 3 — Brainstorm Causes in Each Category
For each M category, ask: "How could [this category] cause or contribute to the effect?" Capture every suggestion without filtering. Typical prompts:
- Machine: Has any equipment setting changed? When was this machine last calibrated or serviced?
- Method: Is there a documented procedure? Are all operators following the same steps?
- Material: Has a new batch or supplier been introduced recently?
- Manpower: Is this defect shift-dependent? Who was operating the machine when defects peaked?
- Measurement: Is the measurement system capable? Could the "defect" be a measurement error?
- Mother Nature: Has ambient temperature or humidity shifted? Is there a vibration source nearby?
Add each cause as a sub-branch on the appropriate bone. Causes can have sub-causes — branch them further inward.
Step 4 — Look for Repeated Causes
Causes that appear under more than one category are high-priority candidates. If "no calibration record" shows up under both Machine and Measurement, that convergence signals a likely contributor.
Step 5 — Prioritize for Investigation
Circle the three to five causes the team believes are most likely based on their process knowledge. These become the items to investigate with data — trend charts, process logs, PLC fault histories, measurement system analysis, or direct observation.
Step 6 — Gather Data to Confirm or Eliminate
A fishbone is a hypothesis map, not a conclusion. Each prioritized cause needs a test:
- Retrieve PLC event logs and alarm histories for the period in question
- Pull quality data stratified by shift, machine, and material lot
- Perform a gauge repeatability and reproducibility (GR&R) study if measurement is suspect
- Run a trial with controlled material, operator, and method to isolate variables
Step 7 — Document and Act
Once data confirms a root cause, link it to a corrective action with an owner and due date. Keep the fishbone diagram as part of the corrective action record — it shows the investigative rigor and prevents the same team from repeating the exercise six months later.
A Worked Manufacturing Example
Problem statement: Sheet metal press on Line 5 produced 6.8% out-of-tolerance parts (±0.3 mm spec, actual deviation +0.5 to +0.8 mm) during the night shift on 10 June 2026.
Here is how the team populated the 6M bones:
| Category | Causes Identified |
|---|---|
| Machine | Press tonnage drifted upward; hydraulic pressure relief valve worn; last PM was 14 weeks ago (scheduled 8 weeks) |
| Method | Night-shift SOP omits the pre-shift die inspection step; no procedure for checking tonnage at shift start |
| Material | New steel coil from alternative supplier (higher yield strength); material cert not reviewed before production |
| Manpower | Night shift operator recently moved from a different line; no formal sign-off on this press type |
| Measurement | CMM calibration expired 3 days prior; in-process gauge not zeroed at shift start |
| Mother Nature | Plant ambient temperature drops ~8°C overnight; thermal contraction of die not accounted for in setup |
Convergence signals:
- Measurement (expired CMM + unzeroed gauge) and Machine (pressure relief valve wear) both point toward process drift going undetected.
- Material (higher yield strength) and Machine (tonnage drift) interact — the press was not retuned when the new coil was introduced.
Prioritized causes for investigation:
- Hydraulic pressure relief valve condition — check pressure logs and inspect valve
- New material yield strength vs. press setup — compare material cert to press settings
- CMM calibration status — verify and recalibrate
The team confirmed the root cause combination: the alternative-supplier coil required 12% higher press force, but the pressure relief valve had degraded and was allowing uncontrolled tonnage creep. The measurement system failure masked the trend until defects escalated.
Automation Example: Intermittent Sensor Fault Categorized by 6M
Fishbone diagrams are equally useful for diagnosing control system faults. Here is an example from a bottling line where an inductive proximity sensor on a conveyor gate intermittently failed to detect bottles, causing the PLC to trigger false reject pulses.
| Category | Causes Investigated | PLC/Control Detail |
|---|---|---|
| Machine | Sensor mounting bracket vibration; sensor body out of spec sensing range; cable shield damaged at drag chain entry | PLC I/O card input filter time set too short (2 ms) — legitimate slow targets filtered out; drive-generated EMI coupling into sensor cable |
| Method | Sensor replaced without verifying sensing distance against new bottle geometry; no functional test step in maintenance procedure | PLC program logic: sensor debounce timer (T_ON) set to 0 ms after a hurried repair — single noise spike registered as valid signal |
| Material | Bottle colour changed from clear to dark amber — reduced reflectivity for capacitive sensor (wrong sensor type flagged) | N/A |
| Manpower | Maintenance tech replaced sensor with a different part number; datasheet not checked | N/A |
| Measurement | No oscilloscope capture of sensor output waveform during fault; fault logged as "intermittent" without timestamp resolution | PLC fault log polled at 1-second intervals — sub-second glitches not captured; historian trend not enabled for that I/O point |
| Mother Nature | Condensation on sensor face during morning startup (plant runs cold overnight); EMI from adjacent VFD on same cable tray | N/A |
Root cause confirmed: PLC I/O filter time (2 ms) and debounce timer (0 ms) were insufficient to reject EMI spikes from the adjacent VFD, which had been upgraded to a higher-power model three weeks before faults began. The sensor itself was functional. Corrective actions: increase I/O filter to 8 ms, set T_ON debounce to 20 ms, route sensor cable on a separate cable tray, and add the I/O point to the historian with 100 ms resolution.
This example illustrates why Machine and Measurement bones are the most productive starting points for intermittent PLC faults — hardware state and data visibility are usually where the investigation gains traction fastest.
Fishbone vs 5 Whys
Both tools are used in root cause analysis, and they work well together. Understanding the difference prevents teams from misapplying either.
| Dimension | Fishbone Diagram | 5 Whys |
|---|---|---|
| Structure | Broad, categorical — explores many cause branches simultaneously | Linear — follows one causal chain downward |
| Best for | Complex problems with multiple possible causes; team brainstorming | Simpler, well-understood problems; drilling into one confirmed cause |
| Team size | Works best with 4–10 people from different functions | Can be done by 1–3 people; less facilitation overhead |
| Risk | Can become unwieldy if every branch is treated equally | Can reach a "wrong" root cause if the initial direction is incorrect |
| Typical pairing | Fishbone first to generate hypotheses → data to shortlist → 5 Whys to drill into the confirmed cause | Standalone for quick investigations; as a follow-on after fishbone narrows the field |
The most effective approach for complex manufacturing or automation faults: use the fishbone to map the landscape, data analysis to shortlist, and the 5 Whys to drill down into the confirmed causal chain.
Tips and Common Pitfalls
Tips for a productive fishbone session:
- Appoint a neutral facilitator who keeps the group on track without advocating for particular causes
- Use physical sticky notes on a whiteboard rather than a laptop — it keeps everyone standing, engaged, and contributing simultaneously
- Time-box each M category to roughly 8–10 minutes to maintain energy and prevent any one branch from dominating
- Invite diverse roles — operators, maintenance technicians, quality engineers, and process engineers see different facets of the same problem
- Date and sign the diagram so it becomes a traceable quality record
Common pitfalls to avoid:
- Jumping to solutions during brainstorming — the fishbone session is for causes only; action items come after data confirms the root cause
- Treating opinions as data — a cause written on the diagram is a hypothesis until evidence supports it; avoid the fallacy of acting on the most senior person's favourite branch
- Leaving the diagram as the deliverable — a finished fishbone with no follow-up data collection is a wall decoration, not a corrective action
- Reusing a fishbone from a previous incident — similar-sounding problems often have different root causes; start fresh each time
- Ignoring the Measurement bone — teams with strong process knowledge often skip straight to Machine or Manpower; undetected measurement system variation has caused more false corrective actions than almost any other single factor in manufacturing quality
Frequently Asked Questions
What is a fishbone diagram?
A fishbone diagram (also called an Ishikawa or cause-and-effect diagram) is a visual tool that maps all potential causes of a problem into a structured skeleton shape. The problem is written in the "head" of the fish; the "bones" represent categories of causes such as Machine, Method, Material, Manpower, Measurement, and Environment. It is used in manufacturing, quality management, and process improvement to organize brainstorming before a root cause investigation.
What are the 6Ms in a fishbone diagram?
The 6Ms are the six standard cause categories used in manufacturing fishbone analysis: Machine (equipment), Method (process/procedure), Material (inputs), Manpower (people), Measurement (inspection and sensing), and Mother Nature (environment). Each category prompts the investigation team to consider a different family of potential causes so that nothing is overlooked.
How do you make a fishbone diagram?
Draw a horizontal arrow pointing to a box that contains a precise problem statement. Add six diagonal bones from the spine — three up, three down — and label them with the 6M categories. In a facilitated team session, brainstorm potential causes under each category and add them as sub-branches. Identify causes that appear in multiple categories, then prioritize three to five for data-driven investigation.
When do you use a fishbone diagram vs 5 Whys?
Use a fishbone diagram when the problem has multiple plausible causes and the team needs to explore broadly before committing to an investigation path. Use the 5 Whys when the cause family is already narrowed and you need to drill down along a single causal chain. For complex automation and manufacturing faults, the most effective sequence is fishbone first → data to shortlist causes → 5 Whys to find the root.
Related Guides
- Root Cause Analysis: Methods, Steps, and a Worked Example — the broader RCA process that the fishbone feeds into
- What Is FMEA? Failure Mode and Effects Analysis Explained — the proactive complement to fishbone; quantifies risk before failures occur
- PLC Troubleshooting: Complete Guide — practical fault-finding methods for PLC and control system faults, including how to read event logs and alarm histories
