Ishikawa Fishbone &
5-Why Analysis

Two small failures, two different lessons

1 — Root cause is not the symptom

A symptom is the detected nonconformity: three 28-day cubes average below the applicable acceptance requirement, honeycombing appears after stripping, or a calibration is overdue. A contributing cause increases the likelihood or severity but may not create the event alone. A root cause is the deepest evidenced cause within the organisation's reasonable control whose removal materially reduces recurrence.

The definition must remain practical. Investigators can descend indefinitely—“why was the procedure absent?” eventually reaches company history. Stop when the team reaches a controllable system condition, evidence supports the link, and a countermeasure can be assigned and tested. There may be more than one root cause.

This matters in NCR/CAPA because correction, corrective action and verification are different. Quarantining affected cubes or reviewing a member is containment. Recalibrating a dispenser is correction. Changing the calibration-control process is corrective action. Checking subsequent batches proves effectiveness. IS 456 governs concrete production and acceptance; it does not prescribe fishbone or 5-Why. IS/ISO 9001:2015 Cl 10.2 requires the organisation to react to nonconformity, evaluate causes, implement action and review effectiveness.

2 — Ishikawa's 6M: widen the search before narrowing it

Kaoru Ishikawa's cause-and-effect diagram, published in his 1968 Guide to Quality Control, places one defined effect at the fish's head and potential causes on branching bones. Construction teams commonly use six prompts:

A 4M version uses Man, Machine, Material and Method for simple production problems. Eight-M variants add prompts such as Management and Maintenance, or Mother Nature and Money; names vary by industry. Choose labels that stimulate relevant questions, state your chosen definition, and do not force every branch to contain a cause.

3 — How to draw a fishbone that can be investigated

Write a specific effect on the right: include member or process, batch or sample identity, date, value and requirement. “Cube failure” is weak; “28-day set C-214 from M30 slab pour S4 averaged 24.8 MPa; acceptance review required” is investigable. Draw the spine pointing to that effect. Add category bones on the left, then sub-causes beneath each category.

Phrase each sub-cause so evidence could confirm or reject it: “sand moisture correction not updated after rain” is testable; “bad materials” is not. Ask “what else?” before debating probability. Then mark each item: ✓ confirmed by a record/test, ? plausible but unverified, or × contradicted. Link evidence IDs—batch ticket BT-214, curing log CL-07, calibration certificate CTM-03—rather than relying on memory.

The diagram is not complete merely because it looks full. It is complete enough when the cross-functional team has considered relevant process families, recorded competing explanations and assigned evidence actions. Preserve rejected candidates: traceability explains why the team did not pursue them.

4 — The 5-Why method: follow one causal chain

The 5-Why method is associated with the Toyota Production System and its practice of repeatedly asking why a problem occurred until the process cause becomes actionable. Each answer must causally explain the statement before it. Convert opinions into checks: interview the person, inspect the item, review the timestamp, reproduce the measurement or compare unaffected batches.

Five is a guideline, not a quota. Three questions may reach an evidenced control failure; seven may be needed where procurement, maintenance and production controls interact. Stop when the cause is controllable, supported and countermeasure-ready—not when the form has five rows. If one “why” has two independent answers, branch into two chains. Do not compress both into a convenient story.

5 — Use fishbone and 5-Why together

The tools solve opposite errors. A 5-Why chain started too early creates tunnel vision: the first plausible story determines every later answer. A fishbone used alone creates a wall of possibilities with no depth. First use the fishbone to identify credible candidates across the process. Then prioritise candidates by evidence, consequence and plausibility, and apply a separate 5-Why chain to each material branch.

Return to the fishbone when a chain branches or evidence contradicts it. The final RCA may contain two linked roots—for example, an untriggered moisture retest in production and inadequate cube compaction in sampling. The CAPA must address each confirmed path; one countermeasure cannot close an unrelated cause.

6 — Common failure modes

• Stopping at the first cause: “excess water” describes a mechanism; it does not explain why the process permitted it.
• Blaming a person: replace “technician careless” with questions about competence evidence, workload, instructions, supervision and error-proofing.
• No evidence: a neat chain without records is a hypothesis. Mark uncertainty and collect data before approval.
• One straight chain for a multi-causal event: branch where independent conditions combined.
• Predetermined answer: do not reverse-engineer the diagram to support a contractual position.
• No countermeasure or verification: “toolbox talk conducted” rarely changes a control. Define owner, due date, changed control, measure and review date.

7 — Worked example: a short cube result, end to end

Effect. Hypothetical M30 slab pour S4, batch-linked sample C-214: three 28-day cube results of 24.3, 24.8 and 25.3 MPa, average 24.8 MPa. This result triggers review against the applicable IS 456 acceptance criteria and project specification; it does not by itself prove that the structure has the same strength or identify a cause. The responsible engineer controls any structural assessment.

Containment. Raise the NCR; preserve cubes, sample and test records; identify the member and related batches; prevent unauthorised repair or loading decisions; inform the designer/PMC; review consecutive sample results. Verify specimen identity, age, dimensions, capping/bearing faces, curing and compression-machine records before blaming production.

Evidence review. Cube IDs match the register and batch ticket. Curing log remains 27 ± 2 °C; mould check and compression-machine calibration are current; companion results are internally consistent. Rain was logged before batching. A retained sand sample shows free moisture materially above the value entered in the batch controller. The approved procedure requires one check at shift start but no retest after rain.

5-Why chain. Why was strength short? Evidence indicates higher effective water-cement ratio. Why? Aggregate free water was understated. Why? The moisture value was not updated after rain. Why? The shift followed only its scheduled morning test. Why? The procedure contained no event-triggered moisture retest. Root cause: moisture-control procedure and batch-release control did not require retesting after weather or stockpile change. A separate chain remains open if site-added water or specimen preparation evidence appears.

CAPA. Correction: determine affected batches, update moisture correction and confirm current production. Corrective action: revise the SOP and batch-release checklist to trigger testing after rain, stockpile turnover and observed moisture change; train relevant roles; record controller updates. Verification: review trigger compliance and moisture-versus-batch data over a defined subsequent production period, plus strength trends when mature. Closure requires evidence that the control operated and recurrence reduced—not merely a signed attendance sheet.

8 — Run a 30-minute site-team session

1. 0–5 minutes — define and contain. State the requirement and observed fact on one line. Confirm safety/structural escalation and containment. Ban blame language.
2. 5–12 minutes — map 6M. QC facilitates; site, RMC/production, laboratory and supervision contribute. Add candidates silently first, then clarify.
3. 12–18 minutes — mark evidence. Apply ✓, ? or × and cite record IDs. Assign owners and times for missing evidence.
4. 18–25 minutes — deepen. Run 5-Why chains on the top evidenced candidates. Branch whenever two causes are independent.
5. 25–30 minutes — control. Agree containment, provisional cause status, actions, owner, target date and effectiveness measure. Do not force closure in the meeting.

Attach the dated fishbone, 5-Why sheet and evidence index to the NCR/CAPA. The record should contain: exact nonconformity and requirement; scope and containment; team and date; cause candidates and evidence status; confirmed root/contributing causes; corrections and corrective actions; owners and target dates; effectiveness criteria and review date; approvals and residual open questions. BS EN 62740:2015 provides international guidance on root-cause analysis and traceability of the reasoning path.

Practical field blocks

Questionnaire — five questions

1. Is the effect written with requirement, observed value, date, location and traceable ID?
2. What direct evidence supports or contradicts each high-priority branch?
3. Have production, sampling, curing and testing causes all been considered?
4. Does each “why” causally explain the previous answer without changing the subject?
5. Will removing the stated root cause materially reduce recurrence, and how will that be measured?

Checklist — five records/actions

1. Preserve batch tickets, test sheets, photos, samples, instrument and curing records.
2. Record participants, timestamps, evidence IDs and status marks on the fishbone.
3. Separate containment, correction, corrective action and effectiveness verification.
4. Name an owner and due date for each evidence request and countermeasure.
5. Obtain technical approval and close only after effectiveness evidence is reviewed.

What happens if…

References & further reading

1. IS 456:2000, Plain and Reinforced Concrete — Code of Practice, Bureau of Indian Standards. Concrete production and acceptance context; it does not prescribe an RCA method. Verify the current controlled edition and amendments.
2. IS/ISO 9001:2015, Cl 10.2, Quality Management Systems — Requirements: nonconformity and corrective action, including cause evaluation and effectiveness review.
3. Kaoru Ishikawa (1968), Guide to Quality Control, Asian Productivity Organization. Foundational cause-and-effect diagram reference.
4. Taiichi Ohno, Toyota Production System: Beyond Large-Scale Production. International reading on repeatedly asking why within the Toyota Production System.
5. AIAG & VDA (2019), FMEA Handbook, First Edition. International reading for structured failure analysis.
6. ANSI/ASQ Z1.1, quality-assurance terminology and related ANSI quality-management publications. Consult the current catalogue alongside the AIAG/VDA FMEA Handbook.
7. J. M. Juran (1988), Juran's Quality Control Handbook, Fourth Edition. Diagnosis and quality-improvement reading.
8. BS EN 62740:2015, Root Cause Analysis. International guidance on RCA process, evidence and traceability.

Frequently Asked Questions

Related articles