The instinctive answer is the first one, because it is the fastest, and it is the most expensive when the "culprit" is a long-standing team member or a relative: in Italy, 80.9% of companies with at least 3 employees are controlled by an individual or a family [5].
A professional with two team members, a twelve-person workshop and a hundred-employee company all react the same way: fix the effect, reprimand someone, move on.
The Ishikawa diagram — also called the cause-and-effect or fishbone diagram — was born in Japanese quality control to do the opposite: pin the effect at the head and sort the causes into families before taking action [1].
An Ishikawa diagram does not find the cause: it forces you to stop fixing the effect and to list the verifiable causes, by family, before choosing where to act.
This article shows how to build one in six steps, the 4M-6M categories translated from the shop floor to the office, a filled-in example and the mistakes that reduce it to a poster.
Stop fixing the effect: what a cause-and-effect diagram is and how it differs from other tools
A half-hour meeting about yesterday's complaint, or a sheet with six branches: which of the two keeps the same complaint from coming back next month?
The meeting finds someone responsible in a few minutes; Deming, based on his own experience and with no statistical claim, estimated that 94% of problems belong to the system and only 6% to special causes [7].
Without a method, a recurring problem gets handled the same way every time: fix the effect, look for whoever made the mistake, close the meeting with a reminder.
The Ishikawa diagram — also called the cause-and-effect diagram or fishbone diagram — was born in Japanese quality control to do the opposite: pin the effect at the head and sort the causes into families before deciding [1].
This section defines the tool, distinguishes it from its close neighbors (5 whys, root cause analysis, flowchart, Pareto) and shows why the fishbone shape matters less than the question that generates it.
In Ishikawa's 1976 manual, the diagram is one of the seven tools of quality control [1].
The context is the quality circles that Treccani associates with Ishikawa [2], and the literature credits him with pioneering the diagram in the 1960s [8].
Ishikawa is the author, cause and effect is the function, fishbone is the shape: a head with the effect, a spine, one branch per family of causes, sub-branches for individual causes (short definition in the glossary).
The 5 whys digs deep into a single causal chain, while the fishbone opens several families across the width: they are complementary, the first applied to the sub-branches of the second.
Root cause analysis is the family of methods to which the diagram belongs as one technique.
A flowchart shows the sequence of activities, not the causes; the Pareto chart, one of the same seven tools [1], ranks defects that have already been counted, while the fishbone generates the hypotheses to count.
The sheet answers the opening question: the effect written at the head and the question "why does it keep coming back?" on every branch force you to list verifiable causes, not a person to blame.
What a fishbone diagram is for, and when another method is a better fit
A problem that has been recurring for months, and a breakdown that happened only once: which of the two deserves a fishbone?
Only the first: on an isolated event the diagram produces thirty hypotheses and zero checks, and the same literature that extends its use to risk ties it to events with multiple causes [8].
The diagram does not produce the answer: it produces an ordered list of questions to verify, which is why it only helps where there is more than one plausible cause and the effect keeps recurring.
For organizations with a certified quality system — in Italy there were 101,426 active ISO 9001 certificates at the end of 2024 [4] — analyzing the causes of a nonconformity is an explicit requirement [3]; for micro-enterprises, which make up 78.9% of Italian companies with at least 3 employees [5], it is a choice of method.
This section places the tool in the comparison table of the guide to business problem solving and explains when a 5 whys, a flowchart or simple data collection is a better choice.
Three conditions tell you whether the effect deserves the diagram.
- recurring effect: it has shown up at least twice in the same period;
- measurable effect: it can be written as a number, such as 9 orders out of 40;
- more than one plausible cause: different people point to different causes.
If one is missing, something else works better: a single chain that is already known is dug out with the 5 whys, the time lost along a process shows up in a flowchart, and the defect that weighs the most is found by counting defects.
Choosing between alternative interventions is a matter of decision-making, outside the diagram.
Its field of use goes beyond product quality: the literature extends it to complaints, delays and administrative errors [8].
Once the three conditions are met, what remains is choosing the categories for the branches.
Choosing the categories: the 4M, 5M and 6M translated for the shop floor and the office
The six textbook categories — machines, materials, methods, manpower, measurements, environment — brought into an administrative office: do you keep them as they are or change them?
You keep the questions and change the words: in the office the "machine" is the management software and the "material" is a missing piece of information, and under Manpower the most frequent cause is who is missing, not who makes mistakes — in 2024, 47.8% of the hires planned by Italian companies were hard to fill [6].
The four classic families of the diagram were born in the factory, in line quality control [1]: in common practice they read as Manpower, Methods, Machines, Materials, with Measurements and Environment added in the five- and six-category versions.
Under Manpower, in many companies, the cause is not "who made the mistake" but "who is missing": in Italian industry, hiring difficulties affect on average 55% of planned hires [6].
This section offers two parallel translations of the categories — one for the shop floor, one for the office — with a guiding question for each, so that a business owner who moves from production to the desk can use the same template in both places.
| Category (4M/5M/6M) | On the shop floor (guiding question) | In the office (guiding question) |
|---|---|---|
| Manpower | Who does the work, who knows how to do it, who is missing? | Who handles the file, who knows how to handle it, who is missing? |
| Methods | How is it written that it should be done, how is it really done? | What is the written path from quote to delivery note, and what is the real one? |
| Machines | What equipment, what downtime? | Management software, spreadsheet, printer: what gets stuck? |
| Materials | What components, what suppliers? | What information, documents, master data are missing? |
| Measurements | What instruments, what tolerances? | What gets recorded, where, by whom? |
| Environment | Space, temperature, seasonality? | Workload peaks, interruptions, vacations? |
In manuals, Environment also appears as Mother Nature or Milieu, and Management is sometimes added as a seventh item.
In family-controlled companies, which account for 81.2% of manufacturing companies in Italy [5], Manpower also includes the business owner and their children, and the template lets you write "the business owner promises the date verbally" as a fact, not as an accusation.
Four categories are enough in a three-person office, six are needed on the shop floor; you rename them, you do not multiply them beyond six or seven, because the diagram has to fit on a letter-size sheet or a whiteboard.
Once the categories are chosen, what remains is filling it in, in six steps.
How to make an Ishikawa diagram: the six-step procedure
Filled in by the business owner at the desk in twenty minutes, or in an hour with the people who do the work: which of the two diagrams holds up when checked against the data?
The second, and not because of participation: Ishikawa designed the tool for supervisors and line staff, to be used with data collected on site [1].
The procedure takes six steps: write the effect with a number; choose the categories; collect the causes with the people who do the work — the cross-functional group "from manager to worker" that Treccani describes for quality circles [2]; dig into the sub-branches with the question "why"; mark which causes are verified with data and which remain hypotheses; choose two or three causes and turn them into corrective actions whose effectiveness you will check [3].
The template used is a sheet with the effect at the head and six branches, one per category: the same one that appears filled in in the example section.
- Write the effect with a number and a period. "Deliveries past the promised date: 9 orders out of 40 in July" goes in the head; "poor quality" is a judgment, not an effect. Done right when anyone can recount that number the following month.
- Choose the categories. Take them from the previous section, four in the office and six on the shop floor, written at the top of the branches. Done right when each branch carries its guiding question.
- Collect the causes with the people who do the work. Thirty to forty-five minutes in front of the whiteboard, one cause per sticky note; judgments wait until the branch is complete. On the shop floor a physical whiteboard beats an app: you fill it in in front of the people who know. Done right when every branch has a cause written by someone who does that work.
- Dig into the sub-branches with the question "why". For each cause, ask why until you reach a fact you can act on: "the supplier delivers late" becomes "the order goes out after the down payment". On the sub-branches the complementary technique is the 5 whys. Done right when the last sub-branch is a fact that can be changed, not a person.
- Mark what is verified and what remains a hypothesis. Next to each cause, a letter: V if data confirms it, I if it remains a hypothesis, S if the data rules it out; the data comes from complaints, delivery notes and record sheets [1]. Causes marked V get a weight for probability and impact, even just high, medium or low [8]. Done right when every V also carries a number.
- Choose two or three causes and turn them into corrective actions. For each one, an action, an owner and a date on which to check its effectiveness [3]. An action that changes the way work is done becomes an operating procedure, and the check closes the continuous improvement cycle. Done right when the check date is already on the calendar.
The diagram filled in with the people who do the work holds up, because every V has a number taken from the delivery notes next to it, not an opinion.
Reading a filled-in example: the recurring complaint in a 15-person family business
Before the diagram the complaint was "the supplier's fault"; afterward, three verified causes across six branches: which of the two readings leads to an intervention you can check a month from now?
The second — in the example the heaviest cause was not the supplier but the moment the delivery date gets promised; and checking the effectiveness of the corrective action is what the quality standard requires of certified organizations [3], though you do not need a certification to do it.
A hypothetical case built for teaching purposes: a 15-person family business that makes custom windows and doors, with the business owner alternating between the shop floor and the office and a daughter in administration.
The recurring complaint is delivery past the promised date: 9 orders out of 40 in one month.
This section shows the filled-in template branch by branch, separates the causes verified on the 9 orders from those that remained hypotheses, and arrives at two corrective actions that become procedures.
| Branch | Causes raised | Verification result (V/I/S) |
|---|---|---|
| Manpower | A single outside installer also does the site surveys; "the new guy on the floor is slow" | V (2 out of 9, measurements confirmed late); S (none of the 9 orders go through him) |
| Methods | Date promised verbally before the supplier confirms the sample colors; supplier order placed after the down payment | V (6 out of 9); I (to be verified the following month) |
| Machines | Miter saw down for one day | V (1 out of 9) |
| Materials | Sample profiles: 3 weeks from the supplier versus the 2 taken for granted | V (linked to Methods) |
| Measurements | Promised date in the business owner's calendar and in the production sheet, not matching | V (different dates in 5 out of 9) |
| Environment | Seasonal peak and the installer's vacation | I |
There are five Vs, but weighted for probability and impact, three causes remain to act on: the verbal promise (6 out of 9), dates that do not match (5 out of 9), the installer who also does site surveys (2 out of 9).
The other two Vs matter less or depend on the first ones: the miter saw accounts for 1 order out of 9, and the supplier is late only relative to a date promised before its confirmation.
There are two corrective actions, each with an owner and a check date, as clause 10.2 of the standard requires [3].
The first: the date is promised only after the supplier's written confirmation on special colors (the daughter, check at the end of the month).
The second: a single delivery calendar shared by the office and the shop floor, with site surveys in a fixed weekly time slot (the business owner, same date).
Both become operating procedures.
In a month, delays are recounted on the next 40 orders: that is the verification cycle, not a promise of results.
Common mistakes that empty out a fishbone diagram, and how to avoid them
A diagram full of branches, or a diagram with five verified causes: why is the first one shown off with pride while the second one changes procedures?
Because completeness is the trap of this tool: without probability and impact assigned to the causes, the diagram remains an inventory [8].
The mistakes that empty out a fishbone look alike from one company to the next: the effect written vaguely, the diagram filled in by one person alone, causes confused with symptoms or with interventions already decided ("we need software"), factory labels copied into the office, the verification skipped, the analysis closed without an action.
This section puts them in a table with the signal that lets you spot them in advance and the correction next to it, before an hour of meeting time has been spent.
| Mistake | How to spot it | Correction |
|---|---|---|
| Vague effect, no number | "Poor quality", "dissatisfied customers" | Rewrite it with data and a period |
| Filling it in alone | The diagram is ready before the meeting | A session with the people who do the work, a cross-functional group [2] |
| Cause confused with symptom or intervention | "We need software", "we could use one more person" | Rephrase as an observable fact |
| Factory labels in the office | Empty "Machines" branches in administration | Translate the categories for the office |
| Thirty causes with the same weight | No marks or numbers next to the causes | Mark V/I/S, weigh probability and impact [8] |
| Stopping at the diagram | Sheet pinned to the wall, no owner | Step 6: action, owner, check date [3] |
| Diagram as an indictment | Names appear on the branches, not facts | Write facts; look for the staffing constraint first |
The last mistake costs the most in family businesses, the vast majority of companies in Italy [5]: the names of a long-standing team member or of a son or daughter end up on the branches.
Under Manpower, the hiring difficulty reported to Unioncamere, the Italian chambers of commerce system [6], is a staffing constraint, which no reprimand will fix.
Seven checks do not add branches: they remove from the diagram what turns it into a poster.
Limitations and conditions of applicability
The diagram generates hypotheses and does not verify them: without step 5 it remains an inventory of possible causes, which is why the literature proposes weighting them for probability and impact [8].
A cause marked V is confirmed by data on the 9 orders, not proven to be the only cause: the effectiveness check after the corrective action is the test that closes the loop.
Deming's 94/6 estimate, mentioned at the beginning of the article, is stated by its author as the result of experience, not of measurement, and should be read as guidance, not as data [7].
The original sources [1] and [8] come from industrial and product-quality contexts: translating the categories for the office is a proposal by the editorial team, not a finding from the literature.
The windows-and-doors company case is hypothetical and for teaching purposes: the numbers serve to show the method, not to estimate results.
The cause-analysis requirement applies to certified organizations [3]; for other companies the diagram is a choice of method, with no obligation and no prescribed format.
FAQ
What is an Ishikawa diagram used for?
It is used to sort into families the possible causes of an effect that recurs and can be measured, so you can choose which ones to verify with data before acting; it is one of the seven tools of quality control [1].
What is the difference between an Ishikawa diagram and the 5 whys?
The Ishikawa diagram opens several families of causes across the width, while the 5 whys digs deep into a single causal chain: they are used together, the second on the sub-branches of the first.
How many categories should you use: 4M, 5M or 6M?
Four categories are enough in an office of a few people, six on the shop floor; the 5M add Measurements and the 6M add Environment, and the labels can be renamed as long as the diagram fits on one sheet.
Can you make an Ishikawa diagram in Excel or PowerPoint?
Yes: a spreadsheet or a presentation program (as examples of the category and without recommendation: Excel, Google Sheets or LibreOffice Calc; PowerPoint, Google Slides or Keynote) is enough for the copy you file, but filling it in works better in a session, on a whiteboard or a letter-size sheet, in front of the people who do the work.
Key takeaways
An Ishikawa diagram starts from an effect written with a number and a period, not from a judgment.
The categories on the branches are chosen from the 4M, 5M or 6M and translated into the vocabulary of the place where the effect occurs, shop floor or office.
Causes are collected in a session with the people who do the work, one per sticky note, holding judgments until the branch is complete.
Each cause is dug into with the question "why" until you reach a fact you can act on.
Each cause gets a letter, V, I or S, and verified causes get a weight for probability and impact.
Two or three causes become corrective actions with an owner and a check date, and an action that changes the way work is done becomes a procedure.
The diagram is useful when the effect is recurring, measurable and has more than one plausible cause; in other cases another method works better.
Conclusion
The Ishikawa diagram is not a drawing to hang on the wall: it is the moment when a company stops fixing the effect and lines up the verifiable causes before choosing where to act.
The fishbone in the example works for a 15-person window and door maker just as it does for a firm with two team members: the words on the branches change, the question does not.
The comparison table of methods is in the guide to business problem solving.
The cycle that turns corrective actions into habit is in the article on continuous improvement in business.
A company that fills in a fishbone when a complaint comes back for the second time builds, month after month, a map of its own recurring causes.
Meetings about "who did it" give way to meetings about "what do we verify", long-standing team members stop feeling accused, and the date promised to the customer becomes a real date again.
Sources and references
[1] Ishikawa, K., "Guide to Quality Control", Asian Productivity Organization, Tokyo, 1976 (2nd revised ed.; revised translation of "Genba no QC Shuho"). Available at: https://archive.org/details/guidetoqualityco00ishi
[2] Treccani, "Qualità, circoli della", Dizionario di Economia e Finanza, Istituto della Enciclopedia Italiana, 2012. Available at: https://www.treccani.it/enciclopedia/circoli-della-qualita_(Dizionario-di-Economia-e-Finanza)/
[3] UNI, "UNI EN ISO 9001:2015+A1:2024 — Sistemi di gestione per la qualità — Requisiti", Italian standards body (Ente Italiano di Normazione), in force since October 16, 2024. Available at: https://store.uni.com/uni-en-iso-9001-2015-a1-2024
[4] Accredia, "ISO Survey 2024: le imprese accelerano su sostenibilità e digitalizzazione", AccrediaHub, 2025 (data as of December 31, 2024). Available at: https://www.accredia.it/comunicazione/accrediahub/iso-survey-2024-le-imprese-accelerano-su-sostenibilita-e-digitalizzazione/
[5] ISTAT, "Censimento permanente delle imprese 2023: primi risultati", press release, November 14, 2023 (reference year 2022; about 280,000 responding companies, representative of 1,021,618 companies with at least 3 employees). Available at: https://www.istat.it/comunicato-stampa/censimento-permanente-delle-imprese-2023-primi-risultati/ — PDF: https://www.istat.it/it/files/2023/11/REPORTCensimprese.pdf
[6] Unioncamere – Ministero del Lavoro e delle Politiche Sociali, Sistema Informativo Excelsior, "La domanda di professioni e di formazione delle imprese italiane nel 2024", 2025. Available at: https://excelsior.unioncamere.net/pubblicazioni/2024/domanda-di-professioni-e-di-formazione-delle-imprese-italiane-nel-2024 — PDF: https://excelsior.unioncamere.net/sites/default/files/pubblicazioni/2024/Domanda_professioni_formazione_imprese.pdf
[7] Deming, W. E., "Out of the Crisis", MIT Press, Cambridge (MA), 1986 (2nd ed. 2000, p. 270). Estimate stated by the author as the result of experience; quotation verified on the W. Edwards Deming Institute page: https://deming.org/quotes/i-should-estimate-that-in-my-experience-most-troubles-and-most-possibilities-for-improvement-add-up-to-the-proportions-something-like-this94-belongs-to-the-system-responsibility-of-management6-sp-3/
[8] Ilie, G., Ciocoiu, C. N., "Application of fishbone diagram to determine the risk of an event with multiple causes", Management Research and Practice, vol. 2, no. 1, 2010, pp. 1-20 (Academy of Economic Studies, Bucharest). Available at: https://econpapers.repec.org/RePEc:rom:mrpase:v:2:y:2010:i:1:p:1-20 — PDF: https://mrp.ase.ro/no21/f1.pdf