8D (Eight Disciplines) is a structured problem-solving method for customer complaints and major nonconformances. The report follows eight steps: form a team (D1), describe the problem in measurable terms (D2), contain it (D3), identify and verify the root cause (D4), choose and confirm permanent corrective actions (D5), implement and validate them (D6), prevent recurrence (D7), and close the case and recognize the team (D8). Some companies add a D0 step in front to decide whether the issue needs a full 8D at all.
The report has two audiences. Your internal team uses it to stay aligned, and your customer uses it to judge whether you have really found the cause. One quality society’s page calls 8D a communication tool that ties actions to root causes. In practice, the step where reports most often get stuck is D4: causes are listed but not verified, and the report never answers why your own inspection did not catch the problem.
- Decide first, then form the team. Do a full 8D when the customer requires one. Without that requirement, do a full 8D when the cause is unclear, several departments are involved, or the issue is a customer complaint or major deviation (Stage 1).
- Make D2 measurable, and contain in D3 without waiting for the root cause. Describe the problem with 5W2H and attach photos and physical evidence. Stock at the customer, in transit, in the warehouse and on the line each needs its own isolation and screening (Stage 2).
- Verify the D4 root cause and find the escape point. Listing causes is just the first step. Verify by reproducing the defect or removing the cause, and ask why the control that should have caught it did not (Stage 3).
- Back D5 to D8 with evidence. Confirm actions with data first, validate them after launch using the same yardstick, state when interim actions end, then roll out to similar areas, update documents and close (Stages 4 and 5).
This guide starts when a customer complaint arrives and you owe an 8D reply, and ends when the actions are validated, documents updated and the case filed. For choosing an analysis method once you have the failed parts, see “How to Inspect BGA Solder Joints: X-ray, Dye-and-Pry or Cross-Section, in 5 Stages.” For the improvement evidence auditors ask for, see “What an SMT Process Audit Asks: 6 Stages of Line Documents and Records, Starting at Solder Paste Printing.” Customer-specific 8D form fields, the full corrective and preventive action (CAPA) process, design of experiments, process capability calculations and software are outside this guide. Each stage below covers how to do it and what public sources say; Stages 2 to 5 also list how JIAHON can help. Sources, exceptions and terms are in the notes at the end.
Figure 1. The five stages of 8D. (1) Decide and form the team: decide whether a full 8D is needed; form the team and name a leader and contact; confirm the customer’s format and deadline. (2) Describe the problem and contain it: write the five Ws and two Hs; compare “is” and “is not”; isolate everything from the customer back to the line — JIAHON can help with microscopes, magnifiers and screening labels. (3) Find and verify the root cause: fishbone diagram and 5 Whys; verify by reproducing or removing; find the escape point — ionic contamination testers, stencil tension meters and inspection machines. (4) Choose, implement and validate actions: compare options and confirm with data; validate with the same yardstick; state when interim actions end — microscopes (before and after), ionic contamination testers. (5) Prevent recurrence and close: roll out to similar areas with a list; update documents and close the escape point; close and file the lessons — traceability labels.
Stage 1: Decide whether a full 8D is needed, form the team, and confirm the customer’s format and deadline (D0, D1)
Decide whether to run a full 8D before deciding who runs it. The customer has usually already set the format and the deadline.
| Decision | A reasonable approach | What public sources say |
|---|---|---|
| Full 8D or not | Follow the customer if they require it. Without that requirement, run a full 8D when the cause is not obvious, several departments are involved, or it is a customer complaint or major deviation. Small issues with a clear cause can go through normal corrective action. | A quality society’s 8D page: especially useful when the root cause is not obvious and the problem crosses departments; commonly used for complaints, recalls and quality deviations. A public online encyclopedia page: in Global 8D, D0 uses assessment questions to decide whether a full 8D is needed, so the team’s effort goes to problems that warrant it. |
| Team | Members from process, quality, equipment and incoming materials; add customer or supplier representatives when they are involved. Name one leader and one customer contact. | 8D page: members need product or process knowledge. Encyclopedia page: members bring different perspectives. The society’s root cause analysis page: choose people from the affected processes, adding managers with decision authority, internal customers and quality improvement specialists as needed. |
| Format and deadline | Ask the customer for their required form and reply deadline, and follow theirs. | Neither public page discusses reply deadlines. A common line practice: the 8D form has a date field at every step, reply deadlines printed beside D3, D4 and D6, and sign-off fields at the end. |
How to do it:
- Check whether the customer has a required 8D form. If so, fill in their fields; do not switch to your own format.
- Write the deadline at the top and plan internal milestones backward from it: D3 containment is due first, and D4 root cause takes time.
- Have one leader and one customer contact. Everyone else focuses on finding the cause rather than answering the customer’s emails.
- Preserve the evidence: seal the failed parts, good parts from the same lot, and the records and parameters from the time, before anyone touches them.
- A small issue whose cause is obvious (an evident input error, for example) does not need a forced full 8D, but still needs a corrective action record.
What public sources say:
- See the table above.
- Encyclopedia page: 8D was developed by a US automaker in the 1980s, with a manual published in 1987; Global 8D in the late 1990s added D0 and the escape point concept. The page also notes that many publications attribute it to the US military, and that this history is disputed.
- Encyclopedia page: effective use of 8D requires training in 8D and in data collection and analysis tools.
Stage 2: Describe the problem in measurable terms (D2) and contain it (D3)
D2 makes sure the team and the customer are looking at the same thing. D3 stops the problem from reaching the customer without waiting for the root cause.
| 5W2H | What to write | Example: discolored solder joint complaint (illustration) |
|---|---|---|
| Who | Who found it, who is affected | The customer’s incoming inspection |
| What | The symptom and how it differs from spec | Discoloration at the edge of solder joints, visible under a magnifier |
| Where | Which station, where on the board, which model | Same model, concentrated on one side of the board |
| When | When it was found; matching production dates and lot numbers | After certain date codes |
| Why | Why it was judged a defect (which spec or drawing) | Visual requirements on the customer drawing |
| How found | Which inspection caught it | Visual check at incoming inspection |
| How many | Quantity, rate and number of lots | Sample size and defect count, not just “a lot” |
Write interim containment separately for each place defective units might be:
| Where defects might be | Interim containment | Evidence to keep |
|---|---|---|
| At the customer (shipped or in use) | Help the customer screen and remove defects; send replacement units if needed so they can keep producing | Screened quantity, rejected quantity, replacement records |
| In transit and in the warehouse | Hold and isolate everything; release after re-inspection | Quarantine area marking, count records |
| Work in progress on the line | Tighten or 100% inspect; switch machines or move to manual work if needed | Inspection records, machine change records |
| Incoming material and suppliers | Notify the supplier and ask them to screen or replace the same lot | Notice and reply records |
How to do it:
- D2: attach photos and evidence. Place defective and good parts side by side at the same angle and magnification. The clearer the description, the faster the team gets up to speed.
- D2: compare “is” and “is not.” Which lots have it and which do not; which line; which location. The differences are your leads for D4.
- D2: do not state causes. “Insufficient cleaning caused discoloration” is a cause; save it for D4.
- D2: prepare a process flow chart so people who do not know the process can see which station the problem occurred at.
- D2: visual defects need an acceptance basis. For reading acceptance criteria, see JIAHON’s article “J-STD-001 vs. IPC-A-610: Process Standard and Inspection Standard Compared.”
- D3: contain the problem whether or not you have found the root cause, and submit it within the customer’s deadline.
- D3: for each interim action, state what is done, who does it, until what date or condition, and how you will show it works.
- D3: screened goods must be identifiable — label them, repack them, or keep them in a separate area. Unmarked goods leave the next person unable to tell.
- D3: interim actions have costs and risks too (100% inspection can miss parts; manual work brings new errors). Stop them once D6 confirms the permanent action works, and record that in the report.
What public sources say:
- Quality society 8D page: D2 uses 5W2H (who, what, where, when, why, how, how many) to quantify the problem; D3 defines and carries out actions to isolate the problem.
- Encyclopedia page: D2 defines the problem in measurable terms, often with an “Is / Is Not” worksheet; whether interim containment is needed depends on where in the product life cycle the problem occurs.
How JIAHON can help:
- Photographing evidence and explaining defects: a 4K 3D digital microscope (its page lists a built-in measurement system and on-screen viewing, and notes that engineers and line supervisors explain defects by working at the screen), a 40x handheld magnifier (folding, with built-in LED lighting), and an LED magnifier lamp with a long arm (10x, three light colors, adjustable brightness).
- Labeling screened goods: custom adhesive labels (size, material and printing made to order).
- For pricing and availability, please contact us.
Stage 3: Find and verify the root cause (D4)
D4 is where reports get stuck. Listing possible causes is easy; the hard part is proving which one is real, and why the problem got past your own inspection.
Figure 2. Two places a problem lives (illustration). The flow runs from incoming material, into the line, through the process station where the problem is created, past the pre-shipment check that should have caught it and did not, and on to the customer, who receives the defective part. The occurrence point is the station where the root cause sits; it answers “why was it created?” — list possible causes (fishbone, 5 Whys), verify them one by one by reproducing or removing the cause, and keep the evidence (photos, data, test conditions). Action 1: stop it from being created. The escape point is the control that should have stopped it; it answers “why was it not caught?” — was the sampling rate enough, was the item included in the check, were the criteria clear and could the measurement method see it. Action 2: add detection. Each location needs its own action.
| Tool | Where it is used | What public sources say |
|---|---|---|
| Fishbone (cause-and-effect) diagram | Sort possible causes by people, machines, materials, methods, measurement and environment, and review each | Quality society fishbone page: categories are usually the 6Ms (materials, machines, methods, measurement, people, environment; a seventh M is cost), and they can be renamed; a cause may appear in several categories; for complex problems the fishbone may be one tool among several, needing other tools and more data; build it with people who know the area. |
| 5 Whys | Follow one cause down, layer by layer | Quality society 5 Whys page: keep asking why until you reach the root cause, which may take fewer or more than five; can be paired with a fishbone; when something goes wrong, shift from “whose fault?” to “what do we change?” |
| Change and barrier review | What changed in people, machines, materials, methods or environment before and after the problem; which control meant to prevent or detect it failed | Quality society root cause analysis page: change analysis applies when system performance changes noticeably; barrier analysis looks at the controls meant to prevent or detect problems; the page also says root cause analysis pays off when it sits inside a broader problem-solving effort. |
Once causes are listed, verify them one at a time:
| Verification | How | What to look for |
|---|---|---|
| Reproduce | Deliberately put the suspected cause into the process and run test pieces | Whether the same defect appears |
| Remove | Take the cause away (put back a good tool, replace suspect material) | Whether the defect disappears, and returns when the cause is put back |
| Compare | Compare affected and unaffected lots, lines and locations item by item | Whether the difference appears just on the affected side |
| Check records | Match the start of the defect against change records | Whether it started after a change |
How to do it:
- Lay out the fishbone, then use the D2 “is / is not” comparison to strike out causes that do not fit.
- Verify one at a time. Doing both — reproducing the defect, and seeing it disappear when the cause is removed — is more reliable than doing one.
- Keep the verification evidence: photos, data, test conditions. If the report just says “analysis shows cause X,” the customer cannot see how you know.
- The cause of occurrence and the cause of escape are often two different things; write them separately. The first is why the problem was created; the second is why it was not stopped.
- Find the escape point: starting from the station where the problem was created, look downstream for the inspection or control that should have caught it, and why it did not (sampling rate too low, item not in the check, unclear criteria, measurement that cannot see it).
- If the evidence is not enough, write “not yet confirmed” with a verification plan and date. Do not force a cause.
- How far down should you go? Stop at the layer where fixing it means the problem will not happen again.
- The more complete your routine data (defect reports, process capability indices, repair records), the faster D4 goes. With nothing to compare, you are left guessing.
- Example: a visual defect is found after shipment. Investigation: run test pieces back through the suspect process and see whether the same defect appears. It does, so the cause is pinned to that station’s process conditions. Why it escaped: the pre-shipment visual check is a sampling inspection, and the chance of sampling this defect was too low. Two actions follow: change that station’s process conditions, and change the inspection so this kind of defect is more likely to be sampled and seen.
What public sources say:
- Quality society 8D page: D4 identifies all possible causes, including why the problem was not noticed; causes must be verified, not just brainstormed; 5 Whys and cause-and-effect diagrams are used within the 8D framework and do not replace it.
- Encyclopedia page: D4 is named “identify and verify root causes and escape points”; the escape point is the earliest control point after the root cause where the problem should have been detected; tools include 5 Whys and the fishbone.
- An article linked from the 5 Whys page notes that knowing when to stop is often overlooked.
- Example: at the user’s site, a module’s readings drift during machine testing. A similar replacement part gives normal readings; putting the first module back brings the failure back, so that part is judged faulty. Returned to the manufacturer, the production history is normal, the exterior shows just minor scratches on a metal surface, and electrical tests pass. The diagnosis is no fault found (NFF), and the manufacturer rejects the claim. An earlier return of the same part had also been NFF; the problem keeps recurring with the cause still unknown. The lesson: a manufacturer’s retest “pass” does not mean there is no problem. If the field can reproduce it and the manufacturer cannot, the conditions at the two ends may differ. Ask the user for the full conditions at the moment of failure, and have the manufacturer retest under those conditions as closely as possible.
How JIAHON can help:
- If you suspect insufficient cleaning or ionic residue: an ionic contamination tester (its page says it measures surface ion concentration with an isopropyl alcohol and deionized water extract at constant temperature, shows conductivity and ion concentration curves in real time, outputs a report when the test finishes, and has static and dynamic modes).
- If you suspect solder paste printing: a stencil tension meter (two measuring ranges, used to check whether stencil tension has dropped) and a semi-automatic stencil inspection machine (its page lists detection of 13 defect types; currently on pre-order). To work back from the print to the consumable to change, see JIAHON’s article “Troubleshooting Solder Paste Printing Defects: Work Back from the Print to the Consumable to Change.”
- If you suspect the inside of a solder joint: for choosing between X-ray, dye-and-pry and cross-section, see JIAHON’s article “How to Inspect BGA Solder Joints: X-ray, Dye-and-Pry or Cross-Section, in 5 Stages.”
- For pricing and availability, please contact us.
Stage 4: Choose permanent actions, implement and validate (D5, D6)
D5 confirms the action really works first. D6 validates it again with the same yardstick after full implementation.
| D5 item | What to write |
|---|---|
| Options | More than one (repair, replace, change the process, change inspection, and so on) |
| Cause addressed | Which D4-verified cause each option targets |
| Side effects | Whether it creates new problems |
| Cost, labor, time | Compare the options side by side |
| Confirmation | How you will show with data that it works (pilot run, trial) |
Figure 3. Interim actions, permanent actions and the follow-up period (illustration, no values). Along a time axis: the complaint arrives, interim actions begin, the permanent action goes live, and interim actions stop after an overlap period, followed by a follow-up period of several consecutive lots without recurrence. Interim actions (screening, isolation, tighter inspection) are costly and can miss parts. They push down the defects the customer sees, while the actual process defect rate (measured with the same yardstick) falls just after the permanent action goes live. If interim actions are stopped before the root cause is fixed, the defects come back.
| D6 check | How | Evidence to keep |
|---|---|---|
| Was the action done? | Confirm item by item: change orders, revised documents, training records, photos from the floor | Document numbers, dates |
| Does it work? | Compare before and after using the same measurement method and criteria as D2 | Before-and-after data or photos |
| When do interim actions stop? | Stop them once effectiveness is confirmed, and record the date | Stop date and condition |
| How long to follow up? | It counts when several consecutive lots show no recurrence, covering shift, material and lot changes | Follow-up period and the result for each lot |
How to do it:
- D5 actions target D4-verified causes. Do not write actions for causes you have not verified.
- D5: list pros, cons, cost, labor and time for each option, pick one and explain why. For prioritizing, you can use scoring similar to FMEA (Failure Mode and Effects Analysis).
- D5: confirm on a small scale before full rollout — a pilot or pre-production run — and look at the results with data.
- D6: before stopping interim actions, keep an overlap period to make sure the new action holds.
- D6: validation by reproduction is preferable: put the cause back and see whether the defect returns. Do this on test pieces, not on product that will ship.
- D6: set your own follow-up period and write it in the report; if the customer specifies one, follow theirs.
- D6: measure before-and-after photos, dimensions and cleanliness with the same equipment, magnification or conditions.
What public sources say:
- Quality society 8D page: D5 selects permanent corrective actions and confirms with data that they will resolve the problem, for example through a pre-production plan; D6 implements them and validates their effectiveness.
- Encyclopedia page: D5 confirms with a pre-production plan; D6 needs evidence of improvement; in Global 8D, actions also address the escape point.
- Neither page gives a general rule for how long, or for how many lots, to follow up.
How JIAHON can help:
- The same yardstick before and after: a 4K 3D digital microscope (built-in measurement system, per its page).
- For cleaning-related actions: an ionic contamination tester (outputs a report when the test finishes, per its page).
- For pricing and availability, please contact us.
Stage 5: Prevent recurrence (D7) and close (D8)
D7 prevents this problem and similar ones from happening again. D8 closes the case and keeps the lessons.
| D7 task | How | Example (illustration) |
|---|---|---|
| Roll out to similar areas | Which other models, lines, suppliers or stations share the same risk | Other models using the same cleaning process |
| Update documents | Work instructions, control plans, inspection specs and training materials; revise and record any change | Add a check item to the inspection spec |
| Close the escape point | Add detection or change sampling at the escape point found in D4 | Raise the sampling rate |
| Traceability | Lot numbers, labels and barcodes, so next time you can tell which lot | Put the lot label in a fixed position |
| Look at the system | Whether the cause lies in rules, human error, organization or training; each needs a different kind of action | Incomplete rules: complete the rules. Human error: add mistake-proofing and double checks. |
How to do it:
- D7: list the rollout targets and owners; do not just write “rolled out.”
- D7: after updating documents, spot-check that the floor follows them. If the documents change but the floor does not, the problem comes back.
- D7: choose the kind of action by the type of cause. A common line practice groups management causes into incomplete rules, human error, organization and structure, training, and insufficient detection of latent risk, each with its own direction (for example, create or strengthen rules when they are incomplete; add mistake-proofing and double checks for human error; revise training materials and qualification methods when training falls short).
- D8: recognize the team; sign off the report (prepared, reviewed, approved) and file it.
- D8: add what you learned to your knowledge base — what the problem looked like, how it was investigated and verified — so the next person does not start from scratch.
What public sources say:
- Quality society 8D page: D7 modifies management systems, operating systems, practices and procedures to prevent this and similar problems; D8 formally recognizes the team’s efforts.
- Encyclopedia page: D7 and D8 as above.
How JIAHON can help:
- Labels for lot traceability: custom continuous labels (three-proof thermal), described on their page as water-, oil- and scratch-resistant, with custom sizes. For choosing a barcode, see JIAHON’s article “Code 39 vs. Code 128: 4 Ways to Choose a Barcode for Electronics Factory Labels.”
- Turning investigation experience into documents: see JIAHON’s article “When Senior Technicians Retire, Does Yield Drop? Five Kinds of Process Knowledge to Document and How to Start.”
- For pricing and availability, please contact us.
Related reading
To cover the related process steps, continue with these articles:
- “How to Inspect BGA Solder Joints: X-ray, Dye-and-Pry or Cross-Section, in 5 Stages”
- “What an SMT Process Audit Asks: 6 Stages of Line Documents and Records, Starting at Solder Paste Printing”
- “Troubleshooting Solder Paste Printing Defects: Work Back from the Print to the Consumable to Change”
- “J-STD-001 vs. IPC-A-610: Process Standard and Inspection Standard Compared”
- “When Senior Technicians Retire, Does Yield Drop? Five Kinds of Process Knowledge to Document and How to Start”
Notes: sources and exceptions
These notes are for readers who want to check the sources; you can follow the steps above without them. Where your customer’s 8D form and requirements differ from this guide, theirs take precedence.
Sources for each practice
| Practice | Content | Source |
|---|---|---|
| 8D steps and when to use them | D0 to D8; when to use; 5W2H; verifying causes in D4; confirming with data in D5; D7, D8 | A quality society’s 8D page (no date shown in the page text; page data marked © 2026 ASQ) |
| Origins and escape point | Step descriptions; Global 8D’s D0 and escape point; Is / Is Not; disputed origins; limitations | A public online encyclopedia page (no date in the page text; page data shows last edited 2026-05-28; the page itself notes it needs additional citations) |
| Fishbone diagram | 6M; categories can be renamed; causes may repeat; not to be used alone | A quality society’s fishbone page (marked reviewed October 2024; page data marked © 2026 ASQ) |
| 5 Whys | Steps; not always five; pair with fishbone; ask “what do we change” | A quality society’s 5 Whys page (no date in the page text; page data marked © 2026 ASQ) |
| Root cause analysis | How to form the team; change analysis, barrier analysis | A quality society’s root cause analysis page (no date in the page text; page data marked © 2026 ASQ) |
| JIAHON’s own material | Product pages, published articles | JIAHON product pages and articles |
| Line practice | Date fields, reply deadlines and sign-off fields on 8D forms; management cause categories and action directions (no companies, machine models or dates named) | Compiled line practice (source documents not public) |
This describes the general case; individual situations may differ.
Exceptions and limitations
- Nature of the sources: this guide draws on four quality society pages and one encyclopedia page. It did not draw on automotive customers’ own 8D manuals and forms, or on the text of quality management system standards (which must be purchased), so their fields and requirements are not covered.
- Sources with a perspective: the encyclopedia page is public and anyone can edit it; the quality society pages are introductory material, not customer specifications.
- Disputed origins: the encyclopedia page says many publications attribute 8D to the US military and that this history is disputed. This guide reports what the page says and draws no conclusion.
- Where sources differ: the number of steps (whether there is a D0: the quality society page says some organizations add one, the encyclopedia page says Global 8D has it); whether interim containment is needed (the encyclopedia page says it depends on where in the product life cycle the problem occurs). This guide lists both without judging.
- Versions and dates: the fishbone page is marked reviewed October 2024; the other three quality society pages and the encyclopedia page show no date in their text, and the dates in their page data are in the sources table.
- This guide’s own examples: the 5W2H example in Stage 2 is an illustration created for this guide, not a customer or real case.
- This guide’s own structure and recommendations: splitting 8D into five stages (decide and form the team, describe and contain, find the root cause, act and validate, prevent and close); “the report has two audiences”; “seal the evidence before touching it”; “one leader and one customer contact”; “write occurrence and escape causes separately”; “go down until fixing it stops the problem”; “state when interim actions end”; “keep an overlap period before stopping interim actions”; “act just on verified causes”; “validate with the same yardstick as D2”; “follow up across shift, material and lot changes”; “reproduce on test pieces, not product that will ship”; “spot-check the floor after documents change”; “list rollout targets and owners.” That reports often get stuck at D4 is this guide’s judgment from experience, not a statistic from public sources.
- Line practice: line practices vary by plant; follow your own process validation.
- Scope: customer-specific 8D form fields, the full corrective and preventive action process, design of experiments and process capability calculations, and software are not covered.
- Equipment and products: specifications are as stated on each product page. The semi-automatic stencil inspection machine is currently on pre-order.
This guide summarizes the general case. It is not a restatement of any standard, not an endorsement of any vendor, equipment maker or service provider, and not a promise about your product or process. Follow your customer’s specifications, local regulations and your own validation; individual situations may differ.