8DEight Disciplines problem-solving

8D is a disciplined sequence — not a substitute for technical problem-solving, but a wrapper that ensures the right people, the right immediate containment, the right root-cause investigation, the right verified corrective and preventive actions, and the right team recognition all happen in the right order. The methodology was created at Ford in 1987 to standardise responses to field defects and warranty claims; the eight-step structure spread through the automotive supply base, into aerospace (AS9100), into medical-device suppliers under IATF/ISO 13485 dual certifications, and is now the de-facto format for complaint and supplier-corrective-action (SCAR) responses in any tier-1/tier-2 regulated supply chain.

1. D1 — Form the team. Cross-functional, owner appointed, skills matched to the problem. Often preceded by a D0 'prepare' step in the modern 9-step variant.
2. D2 — Describe the problem. Quantified, with the 5W2H (what, where, when, who, why, how, how many) and supporting evidence (photos, defect samples, batch records).
3. D3 — Interim containment. Immediate actions to protect the customer: stock sweep, quarantine, 100% sort, replacement shipments. Effectiveness verified before progressing.
4. D4 — Root cause(s). 5 Whys / fishbone / fault tree / Is-Is Not — identify both the technical root cause (why did this happen) and the escape root cause (why did our detection not catch it).
5. D5 — Choose and verify permanent corrective actions. Selection criteria documented; proposed actions pre-tested to confirm they remove the root cause without introducing new failure modes.
6. D6 — Implement permanent corrective actions and verify effectiveness. Actions deployed, deviation closed, effectiveness verified against a defined success metric over a defined window (often 30/60/90 days).
7. D7 — Prevent recurrence. Update FMEA, control plan, work instructions, training records; cascade the lesson to similar processes / products / sites — the 'yokoten' horizontal-deployment step.
8. D8 — Recognise the team. Closure with formal acknowledgement of the team's contribution. Often dropped in practice; it is the discipline that drives engagement for the next 8D.

A defining feature of 8D — and what makes it more rigorous than a simple 5-Why — is the explicit requirement to identify both the technical (occurrence) root cause and the escape (detection) root cause. A part shipped out of spec because the operator misread the gauge (occurrence). A part shipped out of spec because the audit-sample plan was 1-in-50 and the defective lot was outside the sampled subset (escape). Both root causes need their own corrective action: re-train + visual-aid the gauge for the technical side, tighten the sampling plan or add an end-of-line poka-yoke for the escape side. Treating only one of the two is the most common reason a closed 8D produces a repeat complaint within months.

CAPA (21 CFR 820.100, ISO 13485 §8.5.2, EU GMP Ch.1 §1.4.xiv) is the regulatory requirement; 8D is one of several execution formats that satisfy it. A pharma CAPA might be documented on a company-specific multi-page form; an automotive supplier-corrective-action will almost always be on an 8D form because the customer requires it. The discipline content is similar — containment, root cause, corrective action, effectiveness check, prevent recurrence — but 8D's strength is the team-formation discipline (D1) and the cascade step (D7), which CAPA forms often skim. Many regulated manufacturers run CAPA internally and 8D externally for customer-facing SCARs.

An 8D closed without measured effectiveness is paperwork, not improvement. Effectiveness verification requires a pre-stated success metric (zero recurrence in the next 90 days, defect rate below 0.5%, audit finding closed against re-audit) and a measurement period long enough that ordinary variation can't fake success. Premature closure — 'no recurrence in two weeks, closed' — is the second most-cited 8D weakness after the missing escape root cause. Mature programmes track repeat-complaint rate per closed 8D as a leading indicator of effectiveness-verification quality.

1. D1 team is one person — usually the quality engineer doing it alone, missing the cross-functional perspective that's the whole point.
2. D2 is qualitative ('intermittent defects in shipment 4567') with no quantification, no defect rate, no supporting evidence.
3. D3 containment scope too narrow — only the complaining customer's inventory swept, the same defective material still in transit to other customers.
4. D4 escape root cause skipped — only the occurrence root cause documented, escape question never asked.
5. D5 corrective action picked before D4 root cause confirmed — 'we'll add an inspection step' decided in week 1, root cause investigation back-filled to justify it.
6. D6 effectiveness verification 'closed pending data' — never actually re-checked.
7. D7 horizontal cascade skipped — the lesson sits in one product line; the same failure mode shows up on a sibling line a year later.
8. D8 recognition skipped — over time the team that always gets called runs out of patience.

• 8D template per complaint / SCAR / NCR — D1-D8 enforced as gated steps, you cannot skip from D2 to D5.
• D1 team formation pulls cross-functional members from training-record-matched skill profiles; team owner appointment is recorded with e-signature.
• D4 demands separate occurrence and escape root-cause records; closing D4 without both blocks progression to D5.
• D5 proposed corrective actions linked to the FMEA and control plan they will modify; D7 cascade list is auto-generated from products/processes sharing the same FMEA failure mode.
• D6 effectiveness verification scheduled at the chosen window with a pre-defined success metric; the 8D cannot close until the verification check passes.
• Repeat-complaint detection — a new complaint matching the failure mode of a closed 8D re-opens the parent and flags the original effectiveness verification as failed.
• External 8D report PDF generated automatically with the customer-required Ford TOPS-8D template format.