The Quality Assurance SOP— Gate D ── Guaranteeing regulatory fit and never letting one person decide delivery alone

01Do Not Confuse QC With QA ── Drawing the Line

QA fails on most production floors because it becomes QC's "one more time." The same person, with the same eyes, runs down the same checklist twice. That is duplicated inspection, not assurance.

QC (Part 7, Proofreading and Consistency Inspection) looks at internal quality — was it built to spec? Typos, broken links, figure numbers, transcribed values, tone. QA looks at external fitness — can this be released against the four regulatory tiers (the Act, the Advertising Standards, the MSA Guidelines, the JPMA Code), and is the client's intent realized?

02Independence ── The Author Never Says "Ship It"

The spine of the QA SOP fits in one line: separate the author from the QA reviewer. This is not about trust or distrust; it is about the structure of cognition. A writer cannot see the logical gaps in their own prose. Someone convinced they have met the client's intent cannot test the possibility that they have not. When Generator and Evaluator are the same person, evaluation collapses into self-justification.

03Gate D Pass Criteria ── What Earns a "Ship It"

Never sign on "probably fine." Pass criteria are written down in advance, and nothing passes unless every item is PASS. The criteria are ordered in three tiers — regulatory fit, then scientific accuracy, then realization of intent — matching the definition of quality set out in Part 1.

1. Regulatory fit (mandatory; a single NG forces a return): Indications and dosage within the approved range. No exaggerated or absolute claims (Act §66). No advertising of unapproved drugs or off-label use (§68). Information provision kept distinct from advertising and meeting the request-based requirements (§68-2 / MSA Guidelines). Comparative, safety, and economic claims compliant with the Advertising Standards. Procedural requirements of the JPMA Code / creation guide (approval number, citation, conflict of interest) satisfied.
2. Scientific accuracy: Every claim tied to a source. Citations match the originals (no alteration or cherry-picking). Population, conditions, and limitations stated. QC record present.
3. Realization of intent: Meets the SOW and concept fixed at Gate A. Fits the target reader, use scene, and channel requirements.

04The QA Checklist (The SOP Itself)

File the following with the job record, leaving a verdict (P/F/N-A), reviewer initials, and the means of verification (source page number, etc.) for each item. A signature over any blank item is void.

When using N-A (not applicable), leave a one-line reason, e.g. "D-5: contains no comparative claims, N-A." An N-A without a reason is treated as F.

05The Ship-or-Return Logic ── Three Exits

QA has exactly three exits. Do not let a vague "conditional OK" pass by word of mouth; force every case into this taxonomy.

06Escalation Routes ── To Whom, What, and When

Settling escalation as "ask the boss" leaves no record of the reasoning and does not reproduce. Fix the route, trigger, and record in the SOP.

Always attach the QA reviewer's tentative verdict. Do not toss the decision upward; present your read and its basis, then ask the senior party to confirm — this raises both the quality and the speed of the answer. Record the response on the job sheet and accumulate it as precedent for similar cases.

07The QA Sign-Off ── What the Signature Stands For

A signature is not a formality; it makes responsibility explicit. It records who, when, and on what basis said "ship it." At minimum, the signature block carries:

• QA reviewer name and timestamp (a name verifiably different from the author)
• Version / revision number of the deliverable (the signature is bound to that version; swap it and the signature voids)
• Verdict (PASS / any N-A items and their reasons)
• QC completion record reference number (proof of Gate C clearance)
• Reference to any escalation response

Even after sign-off, a newly surfaced regulatory amendment, recall case, or label revision triggers a retroactive re-QA. The signature is assurance at a point in time, not a perpetual guarantee.

08Running QA ── Records, Metrics, Improvement

An SOP becomes quality only once it is operated. Measure and improve the QA process itself.

• Return rate (RETURNs / QA cases): too high points to upstream (authoring, QC) problems; too low with leakage means QA is not working.
• F frequency by regulatory item: a pattern such as frequent D-2 (exaggeration) feeds back into authoring-stage checks and training.
• Leakage count (non-conformities found after delivery): QA's true failure metric. Each one gets a root-cause analysis and an SOP revision.
• Escalation resolution lead time: too long means the route or authority design needs review.

These connect to traceability in the next process (Part 10, in prep). Only when you can trace which version, by whom, and on what basis it was passed does post-leakage investigation and recurrence prevention become possible. The QA SOP does not stand alone; it works by connecting to the records of the whole lifecycle.