Reading the vendor's QC/QA records

01Why reading records is the owner's job

Vol. 3 covered the intake flow and Vol. 4 the scope of verification at receipt. What we should have paused on there is the act of reading the records a vendor attaches. When you outsource production, you also outsource much of the verification to the vendor's QC (Quality Control) and QA (Quality Assurance). What reaches your desk is the deliverable itself plus a bundle of records that claim to back it up.

The common error here is mistaking "a record exists" for "verification happened." Reading a filled-in signature box as proof of verification is like reading a stamp on a delivery slip as proof the contents are correct. Responsibility for what reaches the market stays with the owner (see Vol. 2, Moral Responsibility). If so, refusing to take records at face value — distinguishing what a record guarantees from what it does not — becomes work the owner cannot outsource.

02QC records and QA records are different things

Separate the character of the two records first. Conflate them and you will feel you checked both after reading only one.

QC is "checking what was built"; QA is "assuring the process and deciding fitness to ship." Before pushing to internal review, the owner looks for traces that both layers actually functioned. A thick QC sheet topped by a one-line QA "no issues" is not a two-layer structure. The vendor's quality system itself is the subject of the sister series Production Quality Management; the owner needs only enough resolution to read its output.

03Telling formal records from substantive verification

This is the most practical point. Records come in two kinds: substantive records that transcribe a verification that happened, and formal records produced to claim it happened. The latter is harder to spot the tidier it looks.

The keys are "linkage to a basis" and "traces of friction." When verification really occurs, it bumps into something and small corrections appear. A flawless, frictionless record is itself grounds to suspect the absence of verification.

04Checklist for assessing sufficiency

When you receive the records, assess whether the record itself is sufficient before tracing the substance item by item. Apply a fixed set of questions every time.

1. Independence — Are the QC and QA roles distinct from the creator? Are signatures all from one person or one team?
2. Traceability — Does each judgment tie to a source (approved information, package insert, the MSA guidelines, the advertising standards clause)?
3. Coverage — Are efficacy, dosage, safety information, citations, comparative claims, figure/graph alteration, and audience (HCP-directed?) all present as items?
4. Norm references — Are Act §66 (exaggeration), §68 (pre-approval), §68-2 (the duty of effort on information provision), the advertising standards, the MSA guidelines, and the JPMA Code cited by specific clause, not abstract name?
5. Deviation handling — Where a finding arose, does a correction and re-check follow? If findings are zero, question that separately.
6. Version control — Does the version the record covers match the final delivered version (no QC record from an old version reused for a new one)?

05Flagging the gaps — the usual blanks

A record's weakness shows in what is not written more than in what is. The owner hunts not for empty cells but for "fields that should exist and don't." The frequent gaps:

When you find a gap, treat it not as "the vendor's mistake" but as "an unverified area the owner needs to know about now." Flagging gaps is not for blame; it is the work of filling the blanks before the piece enters internal review.

06How to read the QA disposition

The QA disposition sits at the top of the records. Here the vendor declares "fit to ship." The owner reads the logic of that declaration — not just the conclusion, but the path to it.

• Disposition categories — Just "conforms," or are there "conditional" and "non-conforming, re-judge after correction"? A disposition with a single category likely lacks a criterion for failing anything.
• Deviation handling — Are QC findings consolidated here with each resolution state stated? Is it "fit to ship" with unresolved findings still open?
• Who judged — Is the signer different from the QC performer? Is the judge's scope of authority (what they stand behind) stated?
• Reservations — Are caveats like "X to be finally confirmed by the owner" written in? A blanket guarantee with zero reservations actually blurs the boundary.

The point of the read is whether the disposition is "a boundary line of who guaranteed how far" rather than "a declaration of infallibility." A disposition with a clear line tells the owner what to check next. A line-less "all fine" leaves the guaranteed scope unreadable, and the owner ends up carrying everything.

07Criteria for sending records back

What do you do when reading shows the record is insufficient? Accept it vaguely and the defect slips straight through internal review into the market. Fix the return decision as categories in advance.

A return is not an act that breaks the relationship. The more the criteria are shared in writing, the easier it is for the vendor to meet the bar from the start. Compromising case by case while leaving criteria vague is what erodes quality over time. The return decision and how to record it are covered in Vol. 6 (in preparation).

08Handing records to internal review

The records the owner has read pass on to internal material review. So the owner's reading does not end as a one-off, shape the handoff. Put the reviewer in a position to distinguish "what the vendor checked," "what the owner confirmed," and "what review still needs to see."

1. Intake summary — The sufficiency assessment result (accept / conditional / return history) on one page.
2. Stated unverified areas — Items missing from the vendor record and their completion status, so review can focus there.
3. Norm mapping — A table mapping key claims to cited clauses (§66/§68/§68-2, advertising standards, MSA guidelines, JPMA Code). Misassigning clauses is a source of incidents in review too, so fix it here.
4. Carry over the disposition's reservations — Convert points the vendor marked "owner to finally confirm" into review checklist items.

Reading records is not about policing the vendor. It is the work of keeping the chain of verification unbroken and handing it to internal review, so the owner discharges responsibility for what reaches the market. Passing an unreadable record straight through produces the same result as abdicating that responsibility.