The short answer

A certificate claims a system exists; records prove it runs. Buyers who know which records to request can verify an inspection body's discipline without leaving their desks. This article lists the records that matter, shows how to read them, and explains what evasive answers reveal. It is the due-diligence companion to the earlier parts of this series.

Key takeaways

  1. Ask for the records behind the certificate: assignment logs, sampling calculations, evidence files, review notes, and dispute histories.
  2. Sampling records should show lot sizes, drawn samples, and accept numbers that follow recognised acceptance sampling practice.[1]
  3. Report files should read like acceptance tests held at the factory: clear verdicts on shippable quality backed by traceable evidence.[2]
  4. Review notes from someone other than the visiting inspector prove the four-eyes principle runs in practice rather than on paper.
  5. A provider that produces these records promptly is demonstrating its system; one that stalls is describing it.

Certificates photograph well in sales presentations. Records do not, which is exactly why serious buyers ask for records. Any organisation can hold a certificate on the wall while its daily practice drifts; only organisations whose records match their procedures are actually running the system the certificate describes.

Record verification sounds technical, but the buyer version is straightforward. Request a small set of documents for recent visits, read them comparatively, and watch for the gap between confident claims and thin files. The whole exercise takes an afternoon and reveals more than a year of sales calls.

This fifth part of the series gives buyers the request list, the reading method, and the interpretation guide. It assumes no audit training, only the willingness to ask for paperwork and read it carefully.

The five files that prove the system runs

First, assignment records: who inspected which factory on which date, and whether rotation actually happened. A provider claiming rotation should show it in the log, not merely assert it as policy. Repeated same-inspector same-factory patterns in the log contradict the brochure.

Illustration: The five files that prove the Decorative illustration for the section "The five files that prove the"; visual only, carries no data.

Second, sampling records: lot size, sample size, acceptance numbers, and the verdict for each visit. These figures should follow recognised acceptance sampling practice, with sample sizes rising sensibly with lot size rather than sitting at convenient round numbers.[1] Identical sample sizes for wildly different lot sizes indicate the tables are decorative.

Third, evidence files: the photos behind the findings, tied to cartons and samples. Fourth, review notes showing a second person checked the report before release. Fifth, the dispute file: factory challenges received and how each was decided. Together these five files describe the system's real behaviour, and UTS keeps assignment, sampling, and review records in a form buyers can request and read. The claim stands whether or not the buyer ever asks, but asking is how diligence works.

How to read a report file like an auditor

  1. Match the verdict to the arithmeticRecompute the finding against the stated sample size and acceptance numbers. The verdict should follow from the numbers, not sit beside them as an independent opinion.
  2. Trace every major defect to evidenceEach major finding needs a photo with enough context to identify the unit and carton. Findings floating without evidence are the file's weakest points.
  3. Confirm the acceptance framingThe report should answer whether the lot is fit to ship, with criteria and measurements supporting that verdict. Reports that read like travel diaries instead of acceptance records miss the method's purpose.[2]
  4. Look for the reviewer's fingerprintsCorrections, queries, and sign-off notes from someone other than the inspector show review happened. A file with zero review traces either had a perfect inspector or no review.
  5. Check the dispute trailIf the factory pushed back, the file should show the challenge and the reasoned outcome. Clean files with no disputes across many visits deserve a raised eyebrow rather than applause.

Interpreting what providers send back

Good signs in the response

Complete files arrive within days, numbers reconcile, evidence attaches to findings, review notes appear, and the dispute file shows real cases with reasoned outcomes. Minor imperfections openly acknowledged beat suspicious perfection.

The best responses also explain the system briefly: which procedure each record comes from and who owns it. Providers that can narrate their own paperwork understand it; those that dump files without explanation may merely store them.

Warning signs worth pressing on

Stalling, partial files, sample sizes that ignore lot size, findings without photos, no review traces, and dispute files that are suspiciously empty. Each gap maps to a specific system failure, and buyers should ask for the missing piece by name rather than accepting general reassurance.

The ultimate test is whether the provider treats the request as normal. Mature bodies receive record requests routinely and answer routinely. Defensive or astonished reactions suggest the request is the first of its kind, which says enough.

Illustration: Interpreting what providers send Decorative illustration for the section "Interpreting what providers send"; visual only, carries no data.

Request list and what each file reveals

File to requestCovers which visitsWhat it reveals
Assignment logLast 20 visitsWhether rotation policy runs in practice
Sampling recordsSame visitsWhether tables or convenience set samples
Evidence filesTwo full reportsWhether findings carry traceable proof
Review notesSame two reportsWhether a second person checked the work
Dispute historyLast quarterWhether the system survives factory pressure

Send this list as written. A provider that answers all five rows has little to hide; the pattern of any refusal tells you where the system is thinnest.

Request records for ordinary visits, not showcase inspections. Anyone can stage one perfect visit; the assignment log and the sampling arithmetic for routine weeks reveal how the system truly behaves.

Sources

  1. Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
  2. Wikipedia: Factory acceptance test —— Overview of factory acceptance testing performed before shipment or installation.

Frequently asked questions

How many report files should I request?

Two complete files plus the assignment log for the last twenty visits. That combination shows both depth and pattern: the files reveal evidence and review quality, while the log reveals whether rotation and coverage claims hold across weeks.

What if sampling numbers look identical across different lot sizes?

Treat it as a serious warning. Proper sampling scales with lot size through recognised tables, so flat sample sizes suggest inspectors pick convenient numbers. Ask which tables govern sampling and recompute two visits yourself.

Should an empty dispute file reassure me?

No. Across dozens of factory visits, some findings will be challenged; that is normal commercial life. An empty dispute file suggests challenges are handled informally or suppressed, either of which means the record does not reflect reality.

How quickly should a provider answer a records request?

Within days for standard files, since mature systems retrieve records routinely. Weeks of delay, partial answers, or requests to narrow the scope substantially all indicate records are reconstructed rather than retrieved.