How to Match a COA to a Lot, Batch, and Package

Define the three objects before comparing them

Treat the physical package, the U.S. production batch record, and the laboratory report as distinct objects. Under 21 CFR 111.260, a batch production record connects a finished batch to the lots of packaged and labeled dietary supplement and includes or cross-references test and examination results. An ISO/IEC 17025 laboratory report identifies the testing work and tested item or sample within its scope. Neither source says that a public retail certificate automatically represents every package carrying a similar product name.

Write the proposed connection as a question rather than a conclusion: does this report's stated sample or lot connect, through explicit identifiers or a controlled manufacturer record, to this package? MIHEN's comparison fields (product variation, market, item code, report identifier, laboratory sample, finished-batch value, and package-lot value) form an editorial audit trail. Record only fields shown by the sources under review and preserve blanks. Do not borrow a likely identifier from a neighboring product, image, or certificate.

Capture package identifiers from the item in hand

Photograph the principal display panel, facts panel, manufacturer or distributor statement, universal product code when present, lot or batch marking, and printed date code. Record where each code appears because an outer carton, bottle, cap, seal, or sachet may carry different information. Transcribe punctuation and leading zeros exactly. A retailer stock number or order number identifies a listing or transaction and should not be substituted for a manufacturer lot. FDA's small-entity guide notes that a printed batch or lot number is one traceability method, not the only permitted mechanism.

Package images from a seller can assist preliminary research, but they may be generic or older than the stock delivered. Label the evidence source as physical package, manufacturer image, seller image, or consumer-supplied image. If no physical package is available, the match cannot reach the same evidence state. Keep that limitation visible. Differences in artwork or address can indicate a regional or revised package, but they do not by themselves prove that the package and certificate are unrelated. Seek identifiers rather than relying on visual resemblance.

Extract certificate identifiers without normalizing them away

From the document, capture only identifiers and dates it actually displays: report number, client, product or material description, client item or lot value, laboratory sample identifier, receipt or sampling statement, test dates, and issue date. Store the original strings before creating search-normalized copies. Removing punctuation may help find a candidate match, but two identifiers that become equal only after editing remain a manual-review result rather than an automatic confirmation.

Determine whether the reported item is a finished product, component, environmental sample, or another matrix. In 21 CFR 111.75, a supplier COA is specifically one route for determining whether certain component specifications are met; it is not a rule that converts that component document into finished-product testing. A general specification or example report also lacks a lot-specific link. When the document says client supplied the sample, preserve that statement without inferring how it was selected or whether it represents the whole batch.

Apply a conservative match hierarchy

For MIHEN's editorial classification, reserve confirmed documentary link for cases where the report or a controlled manufacturer record supplies the exact finished-batch or packaged-lot connection and enough product detail to exclude a different variation. This is a conservative publication rule, not a claim that 21 CFR or ISO assigns MIHEN's labels. Product name alone is insufficient for that status. A conflicting item code remains a conflict; a missing lot on one side remains incomplete.

Define evidence states before reviewing a document: explicit documentary link, candidate link with missing fields, unresolved, conflict, or different lot. Record each comparison field as verbatim match, normalized candidate, compatible, missing, or conflicting. Normalization should never overwrite the original value. A field-level table lets another reviewer reproduce the editorial decision and identify the exact missing link, while the underlying manufacturing and laboratory sources retain their own terminology.

Review dates, revisions, and chain records

Dates provide chronology but do not replace identifiers. A certificate issued after manufacture can be plausible, while an issue date before a stated sample receipt date requires clarification or a documented correction. Compare manufacture, sampling, receipt, testing, issue, revision, and package date fields without assuming they all describe the same event. A revision may correct formatting, identifiers, calculations, or results. Retain the superseded report status and identify which version is currently represented as controlled by the issuer.

Where actually supplied, supporting records may include a manufacturer response, sample-submission statement, custody record, or laboratory confirmation. Treat each as a separate source and state only what it explicitly connects. An email confirming a report number may establish issuer confirmation while leaving the package link unresolved when no finished-batch or packaged-lot value appears. Do not infer a chain of custody from an ordinary email. Protect private details and obtain permission before publishing nonpublic records.

Publish the match and its boundary

A public entry should name the exact package variation, transcribe the package lot, identify the certificate and revision, state the match classification, list unresolved or conflicting fields, and show the verification date. Link the complete certificate when lawful and practical. If only a manufacturer-hosted lookup returned the document, preserve the query input and result URL or capture method. Do not call a document batch-specific when the displayed report contains no batch or lot identifier.

Even an explicit documentary link establishes only that the report is associated with the named sample and, through the cited record, the named batch or packaged lot. It does not establish that every package was sampled, that other lots share the results, that the report covers unlisted tests, or that a health conclusion follows. Recheck a replaced report or revised package. A transparent unresolved result is preferable to attaching the wrong technical document to a product page.

• Create separate records for the package, production lot, and laboratory report.
• Photograph and transcribe package codes with punctuation and leading zeros.
• Distinguish manufacturer lot codes from retailer and transaction identifiers.
• Capture report, sample, item, product, client, and date fields from the COA.
• Confirm whether the tested matrix is a finished product or another material.
• Compare fields as exact, normalized exact, missing, compatible, or conflicting.
• Retain revision history and identify the issuer's current controlled version.
• Publish the match state, verification date, and unresolved links.