BRCGS Clause 3.9

This topic is part of the SG Systems Global regulatory & operations guide library.

Updated January 2026 • BRCGS Clause 3.9: traceability, mass balance, mock recall testing, audit-ready records.

BRCGS Clause 3.9 is where “we have traceability” stops being a claim and becomes a timed, testable capability. In BRCGS Food Safety Issue 9, Clause 3.9 is a fundamental requirement: you must be able to trace raw material lots (including primary packaging) from suppliers through all stages of processing and dispatch to customers — and trace the other direction too. That “both directions” point is what catches sites that only track one side of the chain.

The clause is short, but the expectation behind it is not. Auditors don’t want a nice procedure. They want proof that your records and labels create a defensible chain: lots in, lots transformed, lots packed, lots shipped, and quantities that reconcile. If you can’t complete a forward/backward trace under time pressure with a credible mass balance, you don’t have traceability — you have a spreadsheet story that collapses the first time something goes wrong.

Traceability isn’t a document. It’s your ability to answer “where did it go?” and “what did it touch?” before the situation outruns you.

1) What Clause 3.9 really means in an audit

Clause 3.9 is not asking you to say you can trace product. It’s asking you to demonstrate it under test conditions. The audit pattern is predictable:

• Backward trace: Auditor picks a finished lot. You must identify the raw material lots and relevant packaging lots used, plus where/when they were used.
• Forward trace: Auditor picks a raw material lot (or packaging lot). You must identify every finished lot that consumed it and where those finished lots shipped.
• Quantity credibility: Auditor expects a quantity check that closes: what came in, what was consumed, what was produced, what was scrapped, what remains in inventory/WIP.
• Time pressure: You’re expected to complete the trace within your defined time objective (build to a ≤4-hour target).
• Record integrity: The links must be evidence-based, not “tribal knowledge.”

What makes this clause “fundamental” is that it sits under almost every serious scenario: foreign material events, allergen events, label events, supplier issues, customer complaints, and regulatory inquiries. If your traceability is slow or ambiguous, you don’t just get a finding — you lose the ability to execute targeted withdrawals. That’s when “small” issues become expensive ones.

2) Scope: what must be traceable (and what gets missed)

Clause 3.9’s scope is broader than many teams assume. Sites commonly build traceability around ingredients only, then get surprised when the auditor asks about packaging. Under Issue 9, raw materials include primary packaging in scope — and your traceability test must demonstrate links that include packaging where required.

If you want a quick self-test: ask whether you can trace a single printed label roll (or case label lot) forward into finished shipments and backward to the receipt and issuance event. If not, your “ingredients traceability” is incomplete.

3) Traceability vs genealogy vs chain of custody

Teams get tangled because these terms are used interchangeably. They shouldn’t be.

• Traceability is the operational ability to trace and follow lots through receipt, processing, and distribution — forward and backward. It is a capability, not a report.
• Genealogy is the relationship map: which lots were combined, split, reworked, and packaged to produce which finished lots. For BRCGS readiness, you want explicit end-to-end genealogy, not inferred linkages.
• Chain of custody is about control and integrity of ownership/handling through the chain — useful for provenance, identity preserved programs, and customer requirements. It’s related, but not the same as Clause 3.9 traceability. See chain of custody.

Where Clause 3.9 aligns strongly with “genealogy” is the bi-directional expectation: the auditor can pick either end of the chain and expect you to traverse it without guesswork. That’s why “one-up one-down” is rarely enough on its own. It’s a baseline, not a complete internal manufacturing genealogy model. (Still valuable: one-up one-down traceability.)

4) The traceability procedure that actually passes

Clause 3.9.1 requires a documented traceability procedure. Many sites write a long SOP and still fail the test because the SOP doesn’t define how the system works in real terms. A passing procedure is short, explicit, and test-driven.

One subtle but important point: Clause 3.9.1 also expects alignment with legal requirements in the country of sale or intended use. That means your procedure should explicitly state how the site verifies and maintains compliance (e.g., customer-specific traceability expectations, export markets, or additional record requirements).

5) Identifiers: how to label so the chain survives reality

Traceability is only as good as your identifiers. If a lot can be “described” but not uniquely identified, you’ll lose the chain during normal operations. The cure is to standardize identifiers and force them into every movement and transformation record.

At a practical level, that means:

• Product identity uses a stable identifier (e.g., GTIN or internal equivalent) tied to the approved specification.
• Lot identity is present everywhere, consistently formatted, and captured by scan where possible (barcode validation matters because garbage scans create garbage traces).
• Logistic units (pallets) carry a unique identifier (e.g., SSCC) that links cases to shipments.
• Printed label and packaging lots are treated like materials — not like “supplies.”

For many sites, the most practical barcode backbone is GS1-128 at case/pallet level (see GS1-128 case label) and consistent intake labeling (see intake label capture and raw material intake labeling). When you standardize application identifiers (see GS1 application identifiers), you reduce “interpretation” errors across sites and customers.

6) Event model: the records you must be able to produce

Clause 3.9 is easiest to satisfy when you think in “events” rather than “documents.” Every event is a link in the chain. If an event is missing or ambiguous, your chain breaks.

If you implement an event structure cleanly, traceability becomes a query problem, not a scavenger hunt. That’s also where standards like EPCIS can help: they give you a consistent way to capture “what happened, when, where, and to what” across systems.

7) Mass balance & quantity checks: the make-or-break point

Most sites can produce a list of “lots used.” Fewer can prove quantities credibly. That’s why Clause 3.9.3 explicitly expects a quantity check / mass balance where required. Auditors use mass balance as the reality filter: if your quantities don’t close, your trace is not trustworthy.

Mass balance is not one thing; it’s a discipline. A credible approach defines:

• Boundaries: What time window and what process path are included?
• Units: Do you reconcile in kg, cases, eaches, or both? Are conversions governed?
• Loss categories: What losses are expected (trim, evaporation, startup waste, QC sampling) vs abnormal (unexplained shrink, undocumented rework)?
• Inventory truth: Is your on-hand inventory reliable enough to support reconciliation (see inventory accuracy and cycle counting)?
• Decision rules: Who can accept a variance and when does it become an investigation?

Mass balance gets trickier when you run variable-weight processes (meat, produce, bulk) or when you repack/rework. That’s not an excuse — it’s a design requirement. If your operation is catch-weight heavy, build traceability with variable-weight realism (see catch-weight traceability and catch weighing) and ensure your quantity logic matches how the floor actually behaves.

8) Mock recall / traceability test: how auditors test you

Clause 3.9.3 expects you to test the traceability system across product groups at a predetermined frequency (at least annually), retain results, and complete within your target time. If you only test once a year as a box-check, you’ll drift. The best sites test on a cadence that matches operational complexity — and treat failures like real incidents.

Two tools that strengthen Clause 3.9 readiness are routine recall drills and explicit measurement of mock recall performance. These give you a leading indicator: if your trace times are trending up or your mass balance closures are getting weaker, you’ll know before an auditor tells you.

9) Rework and reprocessing: where traceability dies quietly

Clause 3.9.4 is blunt: if you perform rework or reworking operations, traceability must be maintained. This is where many sites unintentionally create “traceability holes” because rework is treated like a convenience rather than a controlled process.

Rework breaks traceability in three common ways:

• Unidentified rework inputs: bins/totes are not labeled to lot identity, or they are co-mingled without rules.
• Unrecorded rework linkages: the finished lot that consumed rework is not linked to the source lots that created it.
• Quantity ambiguity: rework additions are “about a tote” rather than measured, creating mass balance gaps.

If rework is part of your operational model, treat it as a first-class traceability object: assign a controlled identifier, record it like material, and link it into genealogy (see rework traceability and rework/repack traceability). If rework is not controlled, it becomes the easiest route for unknown material to enter finished product — which destroys your recall precision.

10) Packaging & labels: why “ingredients only” is a trap

Packaging is not “just packaging” when it carries legal and food safety information. Clause 3.9 expects traceability to include primary packaging and, in traceability testing where required, printed packaging and labels with legal/food safety information. In practice, that means your traceability and your labeling controls must be aligned.

Three packaging controls routinely separate “passes audit” from “gets written up”:

• Artwork/version governance: ensure the right label/pack revision is issued and used (see labeling control).
• Line clearance evidence: prove the line was cleared of prior materials before start-up (see line clearance and packaging line clearance verification).
• Reconciliation: reconcile printed labels in/out/scrap so unaccounted labels don’t exist (see label reconciliation).

Packaging traceability also ties to movement and unit-load build. If you allow uncontrolled pallet mixing, you may still “have traceability,” but it becomes slow and broad. Mixed pallets can be valid in some operations, but they must be governed and still traceable at the case/SSCC level (see mixed-load segregation).

11) Operational controls that make traceability real

Clause 3.9 is passed or failed on the shop floor. Your systems help, but your operating controls decide whether the data is trustworthy. The strongest sites build traceability into routine disciplines:

• Status discipline: if a lot is on hold/quarantine, it cannot be consumed or shipped. No exceptions without an auditable release.
• FIFO/FEFO compliance: govern consumption rules to reduce partial chaos (see FIFO and FEFO).
• Inventory truth: cycle count and resolve discrepancies fast (cycle counting), especially for printed packaging and high-risk ingredients.
• Identity confirmation: verify materials and components at point-of-use (see material identity confirmation and component identity verification).
• Controlled transfers: internal movements are scanned and recorded (see internal movement scanning).

When these controls are present, the traceability test becomes boring (in a good way). When they’re weak, the test becomes a “who can find the missing binder” exercise — and that’s exactly what auditors are trained to see through.

12) Systems: WMS/MES/ERP and standards like EPCIS

Clause 3.9 doesn’t mandate a specific system, but it effectively mandates system-like behavior: consistent identifiers, event capture, and audit-ready reporting. Most sites use a combination:

• WMS for inventory movement truth and status enforcement (see WMS).
• ERP for procurement, master data, and financial trace (see ERP).
• Execution / shop-floor capture (often MES-like behavior) for transformation events and packaging usage records.
• EDI / ASN for outbound proof and customer integration (see EDI and ASN).

If you’re aiming for future-proofing (especially where customers push digital traceability), adopt a clean event model and consider standardizing exports around EPCIS. EPCIS won’t fix weak operations, but it does prevent every integration partner from inventing a different “trace format.”

Also note the convergence happening in food traceability: Key Data Elements (KDEs) and critical tracking events are becoming more explicit in regulatory and customer expectations. Mapping your Clause 3.9 data to FSMA 204 KDE concepts can reduce rework later, even if your current compliance driver is BRCGS rather than regulation.

13) KPIs, governance, and continuous readiness

Clause 3.9 is not “set and forget.” Your traceability capability degrades unless you measure it. The simplest governance model is:

• Monthly traceability mini-tests: small-scope traces with time recording and variance tracking.
• Quarterly deep tests: include rework, multi-stage WIP, and packaging lots with a full evidence pack.
• Annual formal test: meets the Clause 3.9.3 requirement across product groups and retained for audit.
• Management review: trend test times, mass balance closure, recurring failure modes.

When issues appear, treat them as system failures, not “people mistakes.” If the same failure recurs (e.g., missing label lot linkage), your process design is wrong. Fix the design: tighten issuance, force scans, require reconciliation, and remove bypasses.

14) Common nonconformities (and how to prevent them)

Clause 3.9 findings tend to cluster into the same buckets. If you want to get ahead of the auditor, pre-audit yourself against these.

Notice what’s missing from that list: “buy a new software.” Most fixes are operational — but software can enforce those operations so they don’t depend on heroics.

15) Copy/paste traceability drill script + scorecard

If you want to pressure-test your Clause 3.9 readiness, use a repeatable drill. Don’t let teams “prepare.” Surprise is the point.

16) Industry examples and patterns

Clause 3.9 is consistent across sectors, but the failure modes differ by process type. A few patterns:

• Produce packing: rapid lot turnover, mixed pallets, retailer programs — align to PTI where applicable (see Produce Packing and PTI).
• Meat processing: variable weights, high rework pressure, lethality/segregation constraints (see Sausage & Meat Processing and BRCGS meat controls).
• Dry mixes & ingredients: silo/partial management and cross-contact risks (see Ingredients & Dry Mixes).
• Bakery: rapid changeovers and WIP staging discipline (see Bakery Manufacturing).
• General food processing: multi-stage transformations and packaging complexity (see Food Processing).

The common thread: if you standardize identifiers, enforce scanning at critical points, and govern exceptions, Clause 3.9 becomes routine — regardless of sector.

17) Extended FAQ

Q1. What is BRCGS Clause 3.9 about?
Clause 3.9 is the traceability requirement: you must be able to trace raw material lots (including primary packaging) through processing and dispatch, and trace back from finished product to inputs, with routine testing and retained evidence.

Q2. Do we really need to include packaging and labels?
Yes. Primary packaging is explicitly in scope, and printed packaging/labels with legal and food safety information must be covered where required in testing. Packaging traceability failures are common audit findings.

Q3. What does “traceability should be achievable within 4 hours” mean?
Treat ≤4 hours as the internal benchmark for completing a full forward and backward trace with evidence and credible quantities. If you can’t do it under pressure, you’re not ready for real incidents.

Q4. How often should we test traceability?
At minimum annually per the clause, but operationally you should run smaller tests more often and trend performance. Annual-only testing lets weaknesses accumulate unnoticed.

Q5. What’s the fastest way to improve performance?
Remove ambiguity: standardize identifiers, enforce scanning, control rework, reconcile labels, and strengthen inventory accuracy. Most “slow traces” are really “missing data” problems.

Related Reading
• Traceability Core: End-to-End Lot Genealogy | Mass Balance | One-Up One-Down | Chain of Custody
• Packaging Controls: Labeling Control | Label Reconciliation | Packaging Line Clearance Verification
• Drills & Readiness: Recall Drill | Mock Recall Performance | Recall Readiness
• BRCGS Reference Term: BRCGS Clause 3.9 – Traceability Requirements