A step-by-step walkthrough of a full mock recall — trace one lot back to every supplier input and forward to every customer shipment, reconcile the quantities, and document the timeline and the gaps.
Do these first
Real help from real food people
Chef Diego runs a real food plant. If this page didn't get you there, tell us — a person reads every message.
After this lesson you can run a mock recall from the first minute to the signed record: trace one lot both ways, prove the quantities add up, and hand an auditor the timeline they ask for. The reason to do it now is simple — every gap you hit becomes a corrective action on an ordinary Tuesday instead of a scramble during a real one.
You already picked the lot and named a coordinator in the last lesson. The drill starts the moment you write down the time and give that person the trigger: "Finished lot 2431 is our recall — go."
Pick a lot that will actually stress the system, not the easiest one on the rack. SQF Edition 9 makes this explicit: clause 2.6.3.2 says the recall test should cover products from different shifts and materials — including bulk materials — that are used across a range of products or shipped to a wide range of customers. A lot built from a shared ingredient that went out to a dozen accounts exercises far more of your records than a one-off batch that shipped to a single buyer. Choose the demanding one.
Then note the clock. Not because a stopwatch is the goal — the next lesson untangles where the "trace it in two hours" target really comes from — but because a start time and a finish time are the first things a real recall and a real auditor both ask for.
Clause numbers move — confirm your edition
The clause numbers here are from the SQF Food Safety Code, Edition 9, the basis for audits as of 2026. SQFI has published Edition 10, which phases in later, and BRCGS and FSSC 22000 number their traceability requirements differently. The drill is the same under any GFSI-recognized scheme; check the exact clause against your own certification code before you cite it.
The first leg answers one question: what went into this lot, and where did each piece come from? The regulatory floor is , which clause 2.6.2.1 states as at least one step back from the process to the manufacturing supplier.
But the floor is not where a recall lives. Open the batch record for the lot and list every ingredient lot that went into it, the supplier each one came from, and the date you received it — clause 2.6.2.1 also requires those receipt dates to be on record. Then keep going: for a bulk ingredient that arrived under one supplier lot and got split across several of your batches, you need to know which of your batches drew from it.
This is where a paper trail shows its seams. The receiving log has the incoming lot, the batch sheet has what you made, and the link between them lives in whoever remembers loading the tote. The docs walkthrough on receiving a purchase order shows one way to capture an incoming lot at the dock so it stays tied to everything it becomes — which is the whole difference between a backward trace you can follow and one you have to reconstruct.
The second leg runs the other direction: everywhere this lot went. Start from the same finished lot and list every shipment that carried it — every customer, the quantity each one received, and the date it left your dock. Clause 2.6.2.1 puts the floor at one step forward to the customer, meaning the direct account you sold to, not the shelf the shopper eventually bought it from.
If the ingredient you started from touched more than one finished lot, the forward trace fans out: each of those finished lots has its own shipments and its own customers. Rework makes this harder, and the code accounts for it — clause 2.6.2.1 requires traceability to hold where product is reworked, so a lot that got blended back into a later batch has to carry forward into that batch's shipments too. Follow every branch. A forward trace that stops at the first customer list is the one that misses the account you rushed a partial shipment to on a Friday.
Tracing tells you where the lot went. It does not, on its own, tell you whether you found all of it — and that gap is where a mock recall quietly fails. You can produce a clean-looking customer list and still be missing a pallet, because nothing in a bare trace forces the numbers to close.
is what closes them. Add up the finished units you shipped, the units still in your building, the amount consumed into other batches, and the waste, samples, and rework you can document. That total should match what you started with. When it does not, either a shipment or a use is missing from your trace, or a record is wrong — and you have to explain the difference. Ordinary process loss accounts for a little; a whole missing pallet is not process loss.
Here is the honesty point, and it matters as much as the drill itself: the SQF code never uses the words "mass balance," and it sets no formula. What clause 2.6.3.2 actually requires is that your recall system be "reviewed, tested, and verified as effective." Reconciling the quantities is how you demonstrate that — it is the evidence that your trace is complete, not merely plausible. Treat it as the practice that proves the clause, not as a clause of its own.
The record is the deliverable. Everything above is the work; this is what survives it. Write down when the drill started and finished, the percentage of the lot you accounted for, every point where the trace stalled or a record was missing, and a corrective action — with an owner and a due date — for each gap. Clause 2.6.3.3 requires you to keep records of recall tests, the root-cause work behind any failure, and the corrective and preventive actions you applied. A mock recall with no written outcome is, to an auditor, a mock recall that did not happen.
Be honest in the write-up about what slowed you down. "Took four hours because receiving lots live in a separate binder" is a finding you can fix. A tidy report that hides the scramble teaches you nothing and fails you at the real thing.
The manual version of this drill takes most operations hours — sometimes the better part of a day — and nearly all of that time goes into stitching scattered records back together. A system that holds lot genealogy as links rather than as numbers on separate forms, which is what Bettr Manager does, runs the same two-direction trace in seconds, because the connections were recorded as the work happened. The exercise you just walked is the standard; software is only one way to meet it faster.
You now have two numbers you did not have this morning: how long the trace took, and how much of the lot you could account for. A full mock recall goes one step further than tracing — it also rehearses reaching those customers and pulling the product, which a later lesson covers on its own. But the trace and the reconciliation are its spine, and you have now run both.
The obvious next question is whether your time is good enough. The answer is not in the SQF code, which sets no clock at all — it is in what your customers demand. That is the next lesson.
Do this in your operation