A recall is one of the highest-stakes events a food manufacturer can face. The decisions you make in the first few hours matter enormously: how quickly you can identify affected product, how completely you can account for it, and how confidently you can communicate with distributors, retailers, and regulators.
A mock recall doesn't simulate the pressure of a real event. What it does is test whether your systems (your lot codes, your distribution records, your traceability chain) actually work before you need them to.
For most small manufacturers, running a mock recall once a year is sufficient. The exercise itself takes about two hours. What it reveals can take considerably longer to fix if you wait until a real event to find out.
What a mock recall tests
A mock recall is a traceability exercise. You select a finished product lot, then work through two questions:
Forward trace: where did this product go? Which customers, distributors, or retailers received it, and in what quantities?
Backward trace: what went into this product? Which ingredient lots were used, received from which suppliers, on which dates?
A complete mock recall answers both questions fully and accounts for 100% of the selected lot: product sold, product on hand, and product disposed of or sampled. The gap between what you can account for and 100% is your traceability exposure.
Before you start: what you need in place
A mock recall is only as useful as the records behind it. Before running the exercise, confirm you’ve:
Lot codes assigned to every finished product batch
Receiving records linking ingredient lot codes to suppliers and receipt dates
Batch production records linking ingredient lots to finished product lots
Distribution records linking finished product lots to customers or shipments
A recall plan that identifies who does what and in what order
If any of these are missing or incomplete, the mock recall will surface that gap quickly. That's a useful outcome, but it means the first exercise may be more diagnostic than procedural.
Running the exercise: step by step
Step 1: Select a lot (5 minutes)
Choose a finished product lot at random. Avoid selecting the most recent lot or the one with the cleanest records. The exercise is more valuable when it covers a typical production run, not your best one.
Note the lot code, production date, and batch size.
Step 2: Run the forward trace (30 minutes)
Starting from your distribution records, identify every unit of the selected lot that left your facility. For each destination, record:
Customer or recipient name
Quantity shipped
Ship date
Whether the product has been sold through, is still in inventory, or has been disposed of
Total the quantities. They should equal your original batch size minus any retained samples, internal use, or documented waste. If they don't, identify where the gap is.
Step 3: Run the backward trace (30 minutes)
Starting from your batch production record for the selected lot, identify every ingredient used. For each ingredient, record:
Ingredient name and supplier
Lot code of the ingredient used
Quantity used
Receiving date
Then work back one level further: for each ingredient lot, confirm you’ve a receiving record that links it to a purchase order or supplier delivery. This is the level of traceability a regulator or auditor expects.
Step 4: Calculate your recovery rate (15 minutes)
Tally what you can account for across both traces. A strong result accounts for 99% or more of the selected lot within the two-hour window. Document any gaps and their likely causes.
Recovery rate matters because in a real recall, regulators expect rapid and complete product recovery. Knowing your baseline before an event is the only way to address weaknesses proactively.
Step 5: Document and debrief (30 minutes)
Write up the exercise results: lot selected, time taken, forward trace results, backward trace results, recovery rate, and any gaps identified. This documentation is part of your food safety records and may be reviewed during an inspection.
If gaps were identified, note the corrective action and target date for resolution. A mock recall that surfaces a gap and results in no corrective action is a missed opportunity.
What good looks like
A well-functioning traceability system produces complete forward and backward traces in under two hours, with a recovery rate at or near 100%. The records are retrievable without significant searching, the lot code linkages are clear, and the documentation tells a coherent story from supplier to customer.
If your first mock recall takes four hours and accounts for 85% of the selected lot, that's useful information. It's also fixable. The same result during a real recall, with a regulator waiting, is a different situation.
That’s one exercise. A single mock recall answers one question — could we trace this lot? A program answers a harder and more useful one: can we trace anything, in any of the scenarios that would actually put you into a recall, with whoever happens to be working that day. Most small manufacturers run the first kind. Here’s what the second looks like, and why the difference matters when a supplier calls on a Friday afternoon.
Why one exercise a year is not a program
The annual mock recall has a predictable shape. Someone in QA picks a finished lot, usually a recent one with tidy records, traces it forward and backward, gets a good recovery rate, writes it up, and files it. The exercise passes.
Three things about that are worth noticing.
The lot was chosen by the person running the test, which means it was chosen to be traceable. The scenario started from finished product, which is the direction that is easiest to trace and the direction real recalls least often start from. And it was run by the person who designed the system, who knows where every record lives without looking.
None of that is dishonest. It is just a weak test, and a weak test that passes tells you almost nothing.
The scenario you’re probably not testing
For a small manufacturer, the most likely route into a recall is not your own finished product testing positive. It is a supplier notifying you that an ingredient lot you received is affected.
That scenario runs the trace in the harder direction. You start with a supplier lot code on an email, and you’ve to answer: did we receive that lot, when, how much, what did we make with it, which finished lots did it enter, and where did those go? One ingredient lot can fan out across weeks of production and dozens of customers.
Call this an ingredient-initiated mock recall. It is the exercise most worth running, and the one most rarely run.
The mechanics are the same as a standard mock recall, reversed. Pick an ingredient lot you received four to six months ago. Find every batch that consumed it. From each of those batches, find every finished lot. From each finished lot, find every shipment. Total it, and compare against what you received.
The first time most manufacturers try this, the trace breaks at the batch record, because batch records commonly note the ingredient but not which lot of it.
Building the annual schedule
Two exercises a year is enough for most single-facility operations, provided they are different exercises. A workable pattern:
First half of the year: a finished-product mock recall, run by someone other than the person who maintains the records
Second half: an ingredient-initiated mock recall, starting from a supplier lot code
Once every second year: the same exercise run cold, with no advance notice, during a normal production day
Vary three things across exercises: the direction of the trace, the age of the lot, and the person running it. Holding all three constant is how a program becomes a ritual.
Write the schedule into your recall plan rather than keeping it in someone's calendar. An inspector asking about recall readiness is asking about a documented, repeating activity with results attached, not a one-off.
Where traces actually break
Across small food operations, traces tend to fail in the same handful of places. Knowing them lets you design an exercise that goes looking for trouble rather than avoiding it.
The batch record names the ingredient but not the ingredient lot, so backward tracing stops at "we used sea salt" rather than "we used sea salt lot 4471"
Rework and carryover are undocumented, so product from one batch that went into a later batch is invisible to the trace
Bulk ingredients from multiple deliveries are commingled in one bin or silo with no record of when one lot ended and the next began
Split shipments to the same customer are recorded as one line, so quantities do not reconcile
Samples, staff purchases, farmers market sales, and comped product leave the building without a distribution record, which is where most of the missing percentage in a recovery rate hides
Distribution records live in the accounting or shipping system and are not linked to lot codes at all, so you can see who bought a product but not which lot they received
That last one is the most common structural failure, and it is the reason recovery rates stall around 85 to 95 percent even in operations that feel well organized.
Turning results into corrective actions
A mock recall that identifies a gap and produces no corrective action has cost you two hours and bought you nothing. Worse, the record of it sits in your file showing an inspector that you knew about a weakness and did not act.
Each gap should generate a corrective action with an owner and a target date, in the same log you use for production deviations. If your recovery rate was 92 percent, the missing 8 percent has a cause, and the cause has a fix: a field added to a form, a record linked to a lot code, a procedure that says samples get logged.
Re-test the specific gap at the next exercise rather than waiting for it to resurface on its own.
The short version
One mock recall a year, run by the person who built the system, on a lot they chose, in the easy direction, is a ritual rather than a test. A program varies the direction, the age of the lot, and the person running it, and it schedules the ingredient-initiated scenario that most real recalls actually start from.
The output that matters is not the recovery rate. It is the list of corrective actions the exercise generated and the evidence that you closed them before the next one.
Crown QMS holds lot genealogy, receiving, batch, and distribution records in one place, so a trace runs against live data rather than a folder of spreadsheets, and mock recall reports export as a PDF for your file.
Frequently asked questions
How often should a small manufacturer run a mock recall?
Once per year is a reasonable baseline for a small, stable operation. If you introduce new products, change your lot coding system, add distribution channels, or make significant changes to your recordkeeping, run an additional exercise after the change. The goal is to confirm your systems work under current conditions, not just the conditions that existed when you last ran the exercise.
Does a mock recall need to be witnessed or formally documented?
Under SFCR, your recall plan and its testing are part of your food safety program records. The mock recall results should be documented and retained. There's no requirement for a third-party witness, but the written record of the exercise (including any gaps and corrective actions) should be available for inspector review. Under FDA's Preventive Controls rule, similar documentation expectations apply as part of your written Food Safety Plan.
What's the difference between a mock recall and a real recall?
A mock recall is an internal exercise with no external notification, no product removal from the market, and no regulatory reporting. A real recall involves notifying affected customers, removing product from distribution, and in most cases reporting to CFIA or FDA depending on jurisdiction and the nature of the hazard. The traceability skills the mock recall tests are the same ones a real recall demands, which is exactly why the exercise has value.
