The Safe Food for Canadians Regulations call it a recall simulation, and most people call it a mock recall. At least once every 12 months you run one based on your written recall procedure, write down how it was run and what it found, and keep that record for two years. That's section 84(2). The regulation doesn't say how to run it or what counts as a pass, so the template below covers what an inspector will want to see.
The short version
- Pick a real lot, one you've already shipped, and pretend it has to come back.
- Trace it both ways with your normal records: its ingredients back, its customers forward.
- Time it, and account for every unit: made, still here, shipped, sold, wasted.
- Write it up the same day, fix what went wrong, and file it for two years.
- Do it again before 12 months pass. TaroStack times the drill, keeps the record and reminds you when the next is due.
What the regulation asks, exactly
Section 84(2) says the operator must, "at least once every 12 months", conduct a recall simulation based on the recall procedure, and prepare a document that "sets out the details of how the recall simulation was conducted and the results of the simulation", kept for two years after it's completed.
Three things follow from that wording. It's based on your recall procedure, so you need the procedure first. It's every 12 months, not every calendar year, so a drill in March 2026 and the next in May 2027 is a gap. And a real recall counts: the CFIA's inspection procedure tells inspectors to look for "the latest simulation or actual recall."
Neither the regulation nor the CFIA's recall guide sets a time limit or a percentage to hit. Your buyers or a certification scheme may set their own, and it's worth asking them. My suggestion is to aim for every unit accounted for, in less time than the CFIA's 24-hour records deadline. That's my bar, not the regulation's.
This is general information, not legal advice. Your CFIA inspector has the last word on what applies to you.
How to run one
- Choose the scenario. Pick a finished lot shipped a few weeks ago, or start from an ingredient: "the mill says
rye flour lot
R4471may contain metal." Starting from an ingredient tests more, because it has to go through your batch records. - Start the clock when the person conducting recalls is told, the way they'd hear about a real one.
- Trace back. Which ingredient lots went into it, and from which deliveries?
- Trace forward. Which customers received the affected lots, how much, and when? Pull their contact details.
- Count what's still here. Walk to the shelf and count it. Don't trust the spreadsheet.
- Reconcile. Made = still here + shipped + sold at retail + wasted or used in rework. If the numbers don't add up, find out why before you stop.
- Test the calls. You don't have to alarm customers, but phone one or two to check that the contact name and number on file still reach a person who could act. Tell them it's a drill.
- Stop the clock when you could hand the CFIA a complete distribution list.
- Write it up and fix what went wrong.
The template
| Field | What to write |
|---|---|
| Date and simulation number | |
| Run by | Name of the person conducting recalls (as named in the procedure) |
| Scenario | What "happened", and the starting lot or supplier lot |
| Started and finished | Time each, and the total |
| Product and lots affected | Each lot code, the quantity made |
| Ingredients traced back | Supplier, their lot, delivery date |
| Still on hand | Quantity by location, counted |
| Shipped | Customer, address, quantity by lot, date |
| Sold at retail, wasted or reworked | Quantities |
| Accounted for | Total of the four above ÷ made, as a % |
| Contact check | Who was called, did the details work |
| Problems found | What was slow, missing or wrong |
| Corrective actions | What changes, who does it, by when |
| Signed | Both people named in the recall procedure |
A filled-in example
A 14-person bakery in Ontario runs its yearly simulation in November, starting from a supplier lot of sesame seeds.
| Field | Entry |
|---|---|
| Scenario | Seed supplier reports sesame lot SS-0912 may contain undeclared peanut |
| Started / finished | 9:05 a.m. / 10:30 a.m., 1 hour 25 minutes |
| Lots affected | Sesame burger buns 20261103L1 and 20261104L1, 600 bags of 8 each made per lot, 1,200 bags |
| Still on hand | 180 bags (freezer 160, retail counter 20) |
| Shipped | 1,000 bags to 9 customers, 7 in Ontario and 2 in Quebec |
| Sold at retail, wasted | 14 sold at the counter, 6 bags damaged and thrown out |
| Accounted for | 180 + 1,000 + 14 + 6 = 1,200 of 1,200, 100% |
| Contact check | Two customers called: one number reached a buyer who left in August |
| Problems | Batch sheet for 4 Nov had the seed lot written as "0912", found by matching delivery dates |
| Corrective actions | Update the customer contact (owner, this week); seed lot written in full on batch sheets (production manager, from today) |
The two problems are the useful part. That's what a simulation is for: finding the out-of-date phone number in November instead of during a real recall.
Where this stops working
On paper, the slow part is always the same. Steps 3 and 4 mean reading through batch sheets for a lot number somebody abbreviated, then matching dates against invoices. And the drill itself is easy to forget: it happens once a year, the person who ran the last one may have left, and nothing reminds anyone. The usual way a business finds out its simulation is overdue is that the inspector asks for it.
How TaroStack does it
TaroStack turns the simulation into the same few clicks as a real trace, and keeps the record the regulation asks for.
Start a traceability drill from any lot, yours or a supplier's. TaroStack starts the clock, and the trace it opens is the one you'd use in a real recall: every batch the lot reached, how much is still on your shelves and where, and every customer who received it, with dates and quantities. Because each batch already recorded which ingredient lots went into it, there's no hunting through sheets for "0912". You count what's on the shelf, check it against what TaroStack shows, add a note on what went wrong, and finish. The drill is saved with its start, finish, how long it took and what it found, which is the "how the recall simulation was conducted and the results" that section 84(2) asks you to keep.
The record carries the names of the two people your recall procedure names.
TaroStack reminds you before 12 months pass without a drill, so the gap between March and May never happens. And nothing is deleted, so two years of drills are there when an inspector asks.
| On paper | In TaroStack | |
|---|---|---|
| Tracing the lot | Batch sheets and invoices, by hand | One trace from any lot |
| The time it took | Somebody's watch | Recorded |
| The record | Written up afterwards, if at all | Saved with the drill |
| The next one | Whoever remembers | A reminder before 12 months |
Traceability drills are on every TaroStack plan, from $9.99 a month. It's free for 14 days. If you'd like a hand setting up, ask, and we'll do it with you.
