Mock Recall: A 4-Hour Trace Drill With Pass Marks and a Scoring Sheet
In short: A mock recall is a timed rehearsal of a real one: pick a lot, trace it one step back to the supplier and one step forward to every first customer, reconcile the quantities, test the contact list, and score the result. Canada requires traceability records within 24 hours of a request, the US food traceability rule sets the same limit for the foods on its list, and certification schemes and UK guidance expect the plan to be tested or reviewed at least once a year.
- A well-run drill fits in about 4 hours, which leaves 20 hours of margin before the 24-hour regulatory limit.
- Score time, quantity reconciliation, records format and contacts, not just "did we find it".
- Example plant: line-level lot records shrink the scope of one cheese lot from 80,000 lb to 24,000 lb, worth $87,231 to $343,448 per event.
Guide, updated 14 November 2026. Sources are listed at the end.
Certification schemes and UK guidance expect the recall plan to be tested at least once a year, so many plants run one drill a year for the audit. The better reason is that a mock recall is the only time you find out, without a regulator waiting, how long your records take to answer three questions: where did this lot come from, where did it go, and does the quantity add up. This guide gives a drill a quality team can run in a morning, the pass marks that regulators and standards state, a scoring sheet, and what a slow trace costs in product. The drill tests the records, so the thing to get right before the first run is a lot traceability data model that links inputs, runs and shipments.
What a mock recall has to prove
A mock recall is not a search exercise. It has to show that the plant could carry out the steps of a real recall, with the records it holds today. The Canadian Food Inspection Agency's (CFIA) recall procedure guide for food businesses sets out twelve steps, from assembling the recall team to the post recall review. At step 2 the CFIA recall coordinator may ask for the food name, brand, size and lot codes, the total quantity affected, the quantity that has left your control, where it went and on which dates. That list is a good definition of what a drill must produce.
The pass marks below come from pages we read. Where a mark is our own choice, the table says so.
| What is tested | Pass mark | Where it comes from |
|---|---|---|
| Records handed to the regulator | Within 24 hours of a request | Canada: Safe Food for Canadians Regulations, s. 91. US: FDA food traceability rule, for foods on its list |
| Format of those records | One plain-text file that standard software can open (Canada); an electronic sortable spreadsheet (US) | CFIA traceability guidance; FDA food traceability rule |
| Reach of the trace | One step back to the supplier, one step forward to the first customer | CFIA traceability guidance; SQF Institute tip sheet 15 |
| Quantities | Shipped to customers plus still in your control equals the total affected | CFIA recall guide, step 6 |
| How often | All elements of the recall plan tested at least once a year | SQF Institute tip sheet 15; UK FSA and FSS guidance (review at least annually) |
| Contacts | Team, suppliers and buyers listed, with after-hours numbers | CFIA recall guide, step 1; SQF Institute tip sheet 15 |
| Drill length | About 4 hours | Our choice: a morning's work, with 20 hours left before the 24-hour limit |
The Canadian rule is in force now. The CFIA's traceability guidance says documents must reach the agency within 24 hours of a request, sooner if it sees a risk to health, and if electronic, in a single plain-text file that standard commercial software can import. The US rule is coming. The FDA's food traceability rule requires records within 24 hours and an electronic sortable spreadsheet when the agency needs it during an outbreak or recall. The FDA's page says Congress directed it not to enforce the rule before 20 July 2028, and the rule covers foods on the Food Traceability List. For a plant whose products are not on the list, 24 hours is still the right design target: it is the limit Canada already applies.
Certification schemes set their own tests. The SQF Institute's tip sheet on traceability and recall (2018) tells sites to test all elements of the recall plan at least once a year, to start the mock incident at either end of the process flow, and to be careful that the test does not trigger real publicity. Retailer schemes such as BRCGS and IFS have their own traceability clauses. Check the test frequency and time limit in the issue you are certified to; we could not read those texts for this guide, so we quote no number from them.
The UK's FSA and FSS guidance on traceability, withdrawals and recalls (March 2019) says traceability systems should be reviewed at least once a year, or sooner after a significant change, and that the recall plan should be reviewed annually. It suggests an incident mock exercise involving business customers, because a plan is easier to challenge after a mock than during a real event.
Before the drill: choose a scenario that can fail
A drill that always passes is picking easy lots. The test is only useful if the scenario crosses the places where records break. Choose one of these, and change it each time.
- An ingredient or packaging lot, traced forward. The SQF tip sheet says to start at either end of the flow. Starting from a supplier lot is harder, because it usually fed several runs, lines and products. The CFIA guide warns that ingredients and packaging both trigger recalls.
- A finished lot that spans a shift change or a changeover. Two crews, two sets of paperwork, one lot.
- A lot that took rework. Rework carries the source lot's history into later lots, and it is often logged on a separate sheet.
- A lot sold to a customer who repacks. Your code may not survive their pack. The case below shows why this matters.
- A lot with stock in a third-party warehouse. Off-site stock is where reconciliations come up short.
Decide three things in advance and write them on the drill sheet: who knows the scenario (one person, the drill lead), whether customers will be contacted (if so, tell them it is a drill; the SQF tip sheet warns against triggering publicity), and the start time. Start without warning, on a normal working day, ideally on the shift with the thinnest quality cover.
The 4-hour drill, step by step
Times are targets for a mid-sized plant with one site. Each step has an owner and an output that goes into the drill file.
- 0:00 to 0:15. Start and assemble. The drill lead hands the recall coordinator one fact: a lot code, or a supplier lot number, and a reason ("supplier reports possible contamination"). The coordinator calls the team on the contact list, using the numbers on the list, not phones' own contacts. Output: start time, who answered, who did not.
- 0:15 to 1:00. Trace back. From the finished lot to every input: ingredient lots, packaging lots, processing aids, rework. From a supplier lot, to the receiving record, the quantity received and where it was stored. Output: a list of input lots with supplier, receipt date and quantity.
- 1:00 to 2:00. Trace through the plant. Which line, which run, which hours used the input. This is where paper systems slow down, because the receiving log says when a lot arrived, not when it was opened at the line. Output: line, start and end time of each run that used the lot, and the finished lots it produced.
- 2:00 to 2:45. Trace forward. From each finished lot to every first customer: name, address, contact, date shipped and quantity per lot. Include samples, staff sales and donations. Output: a distribution list in the format the CFIA guide describes at step 6, by food and lot code.
- 2:45 to 3:15. Reconcile. Quantity made, minus quantity shipped, minus quantity in stock on site and off site, minus documented samples and waste. The CFIA guide's tip is the test: if shipped plus still in your control is less than the total affected, review the customer list or the inventory. Output: the reconciliation with the percentage accounted for and the gap explained.
- 3:15 to 3:40. Test the contacts. Call two customers' recall contacts and one supplier, and say it is a drill. Check the after-hours numbers for the team. Output: who was reached and how long it took.
- 3:40 to 4:00. Package and score. Put the records in one file, in the format your regulator would ask for, and score the drill on the sheet below. Output: score, total time and a list of fixes with owners and dates.
If the drill passes the 4-hour mark, keep going and record the real end time. A drill that takes 7 hours is useful data. A drill stopped at 4 hours and marked "nearly done" is not.
Scoring sheet
Print this, fill it in during the drill, and keep it with the drill file. The point values and the time and percentage bands are our proposal; the 24-hour limit and the reconciliation test come from the regulators cited above.
| Item | Full points | Partial | Max |
|---|---|---|---|
| Trace back: every input lot with supplier, receipt date and quantity | All inputs found | 5 if one input or rework is missing | 10 |
| Trace through the plant: line and time window per run | Line plus start and end time | 5 if only the production day is known | 15 |
| Trace forward: every first customer with date and quantity per lot | All customers, incl. samples and off-site stock | 8 if one customer or location is found late | 15 |
| Quantity reconciliation | 99% or more, gap explained | 10 for 97 to 99%; 5 for 95 to 97% | 20 |
| Time to a complete answer | Under 4 hours | 8 for 4 to 8 hours; 3 for 8 to 24 hours | 15 |
| Records in one file a regulator can read | One plain-text or sortable spreadsheet file | 5 if scanned paper or several files | 10 |
| Contacts reached, incl. after hours | All on the first try | 5 if one number is wrong or dead | 10 |
| Fixes logged with an owner and a date | Yes | 0 | 5 |
| Total | Fail regardless of score: over 24 hours, or under 95% reconciled | 100 | |
Track the score and the time from drill to drill. A plant that runs quarterly traces between annual mock recalls sees whether its fixes work, rather than finding the same gap a year later.
Case: the Clover Hill Dairy cheese recall, June 2026
This case is about the trace, not the cause. Nothing below is a judgement on how the companies involved handled it. The facts are from the FDA's outbreak investigation page (last updated 26 August 2026) and the Maryland Department of Health advisory of 14 June 2026.
| Date (2026) | What happened, as the FDA and Maryland report it |
|---|---|
| 13 May | A New York county health department reports two related Listeria illnesses to the state agriculture department; both people bought food at one retailer |
| 14 May, then | The state health department confirms both ate requeson, a soft cheese like ricotta, from that retailer. Of five samples taken there, one, repacked by the retailer, tests positive |
| 27 May | An inspection at the retailer's cheese distributor identifies the maker as Clover Hill Dairy of Mechanicsville, Maryland |
| 30 May | Maryland suspends the facility's operating license |
| 3 June | Clover Hill Dairy recalls its requeson and soft ricotta |
| 5 June | The distributor recalls 1-lb clamshells repacked from batch 2AA051526, sold in New York from 15 to 28 May |
| 18 June | The recall expands to all Clover Hill Dairy cheese on the market, after a second advisory from Maryland on 14 June |
| 26 June | A further firm recalls repackaged requeson with an expiry date of 10 July 2026, distributed from 11 May to 1 June |
| 26 August | The FDA closes its investigation: 15 illnesses in 4 states, 14 hospitalisations, 1 death |
The positive batch was found in an unopened 18-lb bucket with a sell-by date of 14 June 2026 and batch number 2AA051526. The FDA's list of recalled products runs from 3.2-oz snack packs to 40-lb sizes. The expanded recall notice of 18 June says products may be relabeled under other brand names, and tells buyers to look for the manufacturer permit number 24-128 on clamshell containers.
What this case teaches about running a drill
- Fourteen days from retailer to maker. The illnesses were reported on 13 May and the maker was named after an inspection at the distributor on 27 May, because the cheese had been repacked at retail. The forward trace ends at the first customer, but the plant's code has to survive what that customer does. Put one scenario in each year's drill where the customer repacks, and check that your bulk containers carry the plant number and batch in a form the repacker can copy onto their pack.
- Permit number 24-128 was the identifier that survived relabeling. Brands changed downstream; the plant number did not. In the drill, take a finished pack from a customer's shelf, if you can, and see whether the plant and lot can still be read from it alone.
- One 18-lb bucket became 1-lb clamshells. Batch 2AA051526 left the plant in bulk and reached shoppers in a different size, sold from 15 to 28 May. A drill that only lists cases shipped misses the question a regulator asks next: which customers break bulk, and do you know it?
- One product on 3 June, all cheese on 18 June. The notices do not say the expansion was a records problem. With an environmental sample matching the outbreak strain, the site itself was the common factor. The drill question is the shared-path one: for the lot you chose, which other products shared the room, the equipment and the crew, and can your records show it within the morning?
- Dates and expiry per customer. The later notice by a repacking firm gives an expiry date of 10 July and a distribution window of 11 May to 1 June. That is what a downstream business needs from your records to write its own notice. Score the forward trace on whether it gives ship dates and expiry dates per customer, not just a customer list.
The next case will be added under this one, with its date in the heading.
Worked example: what a slow trace costs
Example plant. Every figure in this section is example data except the two unit values, which are sourced. The plant is not any company named above, and the sums are not an estimate of any recall.
A plant cuts and packs cheddar from 40-lb blocks on two lines, two 8-hour shifts a day. Each line packs 1,250 lb an hour, so 20,000 lb a day. On Monday it receives supplier lot B-0412: 600 blocks, 24,000 lb. The quarterly drill starts from that lot: the supplier has reported a possible problem.
The same lot, traced with two levels of record:
| Drill A: day records | Drill B: line records | |
|---|---|---|
| What the plant records | Receiving log; daily batch sheet lists the block lots used that day | Block pallet scanned at the line infeed when opened, with time; pack code carries line and hour |
| What the trace finds | Lot B-0412 used on Tuesday and Wednesday, line unknown | Line 1, Tuesday 06:00 to 22:00 and Wednesday 06:00 to 09:12 |
| Product in scope | 2 days x 2 lines x 20,000 lb = 80,000 lb | 19.2 run hours x 1,250 lb = 24,000 lb |
| Time to a complete answer | 6 h 40 min | 1 h 55 min |
| Reconciled | 96% of the two days' output | 99.6% of the lot; the gap is logged trim and samples |
| Score on the sheet | 63 (passes, with fixes) | 95 |
Drill A is not a strawman. Daily batch sheets that list the lots used meet the one-step-back, one-step-forward rule, and a 6-hour trace is well inside the 24-hour limit. Its weakness is what it cannot rule out: without the line and the hours, both lines for both days stay in scope.
The value per pound, from two sources
- Low: what the cheese is worth at wholesale. The USDA Agricultural Marketing Service's announcement of advanced prices and pricing factors, dated 21 October 2026, gives a September 2026 cheese price average of $1.5577 a pound, one of the product price averages used in federal milk order formulas. It is a bulk commodity price, below what a cut and wrapped pack sells for.
- High: what a shopper pays. The Bureau of Labor Statistics average retail price for natural cheddar was $6.133 a pound in September 2026, US city average.
The sums, against the realistic Drill A baseline:
- Product kept out of scope: 80,000 minus 24,000 = 56,000 lb, or 112,000 packs of 8 oz.
- At wholesale: 56,000 x $1.5577 = $87,231.
- At retail: 56,000 x $6.133 = $343,448.
That is per event, on one lot. The range leaves out freight, retailer charges, destruction, the investigation and lost sales, so it is a floor. It also assumes the product had shipped. If it is still in the plant, the same records turn a recall into a hold of 24,000 lb instead of 80,000 lb, and two days' quality in OEE terms is 70% sellable instead of zero.
Time has its own cost. The 4 h 45 min between the two drills is the time a real team would spend before it could tell customers which product to pull, while product keeps selling. The faster answer does not change what was made, but it changes how much of it reaches shoppers before the notice does.
Book a demo and see OEE and losses per line and shift.
Where drills usually lose time
| Where it stalls | What you see in the drill | The fix |
|---|---|---|
| Receiving to line | The lot's arrival is logged, its use is not | Log the input lot at the line when a pallet, bag or reel is opened, with the time |
| Shift handover | Two sheets for one run; the second crew's lots are missing | One run record that both shifts write to |
| Rework | Rework added to a later batch with no source lot | Rework logged as an input with source lot, quantity and time |
| Code reading | Different sites read the same code differently | One decoder page for every code format |
| Off-site stock | The reconciliation comes up short | Third-party stores on the stock report by lot, checked in the drill |
| Records format | Answer exists, but in scans and five spreadsheets | An export in one file, in the regulator's format |
| Contacts | A customer's recall contact has left; after-hours numbers ring out | Contact list reviewed after each drill and each staff change |
Most of these sit in the plant, between receiving and dispatch. In the example plant the trace back to the supplier and forward to the customers took under an hour each in both drills, because purchase orders and invoices already carry lot numbers. The middle step, which line and which hours, is the one that turns a 2-hour drill into a 7-hour one. It is also the step that decides the scope; how to keep a food recall narrow covers the lot codes and time windows that set it. Production tracking software that records runs, orders and counts per line closes that gap with data the plant collects anyway. The difference between tracing lots and tracing what they became is covered in traceability vs genealogy.
Which record setup fits your plant
There are three usual ways to hold the records a drill reads. The times are example data, in line with the example plant above; your own first drill is the real measure.
| Setup | Smallest scope it can prove | Drill time (example data) | Fits a plant that | Ask before you choose |
|---|---|---|---|---|
| Paper batch sheets and receiving log | A production day, all lines in the room | 6 to 7 hours | Runs one product per room per day, or has very short shelf life and small daily volume | Can we find last month's sheets in 15 minutes? Who reads the handwriting at night? |
| Spreadsheets fed from paper | A production day per line, if the line is written down | 3 to 5 hours | Has one person who owns the file and keeps it current | What happens when that person is on holiday? Is the file the one the regulator would get? |
| Input lots logged at the line, plus line data (runs, stops, counts) | The hours of one line that used the lot | Under 2 hours | Runs several products or lines from shared inputs, or ships to customers who repack | Does the system export one file in the regulator's format? Does it record the time an input was opened, not just received? |
The money in the worked example is the case for the third row: one event on one lot was worth $87,231 to $343,448 of product kept out of scope. A plant that runs one product a day in one room gains less, because the day is already the natural lot.
How the US FDA frames a mock recall
The FDA's guidance on initiation of voluntary recalls (March 2022, non-binding) says a firm that expects complex recalls may want mock recalls to verify its readiness, and that they familiarise staff with the process. It also suggests setting metrics for the recall plan and changing procedures or training if a mock or actual recall falls short. On coding, it says firms should code products well enough to allow positive lot identification, and that coding helps a firm define and limit the scope of a recall. Where a product carries no code, the scope becomes the units distributed in a date range. That last point is the cost of weak coding in one line.
What quality, production and maintenance managers do next
This week
- Quality: pull last year's mock recall file and write down its total time and its reconciliation percentage. If neither was recorded, that is the first fix.
- Production: pick one input lot used last week and ask the shift lead to name the line and the hours it ran, from records only. Time it.
- Maintenance: list which equipment is shared between products in the room with the highest volume: fillers, slicers, conveyors, tools. The drill's shared-path question needs that list.
This quarter
- Run the 4-hour drill with an unannounced start, using a supplier lot that crosses a shift change.
- Score it on the sheet and set a target for the next drill: time down, reconciliation up.
- Log input lots at the line on the highest-volume line first, with the time each pallet or bag is opened.
- Build the export your regulator would ask for, one file, and test that it opens on another computer.
- Add a repacking customer to next year's scenario list, and check what your bulk labels carry.
- Count held product as quality loss in OEE, so a wide hold is visible next to speed and downtime.
The line record is the part day-level records are missing. OEE software records run time, stops and counts per line automatically, and a light MES ties those hours to orders and products, which is what the trace-through-the-plant step needs.
How Fabrico fits
Fabrico is a cloud OEE and light MES platform, not a recall or food safety system: lot records, supplier approval and recall procedures stay with your quality team. What it adds to a drill is the time record of each line: orders, run hours, stop times and counts, with OEE and losses by machine, line and shift, from the PLC or a Fabrico IoT device, and the reasons operators add on a tablet or phone.
On manual stations, AI cameras (an add-on) count output.
Book a demo and see good output per line and shift.
What people ask next
How long should a mock recall take?
Aim for a complete answer, back to the supplier, forward to every first customer and reconciled, in about 4 hours. That is our target, not a legal one. The legal limit to keep in mind is 24 hours: Canada's Safe Food for Canadians Regulations require records within 24 hours of a request, and the US food traceability rule sets the same limit for the foods on its list. A drill that needs most of a day leaves no margin.
How often should a plant run a mock recall?
At least once a year. The SQF Institute tells sites to test all elements of the recall plan at least once a year, and UK FSA and FSS guidance says to review the recall plan annually, suggesting a mock exercise to do it. Short trace tests between annual mock recalls, for example once a quarter, show whether the fixes from the last drill work.
What is the difference between a mock recall and a traceability exercise?
A traceability exercise checks that the records link inputs, runs and shipments. A mock recall tests the whole recall plan: the team, the contact list, the decision, the trace, the reconciliation and the notice. The SQF Institute's tip sheet says to test the full process, including recall contacts, communications and actions on the affected product.
What percentage of product must a mock recall account for?
The pages we read give a test rather than a percentage. The CFIA's recall guide says the quantity shipped to customers plus the quantity still in your control should equal the total quantity affected; if it is less, review the customer list and the inventory. Our scoring sheet gives full marks at 99% or more with the gap explained, and fails a drill under 95%.
Do we have to contact customers during a mock recall?
Not necessarily, but testing the contacts is part of testing the plan. Call a few customers' recall contacts and say it is a drill. The SQF Institute warns not to trigger real publicity, and the UK guidance suggests mock exercises that involve business customers.
What records should a mock recall produce?
A drill file with the start and end time, the input lots and suppliers, the line and hours of each run, a distribution list by food and lot code with dates and quantities, the quantity reconciliation, the contacts reached, the score and the fixes with owners and dates. Keep the trace export in one file, in the format your regulator would ask for.
Sources
All sources accessed on 14 November 2026. Dates are the publication dates shown on each page.
- Canadian Food Inspection Agency, Regulatory requirements: Traceability for food, 26 August 2025
- Canadian Food Inspection Agency, Recall procedure: A guide for food businesses, 27 February 2023
- US Food and Drug Administration, FSMA Final Rule on Requirements for Additional Traceability Records for Certain Foods, 4 November 2026
- US Food and Drug Administration, Initiation of Voluntary Recalls Under 21 CFR Part 7, Subpart C: Guidance for Industry and FDA Staff, March 2022
- US Food and Drug Administration, Outbreak Investigation of Listeria monocytogenes: Soft Cheese (June 2026), 26 August 2026
- US Food and Drug Administration (company announcement), Clover Hill Dairy Expands Recall to Include All Clover Hill Dairy Brand Cheese, 18 June 2026
- Maryland Department of Health, Consumer advisory expanded for all Clover Hill Dairy cheese products, 14 June 2026
- SQF Institute, Tip Sheet 15: Product Identification, Traceability, and Withdraw and Recall, 2018
- Food Standards Agency and Food Standards Scotland, Guidance on Food Traceability, Withdrawals and Recalls within the UK Food Industry, March 2019
- USDA Agricultural Marketing Service, Announcement of Advanced Prices and Pricing Factors, November 2026 (ADV-1126; rolling URL, replaced monthly; the issue read is dated 21 October 2026), 21 October 2026
- US Bureau of Labor Statistics, Average price: Cheddar cheese, natural, per lb, US city average, series APU0000710212 (rolling data page; September 2026 value used), data extracted 14 November 2026
Related articles

Real Fabrico tiles, sample data.










