Food traceability mock recall: a five-step procedure

A food traceability mock recall procedure tests whether you can identify a specific lot, find every unit you received or made, trace where it went, contact the right people, and prove the math works. Under pressure. With real records. Not with a UPC pulled from a case label and a prayer.

The goal is not to perform theatre for an auditor. It is to find the holes while nobody is sick, no regulator is calling, and your customer-service line is not melting down.

For operations handling foods covered by FDA’s Food Traceability List, the stakes are becoming more concrete. FDA says it is directed not to enforce the Food Traceability Rule before July 20, 2028, but the records architecture is not something you build in a weekend. If FDA requests traceability records, covered entities must be able to make the relevant information available within 24 hours, unless FDA agrees to another reasonable timeframe. During an outbreak, recall, or other public-health threat, that can include an electronic sortable spreadsheet.

So run the drill now. Your future self will have enough problems without discovering that “lot number” means five different things in five different departments.

A mock recall does not test whether your team can find paperwork. It tests whether your paperwork describes reality.

What a mock recall is actually supposed to prove

A food safety recall simulation is a controlled test of your withdrawal and traceability system. You select a hypothetical affected product—usually a real product, real lot, and real shipment history—and follow it in both directions:

• Traceback: Where did the affected product or its ingredients come from?
• Trace-forward: Where did the affected product go after it left your control?
• Reconciliation: Can you account for the full quantity without inventing product, losing product, or sweeping discrepancies under the rug?

This sounds simple because people tend to describe traceability as “one step back, one step forward.” That phrase is useful, but it has been abused into near-meaninglessness.

One-step-back one-step-forward tracking means you can identify the immediate supplier and immediate customer for a product. Fine. But a usable mock recall needs the details that make those links operational:

• product description and pack size;
• internal product code and the actual lot, batch, or production code;
• relevant production, receipt, shipment, and use-by dates;
• quantities received, produced, shipped, held, reworked, used, served, destroyed, or returned;
• locations involved, including warehouses, co-packers, distribution centers, stores, and foodservice sites;
• customer and supplier contacts that still belong to living, employed humans;
• a package-label image where possible, because code formats are routinely misread when someone is rushing.

A UPC is not a lot code. A product name is not a lot code. “The blue bagged salad” is definitely not a lot code. Those identifiers tell you what the product is in broad commercial terms. They usually do not tell you which specific production run is implicated.

For FDA Food Traceability Rule-covered foods, the recordkeeping model goes beyond a casual supplier/customer list. FDA identifies seven Critical Tracking Events: harvesting, cooling, initial packing, first land-based receiving, shipping, receiving, and transformation. The rule requires Key Data Elements linked to those events. In plain English: the system must preserve the chain of custody and transformation details that connect a particular traceability lot to what happened to it.

Not every food business is covered in the same way, and not every product falls under the Food Traceability List. Do not treat this as a universal federal mandate for every operation. Do treat it as the direction of travel. The industry is moving from “we have invoices somewhere” toward lot-linked, retrievable records.

Step 1: Review the recall plan and SOPs before you pick a product

The first move is boring. That is why people skip it. Then they discover during the drill that the recall SOP names a quality manager who left two years ago and a warehouse that no longer exists. Very efficient.

USDA’s mock-recall guidance begins with a review of applicable standard operating procedures. It recommends completing that review at least one month before the planned exercise. That lead time matters because an SOP rewrite discovered on the morning of the test is not preparation. It is evidence.

Pull the documents that govern how product and records move through your operation. Depending on your business, that can include:

1. Recall and withdrawal procedures. Confirm who has authority to initiate a hold, approve external notices, stop shipments, and speak with customers or regulators.

2. Receiving procedures. Verify that incoming lots are recorded with supplier identity, lot or batch information, dates, quantities, and receiving location.

3. Production and transformation records. Check whether you can link finished-product lots to ingredient lots, rework, processing dates, and line or shift details.

4. Shipping and transfer records. Make sure internal transfers count. Product moved from Plant A to Warehouse B did not disappear into the ether just because it stayed under the same corporate logo.

5. Inventory, returns, destruction, and waste records. These determine whether your reconciliation will add up—or whether it will expose the usual black hole between “theoretically in stock” and “physically somewhere.”

Choose a scenario that can hurt a little. Do not select the cleanest, smallest, best-documented lot from last Tuesday because it makes everyone feel competent. Pick a product with enough movement to exercise the system:

• a finished product shipped to multiple customers;
• a lot involving a co-packer or third-party warehouse;
• a product made with more than one controlled ingredient;
• a product transferred between company locations;
• a Food Traceability List food, if your operation handles one and is subject to the applicable requirements.

If you process leafy greens, fresh-cut produce, certain seafood, shell eggs, nut butters, or other listed categories, a mock recall should pressure-test the data fields you will need later. The label lot code alone is rarely enough. You need the connected records behind it.

Build the exercise brief, not a mystery novel

Write a short exercise brief for the recall team. It should identify the scenario without giving away all the answers to the people who will retrieve records.

Include:

• the hypothetical hazard or reason for recall;
• the specific product and target lot;
• the date and time the exercise starts;
• the exercise scope: one facility, multiple sites, supplier traceback, customer trace-forward, or all of the above;
• who is evaluating the exercise;
• what counts as an acceptable record;
• the timed target your business has set.

There is no single federal pass/fail clock for every U.S. food business completing a mock recall. Anyone claiming that every company must finish in a magic number of hours is selling simplicity where it does not exist. Set a target that fits your product, number of sites, data systems, and recall plan. Then make it tougher over time.

Step 2: Validate contacts, roles, and the ugly little details nobody owns

A recall moves at the speed of the least current phone number.

USDA’s five-stage model calls for reviewing the emergency notification contact list before conducting the exercise. Do not merely confirm that names appear in the document. Test the list against reality.

Your recall team needs more than a food safety lead and a generic customer-service inbox. Map the actual decisions that must happen during a live event.

Call the numbers. Send the test emails. Confirm escalation paths for evenings, weekends, holidays, and the deeply inconvenient moment when the recall coordinator is on a flight.

Also verify public-health and regulatory contact information appropriate to your operation. USDA’s Harmonized GAP Program Manual expects audited operations to verify current contact details for affected customers and the relevant public-health authority as part of an annual mock recall exercise. That annual requirement applies within that audit framework; it is not a blanket law requiring every food business in America to run the same annual drill. Details matter. Regulators tend to notice when you flatten them.

The lot code verification steps that prevent false confidence

Before the timer starts, make sure your team can interpret the code format on the selected product. This sounds painfully obvious until someone searches “24105” as a calendar date while the plant uses it as a Julian production code.

Document the logic:

• Where is the lot or batch code printed: primary package, case, pallet tag, or all three?
• What does each character mean?
• Does the code identify a production date, line, shift, kettle, supplier lot, or only a packaging run?
• Can a single finished-product lot contain multiple ingredient lots?
• Is rework included, and if so, how is it linked?
• Are code changes controlled when packaging artwork or production systems change?

Take or retain a photo of the relevant package label during the exercise. USDA’s guidance identifies the product description, product code or lot/batch number or GTIN where available, dates, quantities shipped, and ideally a package-label photograph as useful tracking data. This is not paperwork fetishism. It prevents the classic recall-room argument: “Wait, are we talking about code 24-105 or 24105?”

If your lot code needs an oral history to decode, it is not a traceability system. It is a campfire story.

Step 3: Run the traceback and trace-forward, with the clock visible

Now you conduct the mock recall. Start the clock. Record every action, request, response, and delay.

The selected product should trigger two simultaneous workstreams. One team traces backward through receiving and production records. Another traces forward through shipping, transfer, and customer records. If one person does both, that can work in a small operation—provided that person does not become the single point of failure when they are unavailable.

Trace backward: establish what went into the affected lot

Start with the finished product’s lot code. Retrieve the production record and identify:

• the production date and line;
• ingredients and packaging materials used, where relevant to the scenario;
• supplier lot or batch numbers;
• receiving dates and quantities;
• internal movement or staging records;
• any rework incorporated into the batch;
• quality, hold, or deviation records connected to the lot.

For a simple repacker, the traceback may lead directly to a supplier case lot. For a manufacturer, it may require linking a finished lot to several ingredient lots, then connecting each to receiving records. For a grower-packer, the trail may include harvest, cooling, packing, and shipping events.

The point is not to recreate every molecule’s personal journey. The point is to establish defensible affected-product scope. If you cannot tell whether ingredient lot A, B, or both were used in finished lot X, your recall scope will expand. That may be safer than missing product, but it is expensive, disruptive, and avoidable.

Trace forward: establish where the affected lot went

From the same finished-product lot, pull all outbound records. Include:

• shipments to direct customers;
• transfers to company warehouses or other plants;
• product sent to co-packers, distributors, retailers, restaurants, schools, or institutions;
• samples, donations, promotional shipments, and e-commerce fulfillment;
• product still on-site;
• product placed on hold;
• product already used in production, served, sold, returned, destroyed, or otherwise no longer physically recoverable.

If your product went through a distributor, your direct shipment record is not the end of the exercise. Your immediate customer may be the distributor, but your plan should also test how you obtain downstream information when the scenario requires it.

For traceability-rule-covered foods, your retrieval process should be able to assemble the Key Data Elements associated with the applicable Critical Tracking Events. That is where spreadsheet exports and system interfaces earn their keep. A pile of PDFs may technically contain information. Under a 24-hour FDA request, “technically” is not a winning adjective.

What to capture in the event log

Your traceability exercise documentation should record more than the final answer. Capture the route taken to get there.

At minimum, log:

• exercise start and finish times;
• names and roles of participants;
• product identity and selected lot;
• systems searched and reports generated;
• records requested from suppliers, co-packers, warehouses, or customers;
• time each record was received;
• quantities identified at each location or status;
• communication attempts and confirmations;
• missing, contradictory, or delayed data;
• decisions made and who authorized them.

This log gives you something better than “the exercise went well.” It gives you evidence. It also gives you a map of the places where the process only worked because a veteran employee remembered a workaround from 2019. That person is not a control measure.

Step 4: Reconcile the quantities—and do not celebrate 103%

This is where mock recalls stop being administrative cosplay.

You need to reconcile the amount of selected product delivered, produced, or received against every identifiable disposition. USDA’s example calculates mock-recall effectiveness this way:

The target in the USDA example is 100% reconciliation.

That does not mean you get to congratulate yourselves at 100% and ignore everything else. Reconciliation is one leg of the stool. Contact validation, timely communication, accurate scope, record retrieval, traceback, trace-forward, documentation, and corrective actions are the other legs. A three-legged stool is still a bad place to sit.

But the math matters.

A result below 100% means some product was not identified. Maybe it was transferred without documentation. Maybe a case was received under the wrong lot. Maybe a location performed a manual inventory adjustment that never made it back to the ERP. Whatever the reason, you have an unaccounted-for gap.

A result above 100% is not a gold star. It can mean you captured product that was not actually part of the affected lot because the operation cannot distinguish the recalled product from similar inventory. That is a traceability failure too. Over-inclusive scope may be prudent in an emergency, but it exposes a system that cannot accurately separate lots.

Diagnose the mismatch before writing the corrective action

Do not settle for “human error.” That phrase is corporate fog. Find the mechanism.

Common causes include:

• lot codes scanned at receipt but lost during internal transfers;
• finished product linked to a production date but not the actual ingredient lots;
• mixed pallets where cases from multiple lots share a location;
• rework not assigned a traceable identity;
• inventory units recorded in cases at one step and pounds, kilograms, or eaches at another;
• manual shipping changes made after system documents were printed;
• third-party logistics inventory excluded from the internal count;
• product shipped under an old code after a packaging change;
• customer records identifying SKU but not lot.

Then assign a corrective action with an owner and due date. “Train the team” is not always wrong, but it is usually incomplete. If the system allows a warehouse transfer without lot capture, retraining will not fix the system. It will just make the next failure more politely documented.

A useful corrective action statement is specific: configure lot-required transfers in the warehouse management system; revise co-packer data exchange fields; add a reconciliation field for donated product; retrain receiving staff on code verification; conduct a follow-up mini-drill on the corrected process.

Step 5: Debrief, close gaps, and test FDA-ready retrieval

The final stage is the debrief. This is where most organizations waste the entire exercise by writing “no issues noted,” filing the form, and moving on to the next audit ritual.

USDA’s mock-recall model ends with a debrief. The USDA Harmonized GAP Program Manual also expects the annual exercise for audited operations to be documented and to demonstrate the ability to identify affected product, verify contacts, meet applicable legal obligations, and involve the designated recall team. An actual recall may substitute for the mock exercise in that program. Nobody should be hoping for that substitution.

Run the debrief while memories are fresh. Ask blunt questions:

• How long did it take to identify the correct lot?
• Could the team distinguish the selected lot from adjacent production lots?
• Did every site return a usable inventory count?
• Did records connect incoming lots to finished lots, and finished lots to outbound destinations?
• Were customer, supplier, carrier, warehouse, and public-health contacts current?
• Did the team rely on informal knowledge that is not written into the procedure?
• Could the data be assembled in a sortable electronic format if FDA requested it?
• Which records arrived late, incomplete, or contradictory?
• Did anyone confuse product code, GTIN, UPC, and lot code?

For covered Food Traceability List foods, test the 24-hour retrieval requirement as a separate capability. FDA’s non-enforcement period before July 20, 2028 does not make the data problem disappear; it gives businesses time to fix it before the regulator starts expecting performance.

A practical drill is to have someone not involved in the daily traceability process request a defined set of records. Ask for the data needed to understand the lot history, not just raw transaction dumps. Can the team produce the information, explain the terms, and organize it into a usable electronic sortable spreadsheet if required? If the answer is “our IT person can probably do that,” you have not passed the drill. You have identified a dependency.

Document every corrective action. Then verify completion. Then retest the failure point. A mock recall is a control loop, not an annual stage production.