Crisis Management: What to Do in the First 24 Hours of a Food Safety Incident

By GIFSQ Food Safety Team | October 2026

The call comes at 6:40 in the morning. A lab result, a customer complaint, a regulator’s email — something in your product may be making people sick. What happens in the next 24 hours determines whether the incident becomes a controlled correction or a brand-defining disaster. Speed matters, but frantic speed without a plan makes everything worse: destroyed evidence, contradictory statements, product still shipping while the team argues about what to do.

This article is the plan. It assumes you have a written crisis management procedure — if you don’t, treat this as the outline for writing one this week. The sequence below follows the first 24 hours: contain, investigate scope, decide on recall, notify, and communicate.

The First Hour: Contain, Assemble, and Stop the Bleeding

Minute one priority is physical: stop anything suspect from moving. Place all implicated lots on hold — finished goods in your warehouse, product in transit where you can still intercept it, and ingredients if the hazard may have come upstream. Issue the hold in writing through the system your warehouse actually obeys, not by phone call alone, and confirm it was executed. Every pallet that ships after the hold decision becomes a separate liability.

Assemble the crisis team immediately: QA/food safety lead, operations, senior management, legal counsel, and communications. One person owns the incident log from the first minute — every decision, every call, every result, time-stamped. Secure the evidence: retain samples of implicated lots, pull production and sanitation records for the relevant dates, and preserve CCTV if it exists. The golden rule of hour one is simple: do not destroy anything. Discarding suspect product before the investigation closes destroys your ability to find the root cause and looks like a cover-up to regulators.

Tracing the Scope: Which Lots Are Actually Implicated

With the bleeding stopped, define the scope. Pull production records for the suspect lots: raw material lots used, production dates and lines, rework incorporated, and every customer who received the product. This is the one-up, one-down traceability every food facility is expected to maintain, sharpened by FSMA’s traceability rule (21 CFR Part 1, Subpart S): for foods on the Food Traceability List, key data elements must be producible in a sortable electronic spreadsheet within 24 hours of an FDA request.

Run a mass balance while you trace. Units produced minus units on hold minus units shipped should reconcile; a gap means product went somewhere the records don’t show, and that gap becomes the investigation’s priority. Expand or narrow the scope based on evidence, not comfort — shared lines, shared ingredients, and shared shifts widen it; distinct processes and validated clean breaks narrow it. Document the reasoning either way, because “we decided the other lots were fine” without evidence will not survive regulatory review.

The 24-Hour Traceability Test You Should Already Have Run

If tracing your own product takes your team more than a few hours, the crisis has exposed a system failure alongside the incident. GFSI-benchmarked schemes require at least one mock recall per year, and FDA investigators commonly expect a complete trace within two to four hours. Run the drill before you need it: pick a lot at random, start the clock, and trace it forward to every customer and backward to every ingredient. The gaps the drill reveals — a co-packer who answers emails in days, a spreadsheet nobody updated — are cheap to fix on a quiet Tuesday and catastrophic to discover during a real incident.

Hours 2 to 6: Deciding Whether This Is a Recall

Not every incident becomes a recall, but the decision must be made deliberately and fast. Convene the team with the facts gathered: what is the hazard, how confident are you in the result, how much product is affected, where did it go, and who could be exposed? A confirmed pathogen in a ready-to-eat product with wide distribution is a recall; an unconfirmed customer complaint with no corroborating evidence may warrant intensified monitoring and a hold pending results. When in doubt, consult — your regulatory counsel and the FDA district office would rather discuss a potential recall at hour four than read about it at hour forty.

Understand the classification system because it shapes everything downstream. FDA recall classes reflect health risk: Class I covers situations with a reasonable probability of serious adverse health consequences or death; Class II covers remote probability of serious harm or temporary, reversible effects; Class III covers violations unlikely to cause harm, such as minor labeling defects. Separately, FSMA’s Reportable Food Registry (Section 417 of the Food, Drug, and Cosmetic Act) requires reporting within 24 hours when there is a reasonable probability that an article of food will cause serious adverse health consequences — a clock that runs in parallel with your internal deliberations.

Talking to Regulators Before They Talk to You

Voluntary, early engagement with regulators consistently produces better outcomes than waiting to be found. If the hazard evaluation points toward a recall, contact the FDA district office promptly with a prepared package: product description and lot codes, the nature of the hazard and supporting data, distribution scope, the corrective actions already taken, and your proposed recall strategy including the depth (consumer, retail, or wholesale level) and public notification plan. Regulators distinguish sharply between companies that bring them a controlled situation and companies that bring them a surprise.

Keep legal counsel in the loop on every external communication from this point forward. Factual, documented, and prompt beats clever every time — but “factual” should be reviewed before it leaves the building.

Hours 6 to 24: Controlling the Story Without Hiding Anything

By hour six, customers need to hear from you directly — before they hear it from a regulator, a retailer, or social media. Notify direct accounts first, in order of risk: distributors holding the most affected product get the call before the trade press gets the statement. Every notification carries the same core facts: which products and lots, what the hazard is, what the customer should do (stop sale, segregate, await instructions), and a single contact for questions. Consistency across every channel is non-negotiable; contradictions between what you told a distributor and what your website says will surface within hours.

Designate one spokesperson and route everything through them. Prepare a holding statement early — it acknowledges the situation, states the protective actions underway, and commits to updates — because “no comment” reads as guilt and speculation reads as panic. Monitor social media and news actively; correct factual errors quickly and resist the urge to debate causation publicly before the investigation concludes. Never state a root cause externally until the investigation has actually found it. The 2011 Jensen Farms cantaloupe Listeria outbreak, which killed 33 people in the United States, remains the textbook case of how equipment, sanitation, and facility decisions compound — and how the public record of an incident outlives the incident itself.

After the 24 Hours: Investigation, Corrective Action, and the Look-Back

Surviving the first day earns you the right to do the harder work. The investigation now runs to root cause using the preserved evidence: was it a supplier failure, a sanitation breakdown, a process deviation, or a design flaw? Corrective actions must address the cause, not the symptom — retraining the crew that followed a broken procedure fixes nothing. Verify effectiveness over subsequent production cycles with intensified testing and monitoring before declaring the corrective action closed.

Then hold the look-back while memories are fresh. What did the incident log show about decision speed? Where did traceability stall? Did the crisis team have the authority it needed, or did decisions wait for someone’s approval? Update the procedure with every lesson, re-run the mock recall against the new version, and brief the whole organization — not to assign blame, but to make the next first hour faster than this one. The companies that handle incidents well are rarely the ones that never have them. They are the ones that rehearsed.