Product recall process flowchart (recall vs withdrawal)
Product recall process flowchart: health risk assessment, recall or withdrawal, classification, regulator and public notice, retrieval, effectiveness check.
How it works
Rename the lanes to your real roles
Replace Quality, Recall committee, Regulatory affairs, Supply chain / Logistics and Customers and public with the functions you actually have. Small manufacturers usually merge Regulatory affairs into Quality; distributors often split Supply chain into a warehouse lane and a customer service lane. Keep the Customers and public lane even though it holds only two nodes, because it marks the boundary where you stop controlling the process and start measuring it.
Write the recall-or-withdrawal test onto the decision
The "Recall required?" node is the one that decides how far the action reaches, so it needs criteria rather than judgement. The usual test is whether the product has reached end users or consumers and whether there is a risk to health. Record who may make the call, who deputises out of hours and how they are contacted, because this decision is often needed on a Friday evening.
Set the notification window and name the authority
Add your own deadline and named regulator to "Notify the regulator within the window". Windows are sector-specific and sometimes fixed in law, so cite the instrument you work to rather than a rule of thumb, and note whether the clock starts at awareness or at confirmation. Keep the submitted notification with the recall record, since it is the first thing an inspector asks for.
Decide in advance what triggers a public notice
"Public notice required?" is drawn as a decision because a withdrawal that stops at the trade and a recall that must reach consumers need different communications. Write down the trigger, who signs off the wording, which channels are used (press notice, point-of-sale posters, website, direct contact) and who handles the media and customer enquiries that follow.
Fix the effectiveness target before the recall starts
"Reconcile returns against quantity distributed" only works if the distribution list is complete and the target is agreed up front. Decide the contact level (every consignee, a defined sample, or none for the lowest-risk cases) and the recovery percentage that counts as effective. Then define what "Extend the recall and re-notify" means in practice, so the shortfall branch is a real action rather than a shrug.
Close the loop into CAPA and rehearse the map
The chart ends at "Issue closure report and CAPA" because the recall contained the problem but did not remove its cause. State what the closure report must contain: the risk assessment, the classification, quantities distributed and recovered, the disposition of recovered stock, and the CAPA reference. Then rehearse the map in a mock recall at a set frequency, and version it like a controlled document. If you run it in QueryChart, the approval workflow and version history keep the current authorised version and its sign-off available as evidence.
Frequently asked questions
What are the steps in a product recall process?
A complete recall runs: detect and verify the signal, whether it came from complaints, routine or stability testing, or a supplier alert; retain samples and record the date of awareness; convene the recall committee; assess the health risk; confirm reporting obligations; trace the affected batches and build a distribution list; decide whether the situation calls for a recall, a withdrawal or no market action; classify the recall by risk level; notify the regulator within the applicable window; notify direct customers and distributors; decide whether a public notice is needed and publish it; retrieve and quarantine returned product; reconcile what came back against what was distributed; extend the recall if the recovery falls short; dispose of recovered stock; and close with a report and a CAPA. The chart above keeps the branches teams argue about as real decisions rather than footnotes.
What is the difference between a product recall and a product withdrawal?
A withdrawal removes product from the supply chain while it is still in the trade, so the action stops at your direct customers and distributors. A recall retrieves product that has already reached end users or consumers, which is why it normally needs a public notice and a route for consumers to return or dispose of what they hold. EU and retained UK food law makes this split explicit: Regulation (EC) 178/2002 requires a food business to withdraw non-compliant food from the market, and to inform consumers and recall it when the product has already reached them. In this chart the distinction is the "Recall required?" decision, and it changes the path: the withdrawal branch skips risk classification and the public notice and goes straight to notifying customers and distributors.
How are product recalls classified by risk level?
Classification follows the health risk assessment and sets how urgent and how public the action has to be. In the United States, FDA classifies recalls in three levels: Class I where there is a reasonable probability that use will cause serious adverse health consequences or death, Class II where the consequences are temporary or medically reversible or the probability of serious harm is remote, and Class III where use is not likely to cause adverse health consequences. Other regimes use comparable risk bands and add a depth of recall, meaning how far down the chain the action reaches: wholesale, retail or consumer level. Whichever scheme applies to you, put your own definitions in the comment on the classification node, because consistency matters more to an inspector than which label you chose.
How do you check that a recall was effective?
An effectiveness check verifies that the people who received the product were reached, understood what to do and acted on it. It is measured against the distribution list, not against the volume that happened to come back, which is why an incomplete list caps the recovery rate you can ever evidence. FDA's recall guidance describes effectiveness checks at defined levels of coverage, ranging from contacting every consignee down to no checks for the lowest-risk situations, and the level is normally agreed at the start of the recall rather than at the end. Set both a contact level and a recovery target before the notifications go out, and decide what happens when the target is missed. In this chart the Shortfall branch on "Recall effective?" extends the recall and re-notifies rather than allowing the record to be closed on a low return rate.