When a voluntary recall lands on FDA's public database, the quality failure driving it is already weeks or months old. The Mars Petcare announcement on July 2, 2026 — covering two specific lots of PEDIGREE® Can High Protein Chopped Chicken & Duck Flavor 13.2oz wet dog food — is actually a relatively contained situation by recall standards. Two lots. Voluntary. Self-reported. In my experience, that level of containment doesn't happen by accident. It means someone's quality system was working well enough to catch the problem and scope it precisely.
But "well enough to catch it late" is a different standard than "well enough to prevent it." The product reached distribution before the signal was acted upon. And that gap — between detection and prevention — is worth understanding, because it's the gap that costs manufacturers the most in product loss, investigation hours, and reputational exposure.
This article is about the controls that close that gap. Not the recall itself, but the quality systems and testing protocols that, when functioning correctly, tend to prevent lot-level problems from reaching the market in the first place.
What FSMA Actually Requires of Pet Food Manufacturers
The governing framework for pet food manufacturing changed materially with the FDA Food Safety Modernization Act, specifically through the final rule codified at 21 CFR Part 507: Current Good Manufacturing Practice, Hazard Analysis, and Risk-Based Preventive Controls for Food for Animals. This rule took full effect for large manufacturers by September 17, 2016, and for small businesses by September 17, 2017, with some operations qualifying for an additional year.
What Part 507 actually requires is a written food safety plan that does four things: identifies hazards reasonably likely to occur in each product and process; establishes preventive controls for those hazards with specific monitoring parameters; defines corrective action procedures when monitoring shows a deviation; and includes verification activities confirming the controls are functioning as intended. That four-part structure — hazard identification, preventive controls, corrective actions, verification — is the skeleton of FSMA compliance for animal food.
For canned wet pet food with high-protein animal ingredients, the hazard categories most relevant to a lot-level recall include physical hazards (foreign material — metal, bone, dense plastic), chemical hazards (elevated mineral levels, heavy metals in protein ingredients, pesticide residues), and biological hazards (pathogen survival if thermal processing is inadequate). Each category requires its own preventive control analysis. FDA's 2019 guidance document, Hazard Analysis and Risk-Based Preventive Controls for Food for Animals, gives category-specific examples for each.
Under 21 CFR Part 507, every animal food manufacturer subject to FSMA must maintain a written food safety plan that identifies hazards reasonably likely to occur and establishes preventive controls with documented monitoring, corrective action, and verification procedures. That sentence describes the legal floor. The practical ceiling is a system that actually catches problems before distribution — and the distance between those two standards is where most recalls happen.
According to FDA's Center for Veterinary Medicine, approximately 18,000 animal food facilities are subject to FSMA preventive controls requirements. Yet voluntary recalls in the animal food sector have continued at a meaningful rate across multiple product categories since the rule's full implementation. That tells you something about the gap between having a food safety plan on paper and operating one with consistent execution on the floor.
Where Hazard Analysis Tends to Break Down
The most common failure I see in food safety plans for wet-format, high-protein products isn't a missing hazard category — it's one that's undercharacterized. A manufacturer will identify "physical hazards" in their HARPC analysis and list metal detection as the preventive control. What they miss is the upstream gap: ingredient receiving.
Metal detection at the end of a filling line catches some foreign material. It doesn't catch non-ferrous materials reliably — high-density plastic pieces, bone fragments, certain glass types — and it does nothing about contamination that entered through incoming protein ingredients before your facility even received them. If your chicken or duck meat supplier is introducing physical contaminants upstream, end-of-line detection is a secondary safeguard, not a primary one. Calling it your primary control is the kind of thing that holds up fine until it doesn't.
A complete physical hazard analysis for canned high-protein wet pet food would cover all of the following:
- Incoming ingredient inspection and testing protocols for each protein source, with defined acceptance criteria and hold procedures for non-conforming lots
- Supplier verification records under 21 CFR Part 507 Subpart E, demonstrating that hazards at the supplier are actually controlled — not just that a supplier claims to control them
- In-process monitoring at critical staging points before filling, not only at end-of-line
- End-of-line metal detection with documented sensitivity settings, calibration frequency, and challenge-test records for each shift or run
- Can seam integrity verification after seaming and before the retort — because a compromised seam creates its own post-process contamination pathway that thermal treatment can't fix after the fact
The gap between what a food safety plan says and what actually happens on the floor is where most lot-level recalls are born. The plan has the right language. The execution has a weak link — most often at receiving inspection or at monitoring frequency.
Incoming Ingredient Controls and Supplier Verification
21 CFR Part 507 Subpart E — Supply-Chain Program requires that manufacturers implement verification activities for each supplier of an ingredient that has a hazard requiring a supply-chain-applied control. For high-protein poultry ingredients, the practical minimum looks like this:
- On-site supplier audits at a defined frequency, typically annual for approved suppliers, with records of the audit and any corrective actions taken
- Certificate of Analysis review for each incoming lot, including confirmation that the CoA actually tests the parameters relevant to your hazard analysis — not just general specs
- Incoming lot testing for contaminants specific to that ingredient's known risk profile
- A formal approved supplier list with qualification criteria, re-qualification triggers, and a documented record of verification activities completed
The verification activities have to be real. A CoA from a supplier who doesn't test what's printed on the CoA is not a preventive control — it's paperwork with a legal exposure attached. I've reviewed FDA warning letters issued specifically for supplier verification programs that existed structurally but couldn't demonstrate that verification activities actually occurred. Those letters follow a consistent pattern: the manufacturer had written procedures, but couldn't produce records proving the procedures were executed.
For wet pet food with poultry protein, the biological hazard profile also includes Salmonella. The retort step serves as the kill step — but only if the retort process has been validated for the current formulation, the come-up time is adequate, and deviations are documented and investigated every time they occur. If the thermal process validation was completed several years ago and the formulation or fill weight has changed since, the original validation may not cover current production conditions. That's a gap manufacturers sometimes don't discover until an investigation is already underway.
In-Process Controls for Canned Products: The Thermal Processing Question
Canned wet pet food is thermally processed in retort equipment — pressure vessels achieving sterilization temperatures for a validated time-temperature combination. For low-acid canned products (pH above 4.6), this step is the most critical control in the entire process. A failure here is a safety failure, not a cosmetic quality issue.
21 CFR Part 113 governs thermally processed low-acid foods in hermetically sealed containers for human food. While 21 CFR Part 507 governs animal food, FDA's guidance documentation has made clear that the thermal processing principles underlying Part 113 inform what's expected for animal food products in the same format. In practical terms, that means:
- A scheduled process established by a qualified process authority, on file and available for inspection
- Calibrated temperature and pressure monitoring equipment, with records traceable to each individual production run
- Operator training, documented competency verification, and refresher training on a defined schedule
- Deviation procedures that specify exactly what happens when a retort run doesn't meet scheduled parameters — not "contact supervisor" but an actual written procedure with decision logic, hold requirements, and escalation criteria
A deviation from scheduled thermal process parameters is a serious event under FDA's framework. It triggers a formal deviation investigation, and depending on the findings, may require withholding the affected lot from distribution pending results. If the deviation procedure doesn't exist in writing — or if operators lack clear criteria for when a deviation has occurred and what to do about it — product that didn't receive an adequate kill step can reach distribution. That's not hypothetical. It's the documented basis for multiple high-profile animal food recalls in the last decade.
Can Seam Integrity: The Control That Gets Underweighted
Here's something I think gets consistently underweighted in food safety plans I review: can seam integrity for retorted products. The hermetic seal at the double seam is what maintains sterility after thermal processing. A seam that's too loose, inconsistently formed, or mechanically stressed during transport creates a post-process contamination pathway that bypasses everything the retort accomplished.
Can seam inspection involves both visual monitoring per run and destructive seam evaluation — mechanical tear-down and dimensional measurement against the can manufacturer's specifications. Most manufacturers do this, but inspection frequency and measurement rigor vary considerably. What FDA expects in practice: documented inspection frequency per production run, calibrated measurement equipment, acceptance criteria defined against specifications rather than operator judgment, and trend data reviewed across runs to catch seamer wear before it produces defective seams.
A seamer that's wearing and drifting won't fail dramatically — it degrades gradually, producing borderline seams that pass visual inspection but create risk over time. The only way to catch that drift is consistent dimensional measurement combined with trend analysis across production runs. Manufacturers who treat seam inspection as a compliance checkbox rather than a monitoring system tend to find problems at exactly the wrong moment.
Lot Traceability: The Control That Determines Recall Scope
Two lots versus two hundred lots. When a recall happens, that difference comes down almost entirely to how well the traceability system actually works in practice — and whether hold-and-release procedures run in real time or only on paper.
Under 21 CFR Part 507, animal food manufacturers must maintain records sufficient to identify the lot of finished product, the ingredient lots that went into that finished product, and the direct customers who received each finished lot. That's the regulatory requirement. The execution is only as disciplined as the lot-coding practice at receiving, at in-process staging, and at finished goods.
What fails most often: ingredient lots that get commingled during production without adequate documentation; finished product lot codes that aren't consistently linked back to input ingredient lot codes in a retrievable record; distribution records that capture customer names but not which specific lot shipped to which customer. Any one of those gaps means a manufacturer facing a quality signal can't scope the affected product confidently — and a manufacturer who can't scope confidently has to over-recall, pulling in uninvolved product as a precaution.
The Mars Petcare recall covering two specific lots suggests their traceability system was functional enough to scope the issue rather than cast a wide net. That's a genuine quality infrastructure credit. A voluntary recall limited to two specific lots is only possible when a manufacturer's traceability system is precise enough to identify exactly which finished goods were produced from the affected process run — the kind of lot-level linkage that 21 CFR §507.175 requires but many facilities still execute inconsistently.
The honest diagnostic question for any manufacturer is this: if a quality signal appeared today — a consumer complaint, a lab result, a supplier notification — could you identify every affected finished lot and every customer who received product from that lot within 24 hours? If the answer involves uncertainty about commingled lots or incomplete distribution records, that's worth addressing before the signal arrives.
Preventive Control Categories for Canned Wet Pet Food
| Control Category | Key Regulation | What It Covers | Most Common Gap |
|---|---|---|---|
| Supplier Verification | 21 CFR §507.175 (Subpart E) | Ingredient hazard control at the source | CoAs not verified; audit records absent or not linked to lots |
| Process Controls (CCP) | 21 CFR §507.39(a)(1) | Retort, metal detection, seam integrity monitoring | Outdated thermal process validation; inadequate sensitivity testing |
| Sanitation Controls | 21 CFR §507.39(a)(3) | Equipment cleaning, environmental hygiene | Environmental monitoring program gaps; inadequate swab sites |
| Allergen Controls | 21 CFR §507.39(a)(2) | Label accuracy, cross-contact prevention | Rework handling not documented or not controlled by lot |
| Traceability Records | 21 CFR §507.175(c) | Lot linkage, customer distribution records | Commingled ingredient lots; lot codes not linked forward and back |
| Corrective Action | 21 CFR §507.42 | Deviation response, hold procedures | No written decision criteria; hold applied inconsistently or late |
What a Well-Functioning Food Safety Plan Actually Looks Like
A food safety plan that prevents lot-level recalls isn't a longer document — it's a more honestly constructed one. The plans that actually perform in practice share a few characteristics.
They characterize hazards based on what has actually happened in comparable facilities with comparable ingredients — not just theoretical categories. If physical contamination recalls have occurred at industry peers using similar protein sources in the last five years, that history belongs in the hazard characterization. FDA's guidance is explicit: hazard analysis should be informed by known and reasonably foreseeable hazards in the food supply, including historical industry data.
They define monitoring with specificity — who performs it, how often, with what equipment, calibrated on what schedule, with what corrective action trigger. "Check periodically" is not a monitoring parameter. Vague monitoring procedures produce inconsistent execution, and inconsistent execution doesn't show up in records until an audit or an investigation forces someone to look.
They treat corrective actions as decision trees, not blank forms. When a retort deviation occurs during a night shift, the operator on the floor shouldn't be improvising. They should have a written procedure that specifies exactly: tag the lot, notify the on-call QA manager, initiate the deviation investigation, apply the quality hold to all affected production from that run. An actual procedure with decision logic — not a generic form with a signature line.
And they're verified. 21 CFR Part 507 requires that preventive controls be verified — meaning someone confirms, on a schedule, that monitoring is being performed as written and that the controls are effective. Verification activities include record review, calibration confirmation, and periodic product or environmental testing. A food safety plan that sits in a binder and never gets verified is a liability that stays quiet until it shows up on the FDA recall database.
Getting Your Food Safety Plan FDA-Ready
If you manufacture pet food, animal food, or dietary supplements — and you're not fully confident your food safety plan would hold up under a 21 CFR Part 507 inspection — that's worth addressing before FDA addresses it for you. The framework isn't complicated, but executing it well requires someone who understands both the regulatory requirements and the practical realities of your specific process and ingredient profile.
At Certify Consulting, we've helped 200+ clients across food, supplement, and regulated manufacturing sectors build quality systems that hold up under regulatory scrutiny — with a 100% first-time audit pass rate. If you want an honest assessment of where your food safety plan has gaps, start with a free consultation with Jared Clark.
Last updated: 2026-07-15
Reference: FDA Voluntary Recall Notice — PEDIGREE® Can High Protein Chopped Chicken & Duck Flavor Wet Dog Food, July 2, 2026
Jared Clark
Principal Consultant, Certify Consulting
Jared Clark is the founder of Certify Consulting, helping organizations achieve and maintain compliance with international standards and regulatory requirements.