Food Safety & SQF Programs

The stake: A documented pest program that survives an SQF audit — Graduate built one in the mid-1990s that turned out to be what the standard later required.

What is SQF, and what does it actually ask of a pest program?

SQF is Safe Quality Food — a food safety certification scheme recognized by the Global Food Safety Initiative and administered by the SQF Institute, a division of FMI. A certified site has been audited against a published code covering its food safety system end to end: the management commitment behind it, the hazard analysis, the prerequisite programs that hold the analysis up, and the records that show the whole arrangement actually running rather than merely existing.

Pest management sits inside that structure as a prerequisite program. The code expects a documented pest prevention program; it expects that program to have been designed for this building rather than transcribed from another one; it expects execution on a stated frequency; it expects the resulting data to be read by somebody; and it expects corrective action when the data says something is wrong. It also expects the applicator to be licensed and appropriately trained, and the facility to have checked that rather than taken it on trust.

The clause numbering, the scoring bands and the grades attached to them are the scheme owner’s material, and they move between editions of the code. Anyone who quotes a clause number at a QA manager should be asked which edition they are quoting and when they last read it. What does not move between editions is the shape of the expectation, and the shape is what a program has to be built around: written, mapped, executed, trended, corrected, verified. Get those six right and the edition in force stops being a live risk.

The general food-plant framework — the preventive-controls background, the audit layer as a whole, the device and reporting mechanics — is set out on our food manufacturing and processing page. This page is narrower and more specific. It is about what happens when an auditing scheme meets a pest program that was already there.

What happened when SQF arrived at a program built in the mid-1990s?

Nothing had to change.

The sequence is worth laying out in order, because the order is the whole point. Graduate has worked alongside the same food manufacturing company since 1989. In the mid-1990s we built that client a food safety pest program: the survey, the device layout reasoned from the building rather than spaced at regular intervals, the written scope, the service frequency, the records, the structural findings raised as findings and tracked until somebody closed them.

That was years before SQF existed as a scheme anyone in this market was audited against. There was no certification to design toward. The program was built the way it was built because that was how the work was done here, and because a plant that runs its pest program any other way spends the following decade repeating itself.

When SQF eventually arrived at that account and began auditing, the program it found was what the scheme required. Not close to it, not adaptable to it — what it required. Nothing on our end had to be rewritten, re-scoped or retrofitted. The standard had codified a practice, and the practice was already the house method.

The only thing that has changed since is the medium. When the technology arrived, we stopped writing the notes by hand. The device map moved off paper. The service records became searchable. The trend data stopped needing to be assembled by somebody with a ruler and started being a query. That is a real improvement in speed and legibility, and it is also, honestly, all it was: the same program, entered differently.

The account today is six facilities under weekly service. It has been continuous since 1989, which is the part that is difficult to acquire and impossible to claim retrospectively.

Why should a QA manager care that the program predates the standard?

Because it separates a method from a marketing decision, and in this trade almost nothing else does.

Every pest control website in this market now carries the phrase integrated pest management. It is free to write. It commits the firm to nothing, because there is no external body checking whether the program behind the phrase is an inspection-and-correction discipline or a bait station route with a nicer word on the invoice. A QA manager evaluating contractors is looking at a page of firms who all describe themselves identically and who cannot all be telling the truth.

What cannot be written for free is a program that predates the standard and passed it unchanged. That is a sequence with dates in it. It says the practice was not assembled to satisfy an auditor, because when it was assembled there was no auditor to satisfy. The scheme showed up later and found the work already conforming.

It also says something about durability that a current certification does not. Any firm can be compliant on the day it is inspected. A program that has run continuously on one account since 1989, through changes of QA staff, changes of plant management, changes of standard, changes of edition and the arrival of an audit regime that did not exist when it began, has been tested by something harder than an audit day.

The industry named this practice integrated pest management about fifteen years ago. Graduate did not adopt it. Arnold Katz was working this way in 1983, when the company was founded — find the source, read the structure, correct the condition rather than the symptom — and the industry eventually caught up and gave it a name. That is set out in more detail on Ryan Katz’s page, including what the family holds by way of credentials and what it does not.

None of which is an argument that a QA manager should take on faith. It is an argument for what to ask a contractor: not whether they do IPM, but what their program looked like before the standard required it, and what they had to change when it did.

What does an SQF auditor read before they walk the floor?

The file, and the file is read as a description of a building that the walk will then test.

Expect the request to cover the written pest prevention program and its scope, the current site map with every device numbered and located, the service history over a meaningful period, the trend data, the corrective-action log, material and label records for anything applied, and the contractor’s licensing and training documentation. Reading that package builds a model of how the plant is supposed to work.

Then comes the walk, and the walk exists to test congruence. The finding that damages a plant is not usually a live infestation — it is a mismatch between the paperwork and the floor. A numbered station that is not where the map puts it. A device sitting behind a pallet where nobody could have serviced it. A service record dated to an afternoon the area was shut down. An item logged as closed with the condition still visibly open twenty feet away.

A single mismatch is a finding and can be answered. The real cost is what a second one does to the auditor’s posture. Once they have two reasons to doubt the file, they stop sampling and start verifying, and a day that would have covered a fraction of the device network now covers all of it. The paperwork risk in a food plant is not that the file is thin. It is that the file is confidently wrong.

Two patterns draw attention regardless of the rest. Uniform zeros across every device over many months, in a building with dock doors, incoming pallets and a yard, reads as a network nobody is genuinely checking. And sustained activity with no corrective action attached reads worse still, because it demonstrates a plant that is measuring a problem and choosing not to act on it. A documented problem being visibly driven down is a stronger file than a spotless one.

What does an auditor do with the device map?

Uses it as the index to everything else, which is why its accuracy decides how the rest of the day goes.

The map is the auditor’s sampling frame. They pick numbers off it, walk to those numbers, and see whether the device is present, serviceable, reachable and recorded. A drawing that matches the floor keeps the exercise a sample. A drawing that does not turns it into a census, and the census will find things the sample would have missed.

So the questions worth asking about your own map are narrow. When was it last revised, and what changed in the plant since? Do the identifiers survive a rack being moved, or were devices numbered in the order somebody installed them? Is there a device on the drawing that nobody has been able to reach for a year because a pallet position was created in front of it? Is any device recorded as checked on a date the area was shut down?

Drift is the mechanism behind almost all of it. Plants change constantly — racking moves, a wall goes in, a bay is bricked up, a night crew relocates three stations to get a sweeper through — and the documentation follows months later, if at all. The discipline that prevents it is unglamorous: the map is revised the week the plant changes, with a revision date on the face of it.

How the network should be laid out in the first place — the layers, the device classes, what each one’s data actually tells you, and how a pheromone grid locates a source rather than counting one — is set out on our food manufacturing and processing page.

How will an auditor read your trend data?

Backwards from the interventions, which is not how most reporting packages are built.

An auditor is not impressed by a low number and not alarmed by a high one. What they are looking for is whether the site noticed something, did something, and can show what happened next. So the reading runs: here is a rise, here is the corrective action attached to it, here is the count afterwards. A package that presents monthly site totals with no locations and no interventions marked cannot support that reading at all, however neat it looks.

Two shapes attract attention wherever they appear. Sustained activity with nothing attached to it says the site is measuring a problem and choosing not to act. Uniform zeros across an entire network for many months, in a building with dock doors and inbound pallets, says nobody is genuinely checking. A documented problem visibly being driven down is a stronger file than a spotless one, and every experienced auditor will say so out loud.

The test to apply to your own reporting is whether a food safety director can answer three questions from it inside a minute: where is the pressure, which way is it moving, and what did we change. The mechanics behind producing that — plotting by device and area, separating season from structure, setting action thresholds in advance — are covered on the food manufacturing and processing page.

What separates a closed corrective action from a note in a file?

Ownership, a date, evidence, and a second question that most logs never ask.

A service note records what was found. A corrective action names the root cause, assigns an owner, sets a real target date, records what was done, and verifies that it worked. The distinction is visible to an auditor in about ninety seconds of reading, and a log made entirely of notes is read as a plant that observes its problems.

Take a run of captures inside the interior door at one bay. Adding devices and increasing inspection frequency there is containment, not correction — it improves the measurement of a problem that is still happening. Correction begins by asking why anything is getting past that door, and the answer is usually several things at once: a compressed bottom seal, an open back corner in the leveller pit, and a habit of leaving the bay open through a long unload. Correction is replacing the seal, closing the pit, and changing the unload practice, each with a named person and a date. Verification is what that device does over the following weeks, recorded and compared against what it was doing before.

Preventive action is the question that follows, and it is where most logs stop. If that leveller pit is open, what about the others? Are all the bay seals the same age and therefore all near the same failure point? Does the unload practice apply plant-wide or only on that shift? A log that only ever fixes the specific device that flagged will work through the building one failure at a time, indefinitely, and an auditor who reads a year of it will say so.

Three practical points. Most corrective items do not belong to the pest contractor — they belong to maintenance, engineering, sanitation or purchasing, and an item without an owner inside the plant does not close. A target date that is quietly re-dated three times is worse in the file than an item honestly recorded as awaiting a capital cycle with the justification written next to it. And verification needs evidence: a photograph, a follow-up count, an inspection note. The word “completed” with nothing behind it is an assertion, and auditors have read a great many of them.

Where do sanitation and structure meet, and who owns each item?

At a small number of interfaces that generate most of the persistent problems in most plants, and where both departments tend to consider the item the other one’s responsibility.

Dock doors and levellers. The highest-yield failure point in almost every facility. Bottom seals compress and crack, side seals tear, and the leveller pit itself is a sheltered, warm, debris-collecting void with an unobstructed route into the building. Pits are also where sanitation genuinely cannot reach, which means the structural fix and the cleaning problem are the same item.

Air curtains. Where one is fitted it needs to be correctly specified for the opening, aimed, balanced, switched on, and verified periodically with the verification recorded. An unbalanced curtain is a decorative appliance that the file describes as a control. Positive pressure in production areas does the same work passively and is worth checking whenever doors keep getting cited.

Personnel doors. Sweeps that no longer touch the threshold, closers that no longer close, and the door propped open in July. Self-closing hardware that has stopped self-closing is one of the most repeated findings in food plants and one of the cheapest to correct.

Roof penetrations. Chronically under-inspected and frequently where the real answer is. Conduit and refrigerant lines cut through and never properly sealed, HVAC curbs open at the corners, failed parapet flashing, standing water, and bird activity on rooftop equipment that becomes droppings near a fresh-air intake. Nobody finds these from the floor.

Drains and wet areas. Floor and trench drains, their covers, condensate lines, and any low spot holding water at the end of a shift. Drain biofilm will support a small fly population entirely independently of anything else happening in the building, and no amount of exterior work touches it.

Equipment voids and hollow framing. Open-ended frame members, motor housings, conveyor undersides, and guarding that cannot be removed for cleaning. Simultaneously a sanitation failure and a harborage, which is precisely why it survives so long.

The ownership question is worth settling in writing at the start of a program rather than during an audit. The contractor owns the survey, the devices, the data, the reporting and the naming of the condition. The plant owns the door, the roof, the drain, the storage practice and the cleaning schedule. Where that split is written down and reviewed on a cycle, items close. Where it is not, the service report records the same finding every week for two years and everybody involved slowly stops reading it.

What can actually be applied in a food-contact environment?

Considerably less than most people outside a plant assume, and the constraint tightens exactly where the risk is highest.

Three layers narrow the options at once. The label is the first and it is law: it governs what may be used in a food-handling area, in a processing area, and near a food-contact surface, and those distinctions are specific rather than advisory. Scheme and customer policy is the second, and it typically narrows things further — most visibly by pushing interior rodent control to mechanical, non-toxic devices as a matter of course. The plant’s own risk assessment is the third, and it should narrow them again, because a contamination event is a materially worse outcome than the pest problem that prompted the application.

What remains has a legitimate and useful role. Targeted crack-and-crevice work where labels permit it. Exterior programs. Specialist interventions such as structural fumigation for an entrenched stored-product insect population, which is a major planned undertaking, disruptive, arranged months in advance, and — this is the part worth saying plainly — a symptom of years of accumulated structural and sanitation failure rather than a routine tool in a functioning plant.

Everything applied is a record. What, where, when, at what rate, under which label, with the label and safety data sheet retrievable. In a food environment that file is examined closely and it is a legal record before it is an audit artefact.

The conclusion a QA manager should draw from all of this is not that chemistry is unavailable. It is that chemistry is least available in the rooms that matter most, which means the plant’s real pest control capacity is structural whether anyone has framed it that way or not.

Why does exclusion outweigh chemistry inside a plant?

Because a sealed opening works continuously, leaves no residue, and does not need a label check — and because the building is most of what a food plant is actually permitted to change.

The exclusion work in a manufacturing facility is not the work you would do on a house. It is dock seals and leveller pits, door hardware and sweeps, air curtain specification and verification, roof penetrations and curb flashing, wall-to-floor junction repair, sealing conduit and pipe sleeves through walls and slabs, screening on vents and intakes, drain cover and trap integrity, and closing the equipment voids that sanitation has no way to clean. A good deal of it is capital work executed by other trades, which is exactly why it has to be specified precisely enough to be scoped, priced and then verified once done — a finding written as “seal the dock area” will not be actioned by anybody.

Materials matter as much here as anywhere. Custom-fabricated 26-gauge galvanised sheet metal cut to the opening. Copper and stainless mesh where an annulus has to be packed. Hardware cloth on any opening that has to go on ventilating. Mortar and hydraulic cement at masonry. Xcluder door sweeps, fill fabric and below-grade barrier where a barrier has to sit against burrowing. Mechanical fasteners throughout, because an adhesive bond is a guess about a surface nobody prepared. We do not use expanding foam in any role — not as a filler, not as backing behind something better, not as an air seal over it. Rodents chew through cured foam, and in an audited facility a foamed penetration is a finding waiting to be written up twice: once as an unsealed opening and once as a repair that misrepresented itself as one.

The underlying reasoning behind all of this is on the structural exclusion page, and the exterior and below-grade rodent picture that drives most perimeter findings is on rodent control. Where a facility is building, expanding or gutting a line, the cheapest version of this conversation happens before the walls close, which is what exclusion consulting is for. Construction is the single strongest predictor of a new pest problem in a stable plant, because every trade cuts penetrations and almost none of them close what they cut.

What do PCQI, HACCP and SQF credentialing change on this account?

They change who Graduate can talk to and in what language, which sounds like a small thing and is not.

Ryan Katz holds all three personally. PCQI is the Preventive Controls Qualified Individual credential, the individual qualification that sits behind a facility’s preventive controls food safety plan. HACCP is hazard analysis and critical control points, the framework the entire food safety discipline is built on. SQF credentialed means trained against the scheme this page is about.

The practical difference shows up in the meetings that are not about devices. A contractor without these turns up, services the network, leaves a report, and has no view on where pest management sits inside the plant’s food safety plan or what the auditor will do with eighteen months of trend data when they ask for it. A contractor who holds them can discuss why a device sits where it sits in the terms the plan already uses, can read a hazard analysis without needing it explained, and can tell a QA manager which of this month’s findings will matter at audit and which will not.

It also changes what happens when the audit is close. A pre-audit walk is only useful if the person doing it knows how the file will be read as well as how the floor will be walked, and can tell the difference between an untidy record and a genuine gap.

Two honesty notes, because credential inflation is endemic in this trade. These are Ryan’s individual credentials, not company badges — certifications belong to people. And Ryan does not hold a degree; his father Arnold Katz holds the B.S. in Entomology from the University of Georgia and still works the accounts as senior technician and staff entomologist, providing identification and diagnostic support where a species determination changes the plan.

What does weekly service across six facilities actually involve?

Consistency between buildings, which turns out to be harder than depth in any one of them.

Multi-site work fails in a specific way. Each plant develops its own conventions — different numbering logic, different report formats, different thresholds for what counts as worth writing down, different relationships between the technician and the maintenance lead — and within a couple of years the corporate quality function cannot compare its own sites. One plant looks worse than another because it records more honestly. Nobody can tell whether a group-wide problem exists because there is no group-wide view.

What prevents that is unremarkable and has to be enforced: one numbering convention, one report format, one definition of a finding, one corrective log structure, and trend reporting that can be read per site and across sites without anyone rebuilding it in a spreadsheet. When a condition is found at one facility, the preventive question is asked at the other five in the same week rather than at each of their next audits.

Weekly frequency is not a badge. It is what a processing environment with live receiving, wash-down areas and ingredient storage generally needs, and it should be defensible from a written risk assessment covering what is processed, whether it is ready-to-eat, the age and condition of the building, seasonal pressure and what the trend is currently doing. It should rise during construction and after any significant finding. If a contractor cannot explain in a sentence why the frequency is what it is, the frequency was copied.

What should a facility ask a contractor before an audit?

Six questions, and the answers are more revealing than any credential list.

Show me the map, dated. If the most recent revision predates the last plant change, the audit has already found something.

Show me a corrective item you raised that cost us money and that you had no commercial interest in. Roofing, doors, drainage. A contractor who has never written one of those is either working a perfect building or is not writing findings that create friction.

Show me trend reporting you produced without being asked for it. Anyone can generate a chart on request. The question is whether it is part of the service.

Which of our recurring findings do you think we will never close, and why? The useful answer names a capital item and a reason. The unhelpful answer is a reassurance.

Is any of our pressure coming from outside our boundary? A neighboring property, a shared yard, a supplier’s pallets. This belongs in the report with the evidence behind it, because unexplained recurring exterior activity reads as program failure while explained recurring activity reads as a managed risk.

Will you walk the file with us before the audit? A pre-audit walk that covers both the paperwork and the floor, conducted the way an auditor would conduct it, is the single most useful thing a contractor does in a certification year.

What no contractor can offer is a result. The pest element is one prerequisite program inside an audit of an entire food safety system, and any firm implying otherwise is selling something that is not theirs to sell.

Where should a food safety director start?

With whatever your last audit report said about the pest element, and with the findings you have carried forward more than once.

Graduate works on buildings across New York City and Long Island, and has done since 1983, on the principle that a pest problem is a building problem. In a food plant that stops being a philosophy and becomes an operating constraint, because the building is most of what you are allowed to change.

If your operation includes retail or foodservice outlets, the inspection regime there is different and is described under restaurant pest control. If you run several buildings of different ages, the consistency problem is covered under property management. The wider commercial framework sits on the commercial pest management page, and the coverage area is listed under locations.

To begin, get in touch with the facility type, the scheme and edition you are certified against, when the next audit falls, what your trend data is currently showing, and any pest finding still open from the last one. The consultation is free and is usually a phone call; a written proposal and plan carries a service fee, and if you proceed with the work that fee comes off the cost of the project.