On September 11, Prosperity Organic Foods of Boise, Idaho, began withdrawing a single lot of Miyoko's Creamery European Style Salted Plant Milk Butter. On October 23 the FDA classified the action as Class II, the tier for products that may cause temporary or medically reversible health consequences where serious harm is considered remote. The recall covers 1,304 cases of six 8-ounce packages each, or 7,824 retail units, shipped to ten states. Local reporting based on FDA enforcement information gives the identifiers: UPC 857554005773, lot 07871 on the retail carton, lot 1007871 on the shipping case, best-by date November 5, 2026.

By the standards of food recalls, that is a tidy perimeter. A lot code, a date and a barcode define it. But a boundary this narrow protects anyone, including the company's balance sheet, only if every party holding butter can answer one question quickly: is what I hold from lot 07871 or not? When the answer takes days, or arrives with doubt, the rational move for a retailer is to pull more than the notice requires. That excess is the subject of this article. Nothing in the public record says it happened here. I found no disclosure of Miyoko's or Prosperity costs, and the recall is still open. The case is useful because it shows how small the legally relevant universe can be, and how much of the cost question sits outside it.

Three capabilities that get confused

Compliance, operational traceability and financial precision are routinely discussed as one thing. They are three.

Compliance is what the law requires. For most foods that still means FDA's long-standing recordkeeping rules, under which firms keep records of the immediate previous source and the immediate subsequent recipient. One step back and one step forward lets an investigator walk a chain link by link. It does not tell any single company where a given lot went beyond its own customers, or what else shared its production run.

Operational traceability is internal and connective. It ties a finished lot to the ingredients, packaging and line it came from, and to the pallets and customers it went to. A firm can build this for its own quality purposes and still be unable to share it in a form a partner can act on.

A financially precise recall is a third thing: the ability to show, quickly and to someone else's satisfaction, which units fall outside the boundary. The test is not whether the manufacturer can find the lot. It is whether a retailer's risk team will keep the neighboring lots on the shelf on the manufacturer's word. That is a relationship asset built on data formats, timing and credibility, and it is where the money turns.

The clearest public illustration comes from a securities filing. In its fiscal 2023 annual report, J.M. Smucker described its 2022 Jif peanut butter recall, which covered select products from its Lexington, Kentucky plant. The company also paused shipments from its Memphis plant, which was outside the recall, “to eliminate confusion while customers cleared their shelves.” It reported roughly $120 million in total direct recall costs, net of insurance recoveries, spanning customer returns, fees, unsaleable inventory and other items. The filing does not say how much of that figure came from the pause or from product that was never implicated. It should not be read as a measure of overbreadth. What it does show is that a recall defined by plant and lot ran alongside commercial decisions driven by uncertainty.

What the $72.7 million figure is, and is not

The most direct attempt to price that uncertainty is a 2025 paper in the Journal of Food Protection, “Costs of Overly Broad Recalls,” by Nyssa Ackerley, Ayesha Berlind, Michael Black, Kevin Kho, Cristina McLaughlin, Aliya Sassi, Aylin Sertkaya and Sheri Walker. It uses a structured expert elicitation to estimate the spillover costs firms incur when a recall following an FDA product advisory sweeps in uncontaminated product. The scenario is a source that is misidentified or cannot be found promptly, so an entire category, romaine lettuce being the authors' example, is treated as suspect.

The reported median costs per event vary widely by firm type and size. Producers and processors face $3.0 million to $72.7 million. Shippers and distributors face $0.1 million to $2.3 million, and nonrestaurant retailers $0.1 million to $3.1 million. Restaurants, outside the scope of this piece, range from $0.04 million to $1.1 million. The authors also find that producers bear the largest costs and that labor is a small share of the total. They are explicit about why they used experts at all: “Industry is often reluctant to provide estimates on the costs of recalls.”

Those numbers invite misuse, so the limits matter:

First, they are elicited judgments, not observed losses. They describe what informed people believe a typical overly broad recall costs, not what any company booked.

Second, they describe a particular scenario, an advisory-driven category recall. A one-lot withdrawal like the Miyoko's action is a different animal. The study speaks to the cost of imprecision when it occurs, not to how often it occurs or whether it occurred here.

Third, each range spans firms of different sizes, so its ends are not a confidence interval. Medians also do not add. Nobody can multiply a per-firm figure by a headcount and call the product the cost of an event, and the study does not claim otherwise.

Fourth, a retailer's $0.1 million to $3.1 million looks small beside a producer's $72.7 million, but there are far more retailers than producers. Per-firm figures say nothing about who carries the larger aggregate burden.

The regulatory ledger

The other large quantitative anchor is the FDA's own regulatory impact analysis for its 2022 Food Traceability Rule. Over 20 years, annualized at a seven percent discount rate, the agency put the rule's cost at about $63 million to $2.3 billion, with a primary estimate of $570 million a year. It estimated non-health benefits from avoiding overly broad recalls and withdrawals at $233 million to $1.8 billion, with a primary estimate of $575 million. Monetized health benefits had a primary estimate of $780 million. All figures are in 2020 dollars.

Read carefully, this is less triumphant than it sounds. At the primary estimates, avoided overbreadth alone roughly offsets the rule's cost. The cost and benefit ranges overlap heavily, so on that component alone the sign is not guaranteed. The case becomes comfortable only when health benefits and the agency's unquantified gains, such as inventory control, faster recall completion and less food waste, are added. These are modeled projections for a defined list of foods, resting on assumptions about recall frequency and scope. They are not a company's return on investment.

The timing also matters. The compliance date, originally January 20, 2026, was pushed to July 20, 2028, and Congress separately directed the agency not to enforce the rule before that date. FDA itself acknowledged, in proposing the delay, that readiness by some firms would not deliver the rule's benefits if others could not comply. That concession is the economic heart of the issue: a recall boundary is only as sharp as the weakest record in the chain.

Who pays, who collects

USDA's Economic Research Service laid out the logic two decades ago in Agricultural Economic Report 830, by Elise Golan, Barry Krissoff, Fred Kuchler, Linda Calvin, Kenneth Nelson and Gregory Price. Firms, it found, build traceability up to the point where their private costs and benefits balance, and the report's own section heading concedes that “Complete traceability is impossible.” The economic implication is easy to miss. The level a firm chooses reflects the benefits that firm captures, not the benefits the supply chain collects.

Consider who gets what when a recall boundary is drawn tightly and believed. The manufacturer avoids destroying unaffected inventory and shortens any production stoppage. The distributor keeps trucks moving and avoids holding or reverse-shipping product that was never implicated. The retailer keeps adjacent lots on the shelf and avoids the sales gap and the administrative scramble. An insurer faces a smaller claim. Smucker's fiscal 2022 filing recorded a $52.3 million accrual against a $49.8 million insurance receivable, a reminder that insurers can absorb most of the direct costs of a recall, though policy terms are private and the filing does not show what a broader boundary would have done to the claim. Consumers avoid needless discarding and empty shelves, benefits the FDA lists among those it did not monetize.

Only some of those benefits come back to the party paying for the system. Contracts can push costs upstream, and Smucker's filing refers to customer returns and fees. Where chargebacks and penalties are enforceable, manufacturers internalize part of the downstream loss and have reason to invest. The unpriced remainder is where underinvestment lives: the retailer's lost sales of lots it chose to pull out of caution and was never compensated for, the carrier's idle capacity, the shopper's discarded butter. Each of those costs is real, yet none appears on the balance sheet of the firm that would have to buy the software, relabel the cases and train the dock.

This also explains why compliance and precision diverge. A manufacturer can satisfy an auditor with records that are accurate but slow to retrieve, or that sit in formats its customers cannot ingest. The cost of that gap is then paid by others at the moment of greatest urgency.

What the public record cannot tell us

The evidence supports some claims and forbids others. FDA's Enforcement Report, the public face of its Recall Enterprise System, shows what was recalled, in what quantity, with what codes and where it went. It does not show what the recall cost, how long it took to define, or whether retailers pulled adjacent product. Securities filings can show direct costs, insurance recoveries and sometimes shipment pauses, but they rarely separate the spillover from the core. Expert elicitation gives a plausible magnitude for one scenario. The regulatory analysis gives a system-level projection. None of them tells you the build cost for a particular company, the probability that any given firm faces an overly broad recall, or whether better records would have changed the outcome of a specific event.

That includes this one. Whether the Miyoko's lot recall incurred spillover is unknown from public sources, and the rule's scope is a separate legal question. It covers only foods on FDA's Food Traceability List, and the recall notice does not say whether a plant-based butter falls inside it. Anyone offering a company-level savings figure from this case would be inventing it.

What remains defensible is narrower and more interesting. Traceability is best understood as an option whose value is realized in rare, expensive months, and only if the counterparties accept the data it produces. The research suggests the largest per-firm stakes sit with producers, but real money sits downstream and with insurers, and the people who pay for the system hold only part of it. A narrow lot is good news only for those who can prove it is narrow. The unresolved question is less about software than about contracts and standards: who is entitled to demand that proof, who must supply it, and who bears the cost when it is missing.