The Spec Debt Behind Every Custom SKU

The number that should have stopped the tour

The plant tour at a specialty foods manufacturer ran the way most do. Hand-packing on the meal-kit line, workers filling containers with sauce and meat, sealing, boxing for the oven. Slow package lines. A manual quality check where someone eyes each product for uniform size, the kind of step that should have been automated three years ago. None of that was the real story.

The real story came when the supply chain manager started talking about ingredients. The operation runs more than 260 ingredient items, and the count keeps growing. Eighty to 130 of them go through fresh RFQs every year. Many of those exist because a customer wanted a customization, which forced a new and unique ingredient, some sourced from as far as France. Every one of those customizations looked like a win at the quote. The customer got what they asked for, the business got booked.

What nobody had was a system holding the specs for any of it. The catalog had grown to 260-plus moving parts, and the only thing tracking them was the buyer's memory and a scatter of documents. That is spec debt, and it had been accruing silently for years.

Variety is a loan you take against your future self

Here is the mechanism. When you accept a custom SKU, you are not just sourcing an ingredient. You are signing up for a permanent record-keeping obligation: this item's spec, its supplier, its allergen profile, its packaging, its substitution rules, its qualification history. One SKU, you can hold in your head. Two hundred sixty, you cannot. The obligation is real from day one, but it never shows up as a line item, because no invoice says "spec management." So the cost stays invisible, and the catalog keeps growing, because the thing that would have made you say no, the carrying cost of variety, is the one number nobody is tracking.

Debt that is invisible still comes due. For this manufacturer the trigger is Extended Producer Responsibility. Packaging legislation is active or pending in California, New Jersey, Oregon, Washington, Minnesota, and Colorado, and compliance is, at its core, a data demand: produce clean packaging-spec data for every SKU, calculate your exposure, report it, pay the fee. A company with no spec database cannot even estimate what it owes. It has to reconstruct hundreds of specs at once, on the regulator's clock, and likely add fractional-to-full-time headcount just to file. The debt that accrued one quiet customization at a time gets called in a single lump.

The shape repeats across the floor. It is the same as an undocumented allergen profile that turns a routine changeover into a guess, or the technical debt in a codebase that no one priced until the migration. The cost was always there. Only the bill was deferred.

Make the spec a gate, not an afterthought

The fix is not a software purchase. It is a policy, and you can set it this week.

First, make spec intake a gate. A new customer-specific SKU does not go live until it has a structured spec record with a named owner. No record, no production. This feels like friction at the moment a salesperson wants to book the business, which is exactly when the friction is worth the most, because that is the only moment you can still price the carrying cost before you take the position.

Second, push spec ownership to the function that sets the spec, not the one that buys against it. A protein processor working through its film protocol recently made this call explicitly: the protocol document, the team agreed, needs to be the client's own document, with quality-spec ownership sitting with the quality team rather than procurement, and the plant's machine-level production data treated as the single source of truth. That is the move in miniature. One authoritative record, owned where the knowledge actually lives, anchored to real floor data instead of a supplier's claim. They tied the spec to baseline and test data, to actual leak testing, not to a number on a sell sheet.

Third, get ahead of the packaging specs specifically, because that is where the regulatory bill is already addressed to you. Every new label is a spec to track, down to the color count; every carton is a dimension and a material. The processors who structure that data now are writing the report they will otherwise scramble to build in 18 months. The work is the same either way. The only variable is whether you do it calmly or under penalty.

What a system without spec debt reads like

Every active SKU has a structured spec record with one named owner. New SKUs cannot go live without it, so the catalog cannot outrun its own documentation. Any packaging or ingredient spec pulls in under a minute. Regulatory reporting is a query against a database, not a six-week archaeology project. And SKU additions get priced with their spec-carrying cost visible at the quote, so the variety that does not pay for itself gets declined before it is booked, not discovered two years later when a state asks for its data.

Price the position before you take it

The same supply chain manager who had no spec system had real discipline elsewhere. He had refused to use over-the-counter instruments to hedge beef, because in 2024 those hedges would have run 30 to 35 percent above the actual market price. He would not pay for paper protection against a risk he could read directly. That instinct is exactly right; it just needs to point at variety too. Every customization is a position. The specs all looked free at the quote, and the debt was real the whole time. Nobody had put variety and its carrying cost in the same model.