6 min read · Operations
COGS rarely blows up all at once. It creeps. A scoop that's slightly generous, a topping applied a little heavier than the spec sheet says, a recipe substitution made once during a supplier shortage that quietly became the new normal — none of it looks like a problem on any single ticket. It only becomes visible weeks later, when the monthly P&L lands and someone asks why margin is down two points and nobody can say exactly when it started.
That gap — between when a cost problem actually begins and when someone notices it — is the most expensive gap in food and beverage operations. Here's why it exists, and what closes it.
The problem is resolution, not effort
Most operators aren't missing COGS data. They're missing it at the right resolution. A monthly P&L tells you COGS went from 29% to 31%. It doesn't tell you which of your 40 SKUs moved, which shift it happened on, or whether it was one location or all of them. By the time a number is wrong enough to show up in a monthly report, it's usually been wrong for three to six weeks already — and every one of those weeks compounds the loss.
The fix isn't more reporting. It's watching at the level where drift actually happens: per item, per shift, per location, continuously — not once a month in arrears.
What portion drift actually looks like
In practice, COGS drift concentrates in a small number of high-volume items. A shift lead ladles sauce a little heavier during a rush because it's faster than being precise. A new team member wasn't trained to the exact portion spec and nobody corrected it in week one. A supplier substitution during a shortage changed a yield ratio slightly, and it was never revisited once the original supplier came back.
None of these are dramatic. Each one might only move COGS by a fraction of a point on its own. But three items drifting at once, across 140 locations, adds up to real margin — and because each individual instance looks like normal variance, it's genuinely hard to catch by eye.
Catching it before the ledger does
This is exactly the kind of problem a deterministic rule is good at, and it's why COGS threshold monitoring sits at Iris's L1 layer (Zen Rules) rather than waiting for a monthly forecast. A rule watches actual cost per item against its expected spec, shift by shift, and flags the moment a location or item crosses its threshold — not the month after.
Once a rule fires, the next question is always "why" — and that's where it gets useful instead of just noisy. Rather than a flat alert, the system traces the variance back to the handful of SKUs actually driving it, so a manager isn't left auditing forty items to find three.
The fix is usually smaller than people expect
The instinct when COGS drifts is to reach for a price increase across the board. Most of the time that's the wrong tool. If the actual cause is portion drift on three items, the fix is a portion re-spec on those three items — retraining, a visual guide at the station, a scale check — paired with a small, targeted price adjustment only where it's warranted. That combination typically closes the gap faster and with far less customer-facing disruption than a blanket price change, because it treats the actual cause instead of the symptom.
The takeaway
COGS drift isn't usually a pricing problem or a supplier problem. It's a visibility problem — the gap between when something starts and when someone's looking closely enough to see it. Close that gap, and the fix itself is almost always smaller and cheaper than operators expect.