Operating cost control: a checklist with a 'done' criterion vs an open list

Why a checklist without a 'done' criterion controls nothing?

No item in a cost checklist ever closes without a verifiable condition behind it, which is why so many control lists end up as a pile of intentions nobody checks off.

Fifty-two percent of restaurateurs name inflation as their top 2026 challenge, and most face it with loose verbs (review, watch, check) that no manager can mark complete because they carry no threshold. Confusing intention with control is the real mistake: MEASURING waste closes nothing; keeping it below 4%, confirmed by weekly inventory, does. The difference reads like wording, but at the register it separates real control from apparent control: what can be verified gets audited, and what gets audited improves. At Masterestaurant no point counts as closed without its final figure, and Diego F. Parra puts it bluntly: what has no 'done' criterion is not done. An item becomes verifiable once it carries three things: a numeric threshold, a unit of measure, and a data source confirming it without anyone's opinion involved.

What makes an item verifiable and auditable?

CONTROLLING food cost is intention; food cost per dish at or below 32%, with a standard recipe and updated input cost, confirmed by the POS report, is a done criterion.

That condition turns a vague task into something any team member can audit. It is the logic behind the checklist's 12 points: each one carries its figure, unit, and source. The first real run usually shows only 3 of 12 were actually closed; the rest was control theater dressed as management. That early discomfort is, paradoxically, the best sign the exercise works: without it, the owner keeps believing costs are controlled while inflation eats the margin. Thirty-two percent is the ceiling, not a target one negotiates by the week: below that line, with a standard recipe, the food cost item closes. Three conditions hold it up: a technical sheet per dish, an updated input cost, and a calculation verified against the selling price.

The food cost item: a 32% ceiling with a standard recipe

A food cost checked BY EYE does not count, because the real number shifts every time an input rises. This is where AI earns its place: real-time per-dish margin alerts fire the moment an increase pushes a dish above 32%, well before the accounting close catches it. This item sits at the heart of the plate block, and closing it properly avoids the trade's costliest habit: raising a dish's price over an increase that is really solved by adjusting the recipe or the mix, not by making the whole menu pricier. Waste is the operation's silent leak: uncontrolled, it runs around 8% of input cost and quietly drains 2 to 3 points of operating margin nobody notices day to day. Closing this item takes one hard condition only: waste at or below 4%, confirmed by weekly inventory, never eyeballed. Almost no owner measures real waste, they swear they 'throw nothing away' while the inventory tells a different story, and that gap is exactly what the done criterion exposes.

The waste item: at or below 4% confirmed by weekly inventory

In an operation audited by Masterestaurant, real waste sat at 9% and fell to 4% in two months through standardized portion control and purchase forecasting; that close freed margin equal to a 5% price increase, without losing a single customer. A 'we watch our waste' with no weekly inventory behind it is, at bottom, an open item wearing a closed one's badge. Somewhere between 5% and 9% of the purchasing budget leaks away from habit, buying from the same supplier without ever comparing, and that dormant margin is exactly what the purchasing item forces awake. Its closing condition demands 3 quotes per category every quarter and a documented consolidation; 'I talked to the suppliers' is not evidence on file. A group audited by Masterestaurant turned that discipline into $1,100 in monthly savings without lowering quality, simply by consolidating three suppliers into one and matching purchase frequency to real turnover.

The purchasing item: 3 quotes per category every quarter

AI helps flag which categories drifted from market price, something almost impossible to catch by hand. Together with waste, this item usually returns more margin than any price increase, without costing a single guest. Closing it every quarter, not once a year, keeps the savings alive. Payroll, rent, and utilities are not charged to the plate: they go to the break-even point, a separation Masterestaurant never negotiates on any checklist. Only once the monthly figure is calculated, signed, and compared against real sales does this item close, not when someone claims to 'have it under control' from memory. Mixing this block with food cost is the checklist's costliest error: it pushes an owner to raise a dish's price to cover a rent increase that is really solved with more volume or lower fixed cost. What would happen if an owner ignored that separation for a full year?

The break-even item: separate the plate from fixed costs

Prices would climb every time a structural cost rose, traffic would bleed out along the way, and margin would end up lower than before any 'control' began. Once break-even is known exactly, every lever stays in its lane. Without a named owner, a review date, and a fixed frequency, any done criterion reverts to a wish wearing a system's clothes. The 12 points run weekly for the operational ones (waste, food cost, purchasing) and monthly for the structural ones (break-even, contracts), because a checklist reviewed once a year controls nothing. That weekly cadence is, in practice, the only thing that gets an owner to 12 of 12 closed before considering a price increase. In operations audited by Masterestaurant, closing the full checklist recovers 3 to 5 points of operating margin in a quarter, without hurting traffic. AI sustains that cadence better than memory ever could: forecasting and alerts turn data the point of sale already generates into timely signals.

Owner, date, and cadence: what closes all 12 points

Strip out the cadence and even the best checklist ends up as a dead document in a drawer. Price is the item almost every restaurateur reaches for first, and in this checklist it goes dead last: it activates only if the other 12 points are closed and margin is still short. Raising prices without closing structure, food cost, waste, and purchasing costs up to 9% of traffic, without recovering real margin. This last item's done criterion is different: not a fixed threshold but a sequence condition (the other eleven closed) plus a method that runs AI scenarios on low-elasticity dishes against the sales mix and competitor pricing. Done this way, the adjustment caps the traffic drop at 4%. Diego F. Parra repeats it in every Masterestaurant engagement: price is the last checklist item, never the first. Today's only action is to run the other eleven and keep this one locked.