Activity-Based Costing for Restaurants: a 2026 Guide to Costing the Time Behind Every Dish

Step 1: map the kitchen's activities before you touch a stopwatch

The first deliverable of activity-based costing is a short list of activities, between five and eight, describing what the brigade actually does with each dish: mise en place, cooking on the griddle or the range, plating, the pass, table service and dishwashing. With more than eight, measurement won't hold up during a real shift, and with fewer than five you blend tasks that cost different amounts, like the sofrito made once for forty plates and the sear done one by one. Each activity gets its station and its owner, because the cost will come from that station's payroll and not from the whole restaurant's. The check is simple: if a cook reads the list and can't tell where something they do every day belongs, the map is incomplete and should be fixed now, BEFORE a single minute is measured.

Step 2: time each activity using two weeks of tickets

Minutes per dish are measured during real service, not estimated in a meeting, so the deliverable is a table with the average time of each activity per dish, drawn from at least two weeks of tickets. The POS timestamp gives you the start and the pass, while the station chef's stopwatch covers what the system can't see, such as assembling a dessert or how long the griddle stays tied up with a thick cut. Measure slow days and packed days, because the same risotto eats more minutes when the kitchen is slammed, and that is where the hidden cost lives. Your head chef will distrust the exercise, and rightly so, if it feels like people are being timed: explain that the dish is being measured. The proof for this step is that the summed minutes of a shift never exceed the brigade's paid hours, and that gap is idle capacity.

Step 3: turn each station's payroll into a cost per minute of practical capacity

Each activity's rate comes from dividing the station's monthly cost by the minutes that station can genuinely work, not by the minutes on the payroll. For a sense of scale, the median hourly wage of food and beverage serving workers in the U.S. was 15.24 USD in May 2025 (BLS). For example, if your grill station costs you 18 USD an hour including payroll burden and works 48 productive minutes out of every 60, the practical minute comes to 0.375 USD, not 0.30. That practical-capacity adjustment is what separates TDABC from classic survey-based ABC, which asks each person what share of the day goes to each task and almost always adds up to more than reality. The deliverable is a per-minute rate for every station, reviewed with your accountant and dated with the payroll period that supports it.

Step 4: load the minutes onto each dish and rank the menu by margin per minute

With minutes and rates in hand, each dish gets its activity cost, reported next to food cost and never added into it. In the Masterestaurant method the recipe is still checked against the 32 % maximum calculated on ingredients alone, and the activity cost answers a different question: whether the contribution margin covers the brigade time the dish consumes. The reading Diego F. Parra proposes is to rank the menu by contribution margin per kitchen minute, because that exposes what classic menu engineering hides, a best-selling star that drops to the bottom of the list. For example, a dish with 12 USD of margin that takes 10 minutes yields 1.20 USD per minute, while one with 9 USD that leaves in 4 minutes yields 2.25. The deliverable is that ranked list, with a written decision for every dish at the tail: reprice it or redesign its process.

Maintenance: recalculate per-minute rates every quarter

Per-minute rates are recalculated whenever payroll or the menu changes, and at least once a quarter, because the cost of time moves faster than the recipe card. The price environment pushes the same way: USDA ERS observed a 3.8 % rise in food-away-from-home prices in the U.S. during 2025 and forecasts 3.5 % for 2026, and anyone who raises the whole menu evenly on those references punishes the fast dish and the slow one alike. My recommendation is to apply the increase where the cost per minute justifies it and protect the price of dishes that go out in a few moves, since those are the ones regulars compare. What happens if you keep the old rate? Every minute is underestimated, the slow dish looks profitable for another quarter, servers keep upselling it and the hole shows up in the cash count with nobody knowing where it came from.

The mistakes that sink restaurant activity-based costing, and how to avoid them

The most repeated mistake is folding labor into food cost so the recipe card looks «real», which leaves you with a metric that can no longer tell whether the problem is the supplier or the brigade. Next comes measuring a single shift, because a quiet Tuesday produces minutes that Friday won't honor and the cost per dish comes out optimistic. Another classic is spreading the station's cost over paid minutes instead of practical ones, so dead time vanishes from the calculation and nobody manages it. And the last one, which I admit I made for years, is trying to cost every drink and every side from the first month; start with your ten best sellers, because precision on dishes that rarely sell doesn't pay for the hours of measuring. All of them are avoided with the same discipline, reporting separately and measuring full weeks.

How to know your activity-based costing is right: the closing checklist?

Activity-based costing is right when five checks pass at once, and a single failure sends you back to the step that produced it.

First, the minutes measured in a shift don't exceed paid hours, and the difference is recorded as idle capacity with its dollar cost. Second, every top-selling dish shows ingredient food cost and activity cost in two separate columns, with food cost still under the method's maximum. Third, the menu is ranked by yield per minute and every dish at the tail has a decision with an owner and a date. Fourth, the rates carry the date of the payroll they come from. The fifth is the one almost nobody checks: the head chef understands the table and uses it to choose the daily special. Once all of that holds, a single action remains, checking after thirty days whether the repriced dish improved its margin per minute.

What really changes when you cost by activity?

Here is the trade's paradox: ABC needs labor inside the calculation, and the Masterestaurant costing rule keeps labor off the plate. Both hold because they answer different questions.

Recipe food cost is a purchasing and portioning KPI measured against the 32% ceiling, and once payroll gets in you can no longer tell whether the problem is the supplier or the crew; activity cost is a capacity KPI read in minutes and in contribution margin per kitchen hour. Calculate them together, report them apart. The least discussed finding in the Heliyon study isn't the cost gap, it's capacity: second-level cooks at that fine-dining restaurant were working at 70.89% of practical capacity. That number changes the conversation with the chef, because spare time is not a cost of the dish but a scheduling decision, and traditional allocation buries it by spreading it over everything sold. What happens if you charge that idle time to the dishes?

What really changes when you cost by activity — in practice?

Every cost rises evenly, the fast high-margin dish starts looking mediocre and menu engineering drops it into the dog quadrant;

you cut it, the kitchen loses the item that filled dead minutes, idle time grows and the next calculation inflates everything again. The spiral ends with a shorter, pricier menu and the same crew, which is why sound ABC uses practical capacity and reports idle time on its own line. I got this wrong for years: I costed labor per dish with a simple ratio on sales and trusted it because it tied out to the P&L. Of course it tied out, the total doesn't change; what changes is WHO pays for it inside the menu. Since then, in Diego F. Parra's work with owners, ABC comes in only after recipes are properly costed and restaurant inventory closes with small variances, because before that, timing minutes polishes a number that is crooked at the source.