Kitchen brigade stations

Why does a kitchen where nobody cooks a whole dish serve whole dishes faster?

In a restaurant running Escoffier's brigade, no single cook makes your plate. The sauce belongs to one person, the protein to another, the vegetables to a third, the cold garnish to a fourth. Not one of them could produce your dish alone in the time available.

Yet this kitchen plates faster, and more consistently at volume, than one where each cook takes a ticket and makes the whole thing. That should feel slightly wrong: we have added handoffs, added coordination, and removed anyone with the whole picture. Where does the speed come from, given that we have plainly made the process more complicated?

The tempting answer is skill: do one thing all night and you get better at it. True, and not sufficient — a good cook can execute any of those four elements well. Skill alone would predict a modest edge, not the gap you actually see. So look for something structural.

Ask first what a lone cook spends time on that is not cooking. She reaches for a different pan, wipes a board, fetches a herb from another fridge, drops the heat on one burner while she starts another. Every one of those is a changeover — the cost of switching a workspace from one kind of task to the next. A station cook pays that cost essentially once, at mise en place, before service. Her board, her pans, her prepped components and her heat are arranged for exactly one family of operations, and she never reconfigures. The brigade did not make the changeovers faster; it removed most of them from service entirely.

Now the larger effect. Take a dish that needs a protein seared for eight minutes, a sauce finished in three, and a vegetable in four. How long does one cook need? She cannot do all three at once with full attention, so her work tends toward the sum. How long does the brigade need? The three run concurrently, so the ticket takes roughly the longest element rather than the total.

That is the real engine, and it is worth naming precisely: specialization here is not primarily about being better at a task. It is about converting a sequence into a parallel operation. The plate's components were never dependent on each other — they only appeared sequential because one pair of hands was doing them.

But parallel work creates a problem the lone cook never had. Three people finishing independently do not finish together, and a plate is only correct if everything arrives hot at the same moment. So ask: what did the brigade have to add to make parallelism safe?

It added synchronisation. The expediter — the aboyeur at the pass — calls the ticket, then calls the fire times so each station starts at the offset that lands its component at the same second. The pass is where the parallel branches rejoin. Notice that this role produces no food at all; it exists purely because parallel work needs a rendezvous. That is the price of the design, paid in a whole person.

There is a third gain, quieter than the other two. Because tickets flow through stations rather than through individuals, a station can batch: the grill cook with four tickets carrying steaks lays them down together. A whole-dish cook cannot batch across tickets without abandoning the one in front of her.

And the costs? They follow just as directly. Handoffs create coupling: if the sauce station falls behind, every ticket touching sauce falls behind with it, and the pass stalls holding hot components. A whole-dish kitchen degrades gracefully because each cook is independent; a brigade degrades as a system. It is also brittle to absence, since a missing chef de partie has no natural substitute, and it is slower to change, because a new dish must be decomposed and distributed across stations before anyone can cook it.

The misconception worth correcting is that the brigade is a hierarchy dressed up as an efficiency. The ranks are real, but the speed does not come from the chain of command. It comes from two things: eliminating changeover during service, and running independent components concurrently.

A second correction: specialization is not always the right answer, and the same reasoning tells you when it is not. Its gains scale with volume and with menu stability, because both are what let a station's setup stay untouched. In a twenty-cover kitchen with a menu that changes daily, there is no changeover cost worth eliminating and no parallelism worth coordinating — and the coordination overhead, which does not shrink, dominates. That is why small kitchens genuinely do run on whole-dish cooking, and are right to.

Finally, note where the throughput ceiling actually sits. It is not the slowest cook; it is the busiest station relative to the menu's demand on it. If tonight's specials all lean on the saute station, adding hands elsewhere buys nothing. The brigade's structure makes that bottleneck visible, which is arguably its most underrated property.

• How the brigade's bottleneck station is identified and rebalanced when a menu changes, and why menu engineering is partly a capacity problem.
• Why fast-casual assembly lines and the classical brigade look similar but optimise for different things.
• What happens to the model when a single cook covers two stations on a slow night, and where the crossover point sits.

Every term the collection defines is gathered in the glossary.