A group opens its third or fourth site and someone works out that three kitchens are each making their own version of the same stock and sauce base and braise, to three different standards. Consolidating that into one production kitchen is usually a good decision. It fixes consistency and concentrates skill in one place, and it takes hours of prep out of sites that were short on space anyway.
What it also does, and what almost nobody plans for, is change the shape of your cost question. Before, an ingredient came from a supplier and went into a dish. Now there is a step in the middle, and that step has a price nobody set deliberately.
Why Groups Centralise Their Prep
The reasons are rarely about cost, which is worth saying up front because it shapes what happens next.
Consistency is what usually starts the conversation. When the same ragù comes out of one kitchen instead of four, it tastes the same in every dining room, and the customer experience stops depending on who was on prep that morning. For a group building a brand across sites, that matters more than a percentage point of food cost.
Then there is the space problem. Restaurant kitchens are small, and prep is the thing that loses the argument for room every time. Moving it out frees up line space, and it often lets a site trade with fewer people on shift.
Buying power is where the cost case tends to get made, and it is the argument most likely to appear in the business plan. One kitchen buying for four sites buys in bigger quantities at better prices, and it holds less total stock than four kitchens each keeping their own safety buffer.
All three are real. What none of them do is tell you whether the central kitchen is actually cheaper once it exists, and that is a different question with a different answer.
What Breaks in the Numbers When They Do
The moment prep centralises, your cost chain grows a link, and most systems were only built for the original two.
Before centralising, the chain is short enough to hold in your head. An invoice tells you what an ingredient cost and the recipe tells you how much of it goes into a dish, so multiplying the two gives you a dish cost you can trust.
Afterwards, the outlet receives braised beef, sauce base, or portioned fish from another part of your own business, instead of buying carrots directly. The invoice that started the chain arrived somewhere else, days earlier, for something that no longer looks like what showed up at the outlet door.
So the outlet’s dish cost now depends on a number somebody has to decide: what the central kitchen charges for the thing it made. And that number carries the ingredients, the yield loss, the labour, and the waste from the production run. Get it wrong in one direction and the outlets look profitable while the central kitchen absorbs a loss nobody has attributed. Get it wrong in the other and the outlets look expensive for reasons that have nothing to do with how they are run.
This is the same failure that makes group averages misleading in general, arriving through a different door. The group total is right. The site-level picture is fiction.
Prebatch and Transfer Pricing Between Sites
The item the central kitchen produces is a prebatch: a made component that behaves like an ingredient everywhere downstream. Getting its cost right is most of the work.
A prebatch cost has three parts. It starts with the ingredients that went in, priced from real invoices. Then it has to account for yield, because thirty kilos of raw shoulder does not become thirty kilos of braise, and the difference belongs in the cost of whatever survived. Production labour is the third part, and it belongs in there if you want the number to reflect what the thing actually costs to exist.
Yield is where most groups are furthest out, because it is the part nobody measures and everybody estimates. A stock reduces by half, a fish loses forty percent of itself to bones and trim, and a braise renders down while it cooks. If your prebatch cost assumes raw weight in equals finished weight out, every dish built on it is priced against a fantasy, and the error compounds across every site that uses it.
The labour question is a genuine choice, and there is no single right answer to it. Including production labour in the transfer price gives you a true cost per portion and makes the outlets carry their share of the central kitchen’s payroll. Excluding it keeps food cost as a clean food-only number and leaves the central kitchen’s labour as a group overhead. Both work. What does not work is doing it inconsistently, or never deciding, which is the common case.
Then there is the ordering rhythm itself. Outlets order from the central kitchen the way they would from any supplier, working to a schedule and ordering against a list. Treating that flow with the same rigour you would apply to an external supplier is what keeps the whole picture honest.
Counting Stock That Moves Between Locations
Physical counting gets harder in a way that catches groups out in their first month.
Stock is now in motion. At any moment some of it sits in the central kitchen as raw ingredients, some as finished prebatch waiting for collection, some in a van, and some already on an outlet’s shelves. Count on the wrong day and the same beef is either double counted or missing entirely, depending on which end you were standing at.
The answer here is a rule, not a piece of technology. Set a group-wide count day and count every location on it. Then decide explicitly whether stock in transit belongs to the sender or the receiver, and never change that answer afterwards, because the comparison across periods is what you are actually protecting.
Transfers themselves need recording as they happen, not reconstructed afterwards from delivery notes. A transfer that leaves the central kitchen and arrives at a site is two movements in your records, and if only one gets logged, the difference lands in variance at both ends with opposite signs. That is a particularly frustrating kind of variance to chase, because the group total nets to roughly zero while both individual sites look wrong.
Groups that count on a shared day and log transfers as they move tend to find their multi-location food cost picture becomes legible for the first time.
Whether Your Central Kitchen Is Actually Cheaper
The whole exercise exists to settle this, and most groups still cannot say either way.
The comparison you want is straightforward to state. For a given component, what does it cost produced centrally, all in, against what it would cost bought finished from a supplier or made on site? Add up the ingredients, the yield loss, the production labour and the transport, then set that against the alternative price.
Some components will win comfortably. Anything with high labour and cheap ingredients, or anything where consistency has commercial value, usually justifies itself easily. Others will not, and the ones that lose tend to be the items where an external supplier has scale you cannot match. A central kitchen making its own stock from bones is often a clear win. A central kitchen portioning something a supplier would portion for pennies is often not.
Running that comparison component by component, instead of reaching one verdict on the whole kitchen, is what turns centralising from an article of faith into a decision you can revisit. It also tells you what the central kitchen should make more of, which is a more useful output than a yes or no.
Frequently Asked Questions
What is a central kitchen in a restaurant group? A dedicated production site that prepares components, sauces and portioned items for several outlets instead of serving customers directly. It is sometimes called a commissary. The outlets order from it and finish the dishes on their own line.
How do you cost a transfer between a central kitchen and an outlet? Price the prebatch on the ingredients that went in, adjusted for real yield. Then decide once, and for good, whether production labour is included. That transfer price becomes the ingredient cost in every outlet recipe that uses it.
Why does variance get worse after centralising? Usually because transfers are not being logged as they move, so stock leaves one location’s records without arriving in another’s. It also shows up when count days differ between sites, because stock in transit gets counted twice or not at all.
Should every group have a central kitchen? It depends on what you make. Components with high labour and low ingredient cost, or where consistency across sites carries commercial value, are the strongest candidates. Items an external supplier can produce at scale often are not worth centralising.
Getting the Middle Step Right
A central kitchen answers the consistency problem, the space problem and the buying-power problem all at once. It is also a new link in your cost chain, and it only stays honest on three conditions: prebatch costs built on real yields, transfers logged at the moment they move, and one count day shared across every site.
Stockifi handles central kitchen production and inter-site transfers inside the same picture as the rest of your stock, so a prebatch carries its real ingredient cost through into every outlet recipe that uses it. See how central kitchen works.