What I got wrong running a commissary at 8,000 meals a day
The production board taught me the business was not what I thought it was
You learn things standing in a commissary at production hour that you cannot learn from a report. You learn which combi oven actually holds temperature and which one somebody has been babysitting for two years without telling anyone. You learn who arrives early and quietly re-does part of the prior night's prep rather than raise it with anybody. None of that is in the system.
What the floor did not tell me, and what took me far longer to see, was that the building was running on a plan nobody had written down. The board on the wall said what to produce. Nobody could tell me where the numbers on the board came from. They came from last week. Last week came from the week before. Somewhere back there was a real decision, made by somebody who had since left, about a satellite count and a menu mix that no longer existed.
That is the actual shape of most commissary problems I have seen since. It is not a bad crew or a bad building. It is a system that was correct once and has been drifting quietly ever since, and no one has been assigned the job of noticing.
What an uncontrolled handoff looks like
The pattern I have chased more than any other starts with a phone call. A satellite rings mid-morning to say product did not come off the van, service is inside the hour, and somebody has to drive. You do what everyone does: send the product, make service, tell the team good work.
Then it happens again. What I have found underneath those calls is almost never theft and almost never sloppiness. It is loading sequence and route order, and a dispatch process where the person loading the van and the person receiving it are not working from the same document. When the van gets tight the loader pulls from whatever is reachable, and product goes off at a stop that did not order it and gets put away without a word.
The cost never shows up as a line item. It shows up as an emergency run, as product written off, and as a site manager who quietly concludes the commissary does not respect her building. That is what an uncontrolled handoff looks like from the inside long before it looks like money.
Four things I got wrong
I am listing these as mistakes rather than principles on purpose. Every one of them felt reasonable at the time, and that is exactly why they lasted.
Mistake 1
I treated the production sheet as a plan when it was really a memory
For a long stretch, the production sheet was built from what we produced the same day last week, nudged up or down by feel. That is not a forecast. That is an echo. It works right up until the mix moves, and the mix always moves. A single catering contract landing on a Thursday would throw the whole board off, and nobody would see it until Friday afternoon when two sites were short and one was buried.
Mistake 2
I let the transfer sheet be a courtesy instead of a record
We had a sheet. Everybody signed it. Nobody reconciled it. If a site was short two pans, they called, we sent more, and the paper never caught up. What that did to my numbers took me months to understand: the waste showed up on the satellite P&L, so I kept coaching satellite managers about a problem I was creating in my own building.
Mistake 3
I trusted cold chain because the people were good
This is the one I am least proud of. I had a strong crew, so when product turned my instinct was that it was an isolated thing. It rarely is. Cold chain does not fail in the walk-in where the thermometer already lives. It fails in the gaps - a van holding too long partway down a route, a dock, the minutes between dispatch and receipt - and none of those gaps had a number attached to them. Trust is not a control. Product you have to dump is what teaches you that.
Mistake 4
I staffed the commissary for the worst week of the year
I scheduled so that peak week would never hurt. Which meant forty other weeks carried labor they did not need, and I could not see it, because the schedule looked stable and stable felt like competence. Flat labor against a moving volume curve is not stability. It is a slow leak with good posture.
What I actually believe now
Every commissary problem I have chased eventually turned out to be a handoff nobody owned.
Production to dispatch. Dispatch to van. Van to site. Site to service. The failures do not live inside those steps. They live in the gaps between them, where two people each assume the other one has it, and where there is no number that would prove either of them wrong.
What I would do differently on day one
If I walked into a commissary tomorrow, I would not touch the floor plan and I would not change a recipe for at least a month. I would spend the first week doing three boring things.
First, I would sit with whoever builds the production sheet and ask them, line by line, where each number comes from. Not to catch anyone out. To find out how much of the plan is a forward signal and how much is memory. In my experience it is mostly memory, and the person building it usually knows that and has been waiting for somebody to ask.
Second, I would ride the route. All of it, one full day, including the stop where everyone is annoyed. You cannot understand a transfer problem from the transfer sheet, because the sheet only records what people intended. The van records what happened.
Third, I would put a temperature number on every handoff and then do nothing with it for two weeks except look at it. Not as a compliance exercise. As a way of finding out where the cold chain actually sags, which is almost never where the thermometer in the walk-in says it is.
After that, the work is unglamorous and mostly about cadence: one reconciliation meeting a week, at the same hour, with a named owner on every open discrepancy, and a production plan built from confirmed forward demand rather than last week's output. The gain is not in any single fix. It is in the fact that somebody is now assigned to notice.
That is roughly the sequence we ran here, with MarketMan rolled out across all 14 sites so a transfer means the same thing at both ends of it. Food waste came down 14 percent over nine months and labor efficiency improved 3 to 6 percent. Neither of those came from a clever fix. They came from handoffs that finally had a number and an owner attached.
Early signals your commissary is drifting
These are the things I wish I had taken seriously six months earlier than I did. None of them are dramatic. That is the point. By the time a commissary problem is dramatic, it has already been true for a long time.
- A satellite calls for product more than once a week and nobody logs the call
- Your production lead can tell you what you made yesterday but not what is confirmed for Thursday
- The transfer sheet has signatures but no reconciliation meeting attached to it
- Waste sits on satellite P&Ls and never on the commissary's
- Nobody can name the person who owns a discrepancy line
- Overtime is discussed at month-end instead of the same hour every week
Related reading
If you want this fixed in your operation
Everything above is what I learned by getting it wrong first. Rebuilding the production plan, the transfer record, the cold chain, and the labor curve as one connected system is the work I do through XenoSoft, and the scope for that lives over there rather than here.