By Matt Wampler, CEO of ClearCOGS
Almost every business case for prep forecasting gets built on the cost side. Less waste, tighter food cost, fewer hours spent counting. Those numbers are real and they are easy to defend, which is why they end up in the deck.
For a concept that lives on lunch, they may also be the smaller half of the story.
An operations leader at a multi-unit brand put it to me plainly a while back. His priority was not food cost. It was ticket times. His reasoning was that the office worker on a lunch break does not have an hour, and every minute the kitchen adds to a ticket makes it less likely that person comes back next week. He was not describing a hospitality problem. He was describing a revenue problem with an operational cause.
That cause starts hours before anyone orders anything.
Slow tickets usually start in the walk-in
Here is the mechanism, and it is worth being precise about because it gets mistaken for a staffing problem constantly.
A line runs on components that were prepared in advance. Portioned protein, cut produce, batched sauces, whatever your build requires. During a rush, a cook assembles. That is the whole design, and it is why a kitchen can put out far more covers per hour at noon than the same people could cook from raw.
Now take one component to zero at 12:15. The cook stops assembling and starts producing. Throughput on that station collapses to the speed of the slowest upstream step, and because tickets are a queue, the delay does not stay on one order. It propagates to every ticket behind it, including the ones that did not want that item.
Notice what has not happened. You did not 86 anything. Nothing shows up in a stockout report. The guest gets the food they ordered. The only visible symptom is that tickets got slow for forty minutes and nobody can say exactly why.
This is why ticket-time investigations so often end in a shrug and a conversation about staffing levels. The failure did not occur during the rush. It occurred at nine in the morning, when someone decided how much to make.
What the wait actually costs
The instinct is to treat this as a service-quality issue, which makes it feel soft and keeps it out of the business case. The research says it belongs in the business case.
Researchers analyzed operational data from 94,404 customer groups at a single busy restaurant over twelve months to isolate what waiting does to behavior. Longer waits were associated with three separate outcomes: customers leaving before being served, customers spending less time dining, and, most importantly for a lunch concept, a longer interval before those customers came back.
They then built a simulation using those empirical relationships to estimate the system-wide effect. In that model, eliminating waiting entirely would have raised the restaurant’s total revenue by nearly 15 percent (De Vries, Roy, and De Koster, Journal of Operations Management, 2018).
Two caveats matter. That is one restaurant, and the 15 percent comes from a simulation of a hypothetical zero-wait scenario rather than an achievable target. Do not put it in a forecast.
What it does establish is direction and rough magnitude. Waiting does not merely annoy people. It shortens the visit, it sends some guests away before they order, and it pushes out the date of the next visit. Those are revenue effects, they compound across a year, and none of them appear anywhere on a P&L line you could point at.
Why lunch is the daypart where this bites
Dinner guests have elastic time. Lunch guests do not.
A weekday lunch customer is working against a fixed budget of minutes, and your ticket time consumes it directly. Cross their threshold and they do not complain, they just quietly stop choosing you on Tuesdays. The revenue does not disappear in a visible event. It thins out.
This is also the daypart where demand is most concentrated, which means small prep errors have outsized consequences. A location that does a meaningful share of its daily covers inside a ninety-minute window has very little room to recover. Running short at 12:15 is not the same event as running short at 3:00, even if the quantity is identical.
Which is why a forecast built on daily totals is close to useless here. Knowing you will do a given number of covers today tells you almost nothing about whether the line will hold at 12:15.
Nobody owns the join
Here is the measurement problem underneath all of this. Most brands already collect everything needed to see the pattern, in four systems that never talk.
| System | What it knows | What it cannot tell you |
|---|---|---|
| Kitchen display | How long each ticket took, by station and minute | Why a given window went slow |
| Point of sale | What sold, at what time, in what mix | What was actually ready to assemble |
| Prep sheet | What the team was told to make, or decided to make | Whether that amount matched the peak |
| Inventory system | What was consumed across a period | What happened between 12:00 and 1:00 |
Every one of those is useful. None of them, alone, can answer the only question that matters: when tickets ran slow yesterday, what had run out.
Joining the first three gets you most of the way. Pull your worst ticket-time windows from the last month, then look at what the prep plan called for on those days and what actually got made. The pattern usually announces itself within a few minutes of looking, and it is almost always concentrated in a handful of components rather than spread across the menu.
What to change
Four things, roughly in order of effort.
- Forecast the peak window, not the day. Demand inside a lunch rush is not a flat share of daily volume, and the items that spike are not always the items with the highest daily count. Production targets should be built from demand in short intervals, because that is the interval at which your line either holds or does not.
- Stage to the peak, not to the shift. The relevant question is not how much you will sell today. It is how much has to be physically ready at the pass by 11:45. Those are different numbers, and the second one is the one that determines ticket times.
- Rank the prep list by consequence. Some components stop the line when they run out. Others are an inconvenience. Your prep list probably lists them in whatever order it was typed in years ago. Put the line-stopping items at the top so that when the morning runs short on time, the right things get made first.
- Review ticket times and prep together. Once a period, put the two in front of the same person. As long as ticket times belong to operations and prep belongs to the kitchen, the connection between them stays invisible.
Run both numbers
If you are building a case for forecasting and you only have the cost side, you are presenting the smaller argument, and possibly a less persuasive one. Cost savings are a defensive story. Throughput is a growth story, and executives hear those differently.
The cost case says you will throw away less. The throughput case says the same locations, the same staff, and the same four-hour window will serve more people and see them again sooner. In a lunch-driven concept, the second number can be larger than the first.
Both come from the same decision, made at nine in the morning, by someone holding a sheet of paper.
This is the work we spend our days on at ClearCOGS: turning a restaurant’s own history into a specific quantity for each item, in the interval where it matters, before the shift starts. If your lunch ticket times are moving in the wrong direction and nobody can explain why, that is usually a prep question rather than a staffing one.
Sources
- De Vries, Jelle, Roy, Debjit, and De Koster, Rene. Worth the Wait? How Restaurant Waiting Time Influences Customer Behavior and Revenue. Journal of Operations Management, 63, 59–78. November 2018. sciencedirect.com
