Every operator I know has argued about closing time. It is a strangely emotional discussion. Someone on the team wants to keep the doors open until midnight because "we get a good hour at 10:30 sometimes." Someone else wants to close at 9 because "we are just paying two cooks to stand around." Both are usually working from anecdote. Neither has run the number.
The reason the argument is so common is that late night sits in a blind spot on the daily P&L. The daily rollup shows total revenue and total labor. It does not show revenue by hour or labor by hour, so nobody can see what the last two hours actually cost. The general manager knows something feels off. The regional director sees a labor variance and cannot pinpoint it. And the debate keeps going, powered by feelings.
The framework below is the one I have used across the Hana Group's 21 franchise units embedded inside big-box retailers and inside the Zareen's Bay Area group where late-night hours look very different on a Palo Alto Tuesday than a Mountain View Friday. It is not complicated. It just requires you to look at the number nobody has been looking at.
Step 1: Chart sales and labor by the hour, not by the daypart
Pull twelve weeks of hourly sales and hourly labor for the location. Not dinner as one number. Not late night as one number. By the hour. Most POS systems can produce this report. Toast, Square, and Aloha all export the raw data if you ask.
What you are looking for is the curve. Not the peak. Not the average. The curve. A healthy dinner service has a build up from 5:30, a peak around 7:15 to 8:00, and a graceful decay to closing. An unhealthy late-night tail looks like a peak that dies at 9:00 and a two-hour flat line at low sales after that. If your tail is flat at 15 percent of peak-hour sales and your labor tail is at 60 percent of peak-hour labor, the last two hours are almost certainly losing money.
Step 2: Contribution margin per hour, honestly
Contribution margin per hour is the only number that matters. It is:
(Sales that hour) minus (food cost of items sold that hour) minus (all labor for that hour, including the closing tail that continues after the door locks) minus (variable overhead for that hour, which is small but not zero).
The closing tail is the piece most operators get wrong. If you close the door at 11 pm but two cooks stay until 11:45 to break down the line and one server stays until 12:15 to close out checks and one manager stays until 12:45 to reconcile the drawer and lock up, you have another hour and forty-five minutes of labor charged to that last hour of service. The rule of thumb: add 30 to 60 minutes of full-line labor and 60 to 90 minutes of manager and closing-server labor to the last operating hour on your CM calculation.
Do that math and you will usually find the number is negative for the last hour, and often negative for the last two hours, unless the trade area actually delivers late-night traffic.
Fig. 1 · Where the money stops being made.
Step 3: Price in the hidden costs
The direct contribution margin math above is only the visible number. There are five hidden costs that make late-night more expensive than it looks.
- Higher turnover on the closing shift. Line cooks who close on Friday and Saturday burn out faster. The re-hire cost is real, usually $2,000 to $4,000 per line cook, and it hits the P&L three to six months later. Amortize it back into the shift that produced it.
- Small equipment breakage. A longer operating day is more wear. Fryer thermostats, dish machine gaskets, walk-in door hinges. Fifty extra dollars a week of small repair, minimum, over a full year.
- Security incidents. The last two hours are when a walkout or a fight or a smash-and-grab is most likely. Not every night. Enough to matter over 365 days.
- Sanitation depth. When the kitchen runs longer, the deep clean happens faster and worse. Health scores drift. Health scores drift into fines, and health scores drift into guest perception.
- Manager cognitive load. The closing manager is running the whole building alone at 11 pm, doing service and cash and safety and staff management. Their next-day performance is worse. This is not a soft cost. It is a real operational drag on the following day.
Assign a dollar tax to each. In most casual full-service concepts I have looked at, the hidden cost stack adds another $60 to $120 per late-night hour on top of the direct labor. That takes an hour that looks flat on the direct math and turns it clearly negative on the honest math.
Step 4: Decide by unit, not by group
The next mistake is uniform hours across a group. Two locations of the same brand can be a mile apart and have completely different correct closing times. A location next to a movie theater has real 10:30 pm traffic on Friday. A location in a suburban strip mall does not. Forcing them to close at the same time either loses money at one or leaves money on the table at the other.
Uniform hours are a marketing convenience. They should not be an operating default. If the brand promise requires it, absorb the cost consciously. If it does not, let each location run the hours that fit its trade area, and put the correct hours on the individual location's Google Business Profile and website.
The three viable late-night patterns
Once you have run the numbers you usually land in one of three patterns, and each has a distinct playbook.
Pattern A: Close at nine
The trade area does not deliver late-night traffic. The last two hours are consistently losing money on contribution margin. The right move is to close earlier than the market says you "should." The catch is transitioning without hurting the brand.
Do it in one move, not in creep. Announce a 30-day advance change. Update every listing (Google, Yelp, DoorDash, Uber, Grubhub, your website) the same week. Tell your regulars in person. Post a friendly sign for the first 30 days. Guests forgive an announced closing time change. Guests do not forgive an unofficial early close where they arrive at 9:45 to a locked door.
Pattern B: Extend to midnight or later, intentionally
The trade area does deliver late-night traffic (downtown, near a college, in an entertainment district) and you have the labor structure to serve it well. The right move is to build a real late-night program, not to leave the doors open on the same operating model.
Late-night as a program has a smaller menu (10 to 15 items that produce fast, hold well, and are alcohol-friendly), a bar-forward labor plan (fewer servers, one strong bartender, a smaller kitchen crew) and a distinct experience (lower lights, music tempo shift, price point that respects the guest is out later). Run right, this is often the strongest CM window of the week because the bar mix is high and the food mix is simple.
Pattern C: Weekend-only late night
The middle case. The trade area supports late night Thursday to Saturday but not Sunday to Wednesday. Split the schedule. Weeknights close at 9 or 9:30. Weekends stay open to midnight or 1 am with a modified menu. This is the pattern that fits most urban full-service concepts and most of the Zareen's units where the office lunch driver on weekdays did not carry into a weekday late-night dinner but the Friday and Saturday date-night traffic did.
The correct closing time is not a policy. It is a math answer, and the math is different at every location.
What to watch after the change
Whichever pattern you land on, track three numbers weekly for the first 90 days after the change.
- Contribution margin per operating hour for the last two hours of the day, versus the prior baseline.
- Guest complaints tagged "hours" or "closed early" from all review channels. Any bump here means the transition messaging is not landing.
- Closing shift turnover among line cooks and closing servers, versus the prior baseline. If pattern B added late-night hours, is the closing crew burning out?
The number that will move fastest is contribution margin. The number that will surprise you most is closing shift turnover. That one determines whether the change is durable.
What I would tell a first-time operator
Do not decide late-night hours from your gut. Do not decide them from what "everyone in the neighborhood does." Do not decide them from the marketing team's preference for uniform hours across the brand.
Decide them from the twelve-week hourly report, adjusted for the closing labor tail, taxed for the five hidden costs, and applied unit by unit. When you do, the answer usually shows up quickly and it is often not what anyone predicted.
The Zareen's Cupertino unit closed 45 minutes earlier for one quarter and picked up roughly $18,000 in operating contribution across the year. The Palo Alto unit extended weekend hours by an hour and picked up roughly $32,000. Same brand. Opposite moves. Both right for their trade area.
Cadence beats charisma, and math beats habit. Late-night hours are a math problem. Solve it that way.