Scheduling
Scheduling to a sales forecast, not to habit
Most restaurant schedules are copied from last week, which was copied from the week before. That is not a plan, it is an inheritance. Scheduling against expected sales is the single biggest lever on labor cost percentage and it costs nothing to start.
Step 1: a forecast you would actually defend
You do not need a statistical model. Take the same week last year, adjust for how this year has been running against last year, then adjust for what you know: a holiday weekend, a street closure, a festival two blocks over, a forecast of rain on your patio days.
Write it down per day, not per week. A weekly total hides the thing you are trying to see, which is the shape. Wednesday and Saturday need different staffing, and a weekly number cannot tell you that.
Step 2: schedule to the shape of the day
Within a day, sales are not flat either. A lunch peak that ends at 1:30 does not need the same coverage at 3pm, and the shift that runs 11 to 7 because it always has is often two hours of paid standing around at the back end.
This is where the savings usually are, and they are less painful than cutting whole shifts. Trimming the tail of three shifts is worth more than cutting one person entirely, and it does not leave a hole in your coverage.
Step 3: check the draft before you post it
This is the step that gets skipped, and it is the one that matters. Once a schedule is posted, changing it costs goodwill and in many jurisdictions actual money under predictive scheduling rules. The check has to happen while it is still a draft.
Three questions are worth asking every time. What does this schedule cost as a percentage of the sales I expect? Is anyone scheduled into overtime I did not intend? And is there a shift where one person is covering something that genuinely needs two?
That last question is not hypothetical. Smooth Ops has caught a single team member scheduled for a restaurant's busiest shift of the week, before the week ran. Nothing about it looked wrong in a grid. It looked wrong the moment it was checked against expected sales.
Price every trim before you make it
"Cut a few hours" is not a decision, it is a wish. Four hours off a Tuesday close is a specific amount of money and a specific reduction in coverage. When both are visible, the conversation with a manager stops being a negotiation about vibes.
It also stops you cutting the wrong four hours. The cheapest hours to remove and the safest hours to remove are not always the same ones, and that tradeoff is only visible when the dollar value sits next to the coverage impact.
How Smooth Ops does this
The Schedule module reads shifts and sales from Square, so the forecast is built from your own history rather than typed in. Next week's draft is checked against it, with suggested trims priced in dollars, and coverage gaps and missed clock-outs flagged with a fix next to each. Labor lands as a percent of sales each week with overtime and payroll tax counted.
Read next to food cost, that gives you both halves of prime cost, and labor cost versus food cost covers telling them apart when the combined number climbs.
$25 a month per location, or $99 for every module. 30 days free, no credit card.
Scheduling to sales, answered
How do I forecast sales for next week?+
Start with the same week last year if you have it, adjust for how this year has been tracking against last, then adjust again for anything you know about: a holiday, a local event, weather that will keep a patio empty. You do not need a model. You need a number that is close enough to schedule against and honest enough that you would defend it.
What if my sales are unpredictable?+
They are usually more predictable than they feel. Day of week is the strongest pattern in most restaurants and it is remarkably stable. Even a forecast that is regularly off by ten percent is far more useful for scheduling than no forecast at all, because the shape of the week is what you are scheduling against, not the exact total.
Is it worth scheduling to a forecast for a small team?+
Often more so. With four people on payroll, one unnecessary eight hour shift is a much larger share of your labor cost than it would be in a thirty person operation. Small teams also have less slack to absorb a shift that was scheduled thin.
How far ahead should I post the schedule?+
Far enough that your team can plan their lives, which in many places is also a legal requirement under predictive scheduling rules. That is exactly why checking the draft matters: once it is posted, changing it costs you goodwill and sometimes money, so the check has to happen before it goes up rather than after.
Does Smooth Ops build the schedule for me?+
It reads shifts and sales from Square, checks next week's draft against the forecast, and suggests trims with dollar values attached. You decide what to cut. A tool that rewrote your schedule without knowing that one person cannot open, or that two of your staff cannot work the same shift, would not survive contact with a real restaurant.
Check next week before you post it
Labor against your Square sales, coverage gaps flagged, and every suggested trim priced.
30 days free, no credit card