Written by The Pragmatic CFO. 15+ years running restaurant P&Ls.
This article is part of The Operating Costs Series.
Your schedule, your food order and your cash plan all start from one number: what you’ll sell next week. Most restaurants guess it. A simple forecast you build every week in 20 minutes beats the guess, and it’s the number your labor budget should come from.
The method
- Start with the same week last year: match by weekday, not date. The second Friday of March lines up with the second Friday of March.
- Apply your trend: compare your last 4 to 8 weeks with the same weeks last year. If you’re up 4%, multiply by 1.04.
- Adjust for the calendar: holidays that moved, a local event, a road closure, weather you can already see coming.
- Split by daypart: lunch and dinner staff differently, so forecast them separately.
A worked example
| Step | Amount |
|---|---|
| Same week last year | $40,000 |
| Trend over the last 8 weeks, up 4% | $41,600 |
| Home game on Saturday, plus 2% | $42,432 |
| Lunch, 35% | $14,851 |
| Dinner, 55% | $23,338 |
| Late night, 10% | $4,243 |
Turn sales into labor hours
Divide the forecast by your target sales per labor hour. If your target is $85, $42,432 divided by $85 is about 499 hours for the week. That’s your scheduling budget before anyone writes a shift. See sales per labor hour for how to set the target.
Do the same by daypart. Dinner at $23,338 gets about 275 hours. If the manager’s draft schedule has 320, you know before the week starts that dinner labor will run high.
Check it every Monday
- Write the forecast down before the week starts.
- On Monday, compare it to what you actually sold, by daypart.
- Aim to land within 5% for the week. If you miss by more two weeks running, your trend number is off. Update it.
- Keep a short list of events that moved sales. Next year’s forecast starts there.
A worked example: the Monday check
Here’s the week from the example above, forecast against what actually happened.
| Daypart | Forecast | Actual | Miss |
|---|---|---|---|
| Lunch | $14,851 | $13,900 | minus 6.4% |
| Dinner | $23,338 | $24,100 | plus 3.3% |
| Late night | $4,243 | $3,900 | minus 8.1% |
| Week | $42,432 | $41,900 | minus 1.3% |
The week landed within 2%, so the trend number is fine. Lunch is the story. At $85 sales per labor hour, the lunch forecast justified about 175 hours. Actual lunch sales only supported about 164. That’s 11 hours you paid for and didn’t need.
One week doesn’t make a trend. If lunch misses low again next Monday, cut the lunch forecast and the lunch schedule with it.
No history yet? Average your last four of the same weekday. Fridays of $6,200, $5,800, $6,400 and $6,000 give you a $6,100 Friday forecast.
The weekly forecast feeds the 13-week cash flow forecast and the labor cost template. If you want it all in one file, use the 13-week cash flow model.
Free: The Restaurant Financial Health Checklist. 30 monthly checks across sales mix, prime cost, 4-wall EBITDA, cash and operations, with the target ranges I use. Written by a former chain CFO. Instant PDF.
Unsubscribe anytime.
More on this: setting a weekly labor target.
Build the schedule from the forecast.
The Toolkit’s Labor Scheduling and Cost Model turns forecast sales into target hours and labor dollars by day, for up to 15 positions.
Get the Toolkit, $67FAQ
How do restaurants forecast sales?
Take the same week last year, multiply by your recent trend, adjust for holidays and events, and split it by daypart.
How accurate should a restaurant sales forecast be?
Within about 5% for the week is a good target. Daypart misses will be bigger.
How far ahead should I forecast?
Weekly for scheduling and ordering, 13 weeks for cash.
What if I don’t have last year’s data?
Use the average of your last 4 weeks for the same weekday and adjust for known events. Switch to last year’s numbers once you have them.