Restaurant sales down? What to check before you cut

Before you cut anything, find out which number fell: how many guests came, or what each one spent. Check it by day, daypart and channel against the same weeks last year, then see how far your sales sit above break-even. The first answer tells you where to act and the second how fast: cut the hours guests no longer fill, and keep what brings them in.

First, find which number fell

Sales are your guest count times your average check. So a sales drop means fewer guests, a smaller check, or both, and each needs a different fix. Your POS has both numbers.

Over the 8 weeks, sales are down 5.5% but guests are down 10%: the higher check hid a guest problem nearly twice the size. And lunch lost 340 of the 480 missing guests, while dinner barely moved.

Check 1: guests or average check?

Whichever fell more is your main problem. Fewer guests means finding out who stopped coming, and when. A smaller check means guests order less, or discounts and comps grew. If you raised prices in the past year, the check should have risen about as much; if it didn't, look at drinks and desserts per guest in your product mix report. Compare checks within each daypart too: when a cheaper daypart loses guests, the overall average rises with nobody spending more.

Check 2: which days, dayparts and channels?

Split the missing guests by day, daypart and channel. If one accounts for more than half of them, start there, not across the whole week: in the example, lunch. If the loss is spread across every shift, look at what touches all of them, such as prices, ratings or a new competitor. If dine-in fell while app orders grew, some guests may have moved channel rather than left, and each app order keeps less of the sale.

Check 3: the same weeks, with the calendar lined up

Compare weeks, not months, going back 364 days so every Friday lines up with a Friday. Months don't line up: August 2026 had four Fridays and five Mondays, August 2025 five Fridays and four Mondays. Holidays move too. Labor Day fell on September 1 in 2025 and September 7 in 2026, so on a 364-day comparison each one lines up with an ordinary Monday. Judge the 8 weeks as a whole, or compare the two holiday weeks with each other.

The rule: if the drop disappears once the weeks line up, there's nothing to cut. If one or two weeks explain it all, find out what happened in them first. Otherwise, 5 or more of the 8 weeks down is a trend; with 3 or 4, watch the next 2 weeks before you cut.

Check 4: did costs move at the same time?

Put food cost % and labor % for the 8 weeks next to last year's, from your invoices, payroll and POS. Labor % up while guests fell usually means the schedule didn't follow them, unless pay rates rose. A higher food cost % can mean supplier prices rose, or that you're still ordering and prepping for last year's covers: check invoice prices, then cut pars and orders to this year's counts. Neither explains why guests left, but both are fixable this week.

How deep is the hole? Check your break-even

Some costs leave with the sale: food, drink, card fees, packaging. Most stay: rent, salaries, insurance, loans and shifts scheduled the same way every week. Break-even is the sales a month needs to pay them all, and your distance above it is your room to work.

Sales fell 5.5%; what's left over after every bill fell about 41%. Each $100 of lost sales took $60 with it, because only $40 of costs left with the sale. Now compare your cushion, sales minus break-even, with the monthly drop:

• Below break-even: every week costs cash. Match hours to your guests this week.
• Above it by less than the drop: one more drop like it puts you under. Act on costs this week.
• Above it by more than the drop: you have time to find the cause. Don't wait for next month's P&L.

The example's cushion is $8,280 against a $5,720 drop: room for one more drop like it, not two.

What to cut, and what to keep

Flex labor to the guests you actually have

Shifts written for last year's guest counts don't shrink on their own. In the example, $33,000 a month of salaries and hourly pay went from 31.7% of sales to 33.6%, with no raises. The rule: if labor % is up on the same weeks last year while guests are down, rewrite next week's schedule from forecast covers, starting with the daypart that lost the most guests, and use a sales-per-labor-hour cut line on the shift. In the example, each $1,000 a month of hours it doesn't need lowers break-even by about $1,667 in monthly sales ($1,000 ÷ 0.60).

Rules on schedule changes and on pay for shifts cut short differ by state and city. If you're not sure which apply to you, ask your state labor office, or your city's labor office where your city has its own scheduling rules, before you change shifts.

Don't cut what brings guests in

Guests notice portions, quality, speed at the rush and a clean room. Cut those on the shifts that held up and a lunch problem can become a dinner problem. The rule: if a regular would see or taste the cut on a busy shift, don't make it. Cut idle hours, prep for covers that don't come and spending no guest sees. If your star rating slipped while guests fell, fix what the reviews name first.

Fix the leaks that cost money at any level of sales

Some leaks have nothing to do with the drop, such as dishes priced off old invoices or app orders at menu prices. Fixing them raises the share of each sale you keep (60 cents in the example), which lowers break-even too. Start with the six usual leaks: they cost money when you're slow, not just when you're busy.

Is it just you? What the national numbers say

In the National Restaurant Association's monthly survey of operators, published September 30, 2026, 51% said same-store sales in August 2026 were higher than in August 2025, and 31% said lower. Traffic was weaker: 43% said customer traffic rose over the same 12 months and 46% said it fell. August was the 18th month in the last 19 in which more operators reported lower traffic than higher.

More operators saw fewer guests (46%) than saw lower sales (31%): in some restaurants, as in the example, rising checks are covering for fewer guests. Some guests are trimming extras too. In the Association's third-quarter consumer survey, published September 17, 2026, 34% of consumers said they were ordering fewer add-ons, such as desserts and drinks. That's the backdrop, not your diagnosis. A national number can't tell you why your guests left; your own reports can.

This week

1. 01Pull the last 8 weeks and the same 8 weeks last year (364 days earlier) from your POS: net sales, guests and average check, by daypart and channel. Do it before you write next week's schedule.
2. 02Enter last month's sales and costs in the break-even calculator. If your cushion is smaller than the monthly drop, act on costs this week.
3. 03Write next week's schedule to forecast covers, starting with the daypart that lost the most guests. Next Monday, compare labor % with the same week last year.