Restaurant sales are down for one of two reasons: fewer guests came in, or the same guests spent less on each order. If your guest count held and sales fell 33%, every order got a third smaller. In late September 2026 that is mostly the season and fuel prices, not your café. Judge the full week against the same week last year, and act only when the week is down too.
Why are my restaurant sales down
Sales are two numbers multiplied: orders times average ticket. Average ticket is what one order spends on average, so total sales divided by the number of orders. Split every bad day into those two numbers before you do anything else.
Here is a normal Thursday and a bad Thursday at one busy café, in US dollars, in 2026:
| Normal Thursday | Bad Thursday | |
|---|---|---|
| Orders | 200 | 200 |
| Average ticket | $12.00 | $8.00 |
| Sales | $2,400 | $1,600 |
| Change | −33% |
The guests did not leave. Many bought only a $5 latte and skipped the $7 sandwich or pastry. Same guests with smaller orders is a ticket problem, not a traffic problem, and the fixes are different. If the number of orders had dropped instead, you would need new guests. Then start with ways to fill empty tables on slow weekdays.
Judge the week, not the day
One day has too many moving parts: rain, a school event, a closed road, the day before payday. A single day can fall 30% and mean nothing at all.
Close your week from Monday to Sunday and read it by Tuesday morning, so you can still change next week. Compare it with the same week last year, not with last week, because last week sits in a different season.
Carry the bad Thursday through its week. A normal week at this café brings $16,000. The bad Thursday took $800 off, so the week closed at $15,200, down 5%. The same week last year brought $15,500, so against last year you are down about 2%. That is a normal slow week.
Use this rule of thumb before you worry:
- One bad day: write down what was different that day and move on.
- Two weeks in a row down against last year: look at orders and average ticket for each part of the day, morning, lunch and afternoon.
- Four weeks down more than 10% against last year: change the ticket, the labor and the hours, as below.
Track five numbers every week: sales, orders, average ticket, food cost and labor cost. The full Monday list is in restaurant KPIs to check every Monday.
Late September is slow for most cafés
Many owners with long careers see the third week of September as the second-slowest week of the year, after the third week of January. Treat that as a rule of thumb, not a statistic. The usual reason is money: back-to-school costs reach the credit card bill in September, and holiday costs reach it in January. Late July and early August come next, when families are away on vacation.
Check your own sales for the last two or three years and mark those weeks on next year's calendar. Then plan around them:
- Don't book a paid band, launch a new menu or start a promotion in the third week of September or January.
- Give staff their vacation days in those weeks.
- Use the dead hours for jobs a busy day never allows: re-cost your 10 best sellers, ask a second supplier for prices, fix the broken shelf, and train a new barista on one add-on.
Guests are spending less in 2026
The season is not the only thing going on. In August 2026, 46% of US restaurant operators reported fewer guests than a year before, according to the National Restaurant Association. It was the 18th time in 19 months that more operators lost guests than gained them. In the same survey, 51% still reported higher sales than a year before, so people are still eating out.
Fuel is a big part of it. In September 2026 the weekly US diesel average passed $6 a gallon for the first time, according to the Energy Information Administration. It reached $6.285 in the week of September 14. Regular gas was about $4.47 a gallon at the end of the month. Your guests pay more to drive to you, so many keep the visit and cut the extras.
Diesel also moves every truck that brings your milk, flour and cups. Ask each supplier if they add a fuel surcharge, an extra fee for fuel, on every delivery. If one adds $15 a delivery and you take three a week, moving to two a week saves $780 a year.
Raise a smaller ticket without discounts
When guests trade down, meaning they pick cheaper things, a lower price will not bring the lost $4 back. Make it easy to add one small item instead.
- Put a coffee and pastry pair on the board at $7.50. Bought apart, they cost $8.75. A guest who came for a $5 latte now spends $2.50 more, and the pastry costs you about $1.10, so you keep $1.40 more.
- Sell a half sandwich for $6. A guest who won't pay $11 for lunch often pays $6.
- Ask one clear question at the till. "Want a cookie with that?" sells more than "Anything else?" because it names a thing.
- Keep one $2.50 item next to the register that costs you less than $0.70.
Do the maths on the bad Thursday. If one guest in four adds the $2.50 pastry, that is 50 × $2.50 = $125 more that day. Over 30 days it is $3,750, from the same guests.
Hold price rises until the slow weeks pass. A guest who already counts every dollar notices them first. More ideas are in how to increase your average ticket.
Cut labor in the dead hours
A slow day does not lower your labor bill, so labor takes a bigger share of sales. Three people on 8-hour shifts at $20 an hour cost $480. That is 20% of a $2,400 day and 30% of a $1,600 day.
Look at sales by hour, not by day. Sales per labor hour is your sales divided by the hours your staff worked. On the normal Thursday it is $2,400 ÷ 24 hours = $100. On the bad Thursday it is $67.
If the 2 p.m. to 5 p.m. stretch sells $150 with three people on, end one shift at 2 p.m. in the slow weeks. Cutting those 4 hours saves $80 and lifts sales per labor hour to $80. Put the shorter shift in the schedule two weeks ahead, so nobody drives in only to be sent home.
US rules before you send staff home early (2026)
Some states make you pay for a minimum number of hours when you send a worker home early. This is called reporting time pay. California, New York, New Jersey and Massachusetts are among them, and each state sets its own minimum.
Some cities, among them New York City, Chicago and Seattle, have fair workweek laws. They make employers pay extra when a shift changes at short notice. Most of them cover only large chains, but check before you change a schedule late. Rules change, so confirm with your state labor department.
When to close a location
Read each location's own profit and loss, not the total for all of them. A strong store hides a weak one. Say one café makes $9,000 a month and another loses $2,000 a month. After six months, the strong one has paid $12,000 for the weak one.
Compare the weak café with its own same months last year. If it loses money in months that made money last year, the season is not the reason. Give it one change and 90 days: shorter hours, a smaller menu, or one less person on weekdays. Write the target down before you start, for example "break even by month three". If it misses, close it or sell it.
Before you decide, read your lease with a lawyer. Look for a break clause, which lets you end the lease early. Also look for the right to sublet or hand the lease to a new tenant.
Count your own hours too. A $100,000 job is about $48 an hour for a 40-hour week. If you spend 15 hours a week fixing the losing café, those hours cost you about $720 a week.
Do this week
- Close last week from Monday to Sunday for each location and compare it with the same week in 2025.
- Split each location's sales into orders and average ticket.
- Put one coffee and pastry pair on the board and one named add-on question at the till.
- Mark the third weeks of January and September, and late July, as slow weeks in next year's schedule.
- Write a 90-day target for any location that lost money for six months in a row.