The real cost of server turnover, worked out line by line

When a server quits, you pay for it several times: in shifts other people cover, in the hours spent hiring, in paid training shifts, and in a few slow, mistake-prone weeks while the new person learns. In our worked example below, that adds up to about $1,070 per departure for a 14-server restaurant, or about $7,500 a year at seven departures. Training time is the biggest line. Your figure will differ, so every assumption is stated and there’s a blank column for yours.

How often do restaurant workers quit?

Short answer: about twice as often as workers in the rest of the private sector, though less often than a few years ago.

The US Bureau of Labor Statistics measures quits through its JOLTS survey. In accommodation and food services, the quits rate averaged 4.2% a month in 2025, against 2.2% for all private industry. It was 5.8% in 2021 and 2022, 5.0% in 2023 and 4.1% in 2024 (BLS JOLTS, Table 22). In 2025 the sector recorded 7.24 million quits, about 20% of the 35.8 million quits across all private industry (BLS JOLTS, Table 21). The latest month, August 2026 (preliminary, seasonally adjusted), shows 3.5% for the sector against 2.1% for all private industry (BLS JOLTS, Table 4).

Three cautions before you use those numbers:

• They cover the whole sector, from hotels to fast food, not servers alone. BLS doesn’t publish a quit rate for servers.
• They count quits only, not dismissals or layoffs.
• The trend is down. The National Restaurant Association reported that turnover “fell to an 11-year low for restaurants and accommodations in 2024” (NRA, Mar 2025), yet “77% of operators say recruiting and retaining employees is still a leading challenge” (NRA, 2025).

Turnover is built into the job. BLS projects “about 423,100 openings for waiters and waitresses” each year on average, mostly to replace workers who leave, against 2,284,400 jobs in 2025 (BLS, 2026). And the workforce is young: 52% of waitstaff are under 25 (NRA, Mar 2026).

What does one departure cost? A worked example at Casa Olivo

Short answer: in this example, about $1,070, and about half of it is training.

Casa Olivo is a fictional US restaurant (an example, not a customer). Every number below that isn’t from BLS is an assumption, marked as one. Replace them with yours.

The assumptions.

• Casa Olivo has 14 servers.
• Server time is valued at $16.94 an hour, the US median wage for waiters and waitresses in May 2025 (BLS). BLS notes the median includes tips, so the cash Casa Olivo pays per hour may be lower. We use it as the value of a server’s hour.
• Manager time is valued at $25 an hour. (Assumption, not a BLS figure.)
• After a server gives notice, it takes two weeks to hire and start someone. Other servers cover 10 extra hours at an overtime premium of $8.47 an hour, half of $16.94. (Assumption.)
• A new server works 4 paid shadow shifts of 6 hours before working alone. (Assumption.)

The worksheet.

Training (lines 7 and 8) comes to $508.20, just under half the total. Hiring (lines 3 to 6) is $390.

How do you count your own departures?

Short answer: from payroll. Count everyone who left in the last 12 months and divide by your average number of servers.

The sector rate is a guide; your own number is better. Pull every server who left in the last 12 months from payroll, for any reason, and divide by the average number of servers on your books. At Casa Olivo, 7 departures ÷ 14 servers = 50%.

Then look at when people left. A server who quits in week three costs nearly the full $1,067.90: you paid for the ad, the interviews and all four shadow shifts, and got a few shifts of work back. If two of Casa Olivo’s seven leavers go within their first month, about $2,100 a year buys almost nothing. Look at those early leavers first: they may point at how you hire or run the first week, which can be cheaper to change than pay.

Staffing has eased but not disappeared. About a third of operators (32%) said they “don’t have enough employees to support existing customer demand”, per National Restaurant Association research reported by FE&S, down from earlier years (FE&S, 28 Apr 2026). If you’re in that third, every departure also means a stretch of short-staffed nights, which the worksheet prices only as overtime.

What does it cost in a year?

Short answer: multiply the cost per departure by how many people leave. At Casa Olivo that’s about $7,500.

How many will leave? The sector’s 4.2% monthly quits rate, multiplied by 12, is 50.4%, or roughly half the workforce in a year. That’s a rough guide for the reasons above, so check it against your own records. For Casa Olivo: 14 servers × 50% = 7 departures.

7 × $1,067.90 = $7,475.30 a year.

Here’s how the total moves when the assumptions change:

The last row shows why training is the line to watch. Four more shadow shifts add 24 hours × $16.94 = $406.56 to every hire. Cutting training to save money can be a false saving, though: a new server sent out too early makes more of the mistakes in line 9, and may leave sooner.

What does the worksheet leave out?

Short answer: the costs that are real but hard to price, so treat the total as a floor.

• Covers you can’t serve. On a night when you’re two servers short, you might close a section or slow the waitlist.
• Service quality while stretched. Longer waits to order and to pay, and tired staff.
• What leaves with the person. Regulars’ names, the wine list in their head, a second language that made them the person you sent to certain tables.
• The manager’s attention. Hiring hours come out of the same week as ordering, scheduling and the floor.
• The next departure. A team that keeps covering for empty shifts gets tired, and tired teams lose people. That knock-on effect is real but impossible to price honestly, so we don’t.

If you want to put a number on lost covers, count the tables you turned away or the section you closed on short-staffed nights over a month, and multiply by your average check per table.

How do you use the number?

Short answer: as the yardstick for any retention idea. A step pays for itself if it prevents enough departures.

At about $1,070 per departure, the break-even is simple: divide the yearly cost of a retention step by $1,070.

Those costs are Casa Olivo’s own estimates, not benchmarks. The point is the method: price the step, divide by your cost per departure, and ask whether it’s plausible that it keeps that many people.

Where does WaiterPilot fit?

Short answer: we don’t claim it reduces turnover. It’s a tool, and any new tool adds to training.

We have no data on WaiterPilot and turnover, and we won’t quote a number we can’t support. What it does: guests order and ask questions at the table from their phone, and the chat answers from your menu; staff confirm orders on a board, and staff make and serve the food. It doesn’t replace servers.

If you add WaiterPilot, or any ordering tool, add its training to line 7 for every new hire. Our setup includes staff training at no charge and a printable staff guide, but the hours your servers spend learning it are still hours. Ask every vendor the same question: how long does it take a new server to learn this on their first shift?

To compare tools by cost rather than staffing, our cost calculator works out flat fees against commission, and flat fee vs commission does it for three restaurants. WaiterPilot is $200 per location per month after 30 days free, with 0% commission; see pricing.

What to do next

1. Fill in the “Yours” column, using your real wage, manager rate and shadow shifts.
2. Count last year’s server departures from payroll records.
3. Multiply, and write the yearly figure where you’ll see it when you plan the next year.
4. Price one retention step and work out its break-even.

For jobs you can take off servers on busy nights, read short-staffed tonight. For the wider staffing picture, see our labor shortage data for 2026. And if you do bring in a new ordering board, plan the training with how to train staff on a new ordering board.