Restaurant employee turnover: what the BLS rate means for your budget

Keep the rate, period and industry together

The Job Openings and Labor Turnover Survey, or JOLTS, measures changes to employers' payrolls. Its accommodation and food services category combines NAICS 721, accommodation, and NAICS 722, food services and drinking places. This is broader than independent restaurants, and it does not separate your kitchen crew from your managers. BLS industry scope

Here are the published annual average rates from the March 13, 2026 release. Each number is a percentage; all are not seasonally adjusted.

Sources: total separations, Table 20, quits, Table 22, and the dated release. Checked October 11, 2026. These are 2021–2025 observations, not a full-year 2026 estimate. BLS revises the data; a small difference between rounded values alone does not establish a statistically meaningful change.

Total separations include voluntary quits, employer-initiated layoffs and discharges, and other departures such as retirement or transfer to another location. Quits are already part of the total: do not add the two columns. Nor does every separation imply that an operator will replace the position. BLS definitions

Why 5.5% and roughly 66% can describe the same year

BLS calculates its annual average separations rate this way:

An annual turnover calculation using average monthly headcount has a denominator one-twelfth as large:

For the same underlying data, that second calculation is 12 times the first. Multiplying the published 2025 figure gives 5.5% × 12 = approximately 66%. This is a derived approximation from a rounded rate, not another percentage published in Table 20. The denominator rule comes from the BLS technical note.

It does not mean 66 of every 100 people employed at the beginning of the year left. These are payroll departure events over time. Several successive occupants of one job can leave during a year. Calculating retention of the starting team requires following those specific people separately.

This conversion uses the annual average's definition. Multiplying one unusually busy month's rate by 12 instead would be a scenario that assumes the month repeats, not an observed annual result. JOLTS treats hires and separations as flows across a period. BLS FAQ

Calculate your own rate before comparing it

Suppose a hypothetical café records 18 separation events during a full year. Twelve are quits, four are layoffs or discharges, and two are other separations. Its 12 consistently measured monthly payroll headcounts sum to 240, so average headcount is 20.

The 90% and 7.5% figures describe the same hypothetical departures. Comparing 90% directly with the published 5.5% mixes denominators. Even the aligned 7.5% is not a verdict on the café: its size, roles and operating pattern differ from the broad national sector.

Use a consistent monthly payroll snapshot, include full-time and part-time employees, and record departure dates and reasons. Headcount is not full-time-equivalent staffing or paid hours. For a closer BLS comparison, use its employment reference: the payroll period including the 12th of the month. A different consistent snapshot can support your internal trend, but disclose the method. JOLTS employment definition

Keep your own employee records private. Review patterns by role, tenure and stated reason before deciding what needs to change; an aggregate percentage does not explain why someone left.

Turn a hiring plan into a cost estimate

Consider a separate, hypothetical quarter. The operator starts with 20 employees and plans to finish with 22. They assume four departures during that quarter based on their own planning information. That means six planned hires: four replacements and two additional positions. This is not a forecast derived from the national rate.

For each hire, assume the following work is additional to the ordinary wages already in the quarter's budget:

For six hires, the extra gross wages are $420 × 6 = $2,520. With an assumed employer burden of 18% on those wages, the added labor is $2,973.60. Recruiting fees are another $240, for a combined $3,213.60 under these assumptions.

These are teaching inputs, not wage recommendations or an industry cost-per-hire benchmark. Replace the hours, rates, burden and fees with your records. If a salaried manager handles interviews within unchanged paid hours, their time may displace other work but does not automatically create an extra payroll payment. Likewise, training already included in the wage budget must not be counted twice. Lost sales, service disruption and waste remain unpriced here; do not invent them to inflate the total.

See what the paid training changes in the labor budget

Assume the quarter's baseline includes $50,000 of gross wages for all planned ordinary work, including the two additional positions, and $200,000 of sales excluding sales tax and tips. The 18% burden is still a hypothetical assumption.

The increase is about 1.49 percentage points, calculated before rounding. Keep the $240 recruiting fees separately identified; this labor calculation applies employer burden to wages, not to those fees. Neither row predicts profit or proves that a retention initiative will save money.

Use the free labor cost calculator to try wages of 50,000, burden of 18%, and sales of 200,000; then change only wages to 52,520. The tool displays percentages to one decimal place, so the second result reads 31.0%. Replace the example's inputs and enter your actual same-period covers if you want labor per cover; its prefilled cover count is not part of this example.

The labor-cost guide explains the period calculation in more detail. Marji is restaurant food cost software, not a payroll, recruiting or employee-retention system. This free calculation can help examine your labor budget; it does not track departures or determine which staffing decision you should make.