The cost of a server's extra walk: table-service labor economics

A server on a busy shift walks more than most desk workers walk in three days. The industry that employs them replaces most of its front-of-house workforce every single year, at a cost nobody agrees on precisely, while nearly two-thirds of operators say they can't find or keep the people they need. This page assembles what's actually measured about server labor — the physical load, the wage, the turnover, and the cost of it — and where the numbers are more solid than others.

How physically demanding is a serving shift, really?

Answer: extremely, in the small number of studies that have actually measured it directly — and the honest gap is that nobody has measured it at scale.

A 2019 accelerometer study of young adult servers (published in the International Journal of Nutrition, n=7) found male servers accumulated 13,589 steps on a working day and female servers 13,141 — more than double their step counts on days off (about 6,200 for both). A separate, CDC-archived ergonomics study of 20 servers across three restaurants found they spent 76% of their shift standing or walking, carried an average of 16.4 kg (about 36 lbs) per hour in trays, and reported the most end-of-shift discomfort in the upper back (55%), lower back (50%), and neck (45%).

Both studies are real, peer-reviewed measurements — and both are small. Seven people in one, twenty in the other. The widely circulated online claims of "servers walk 5 to 8 miles a shift" are plausible and consistent with the measured step counts, but we could not find a large-sample study behind that specific range; treat it as a reasonable estimate, not a citable statistic.

What does a server's time actually cost?

$16.94 an hour, median, before tips — the U.S. Bureau of Labor Statistics' May 2025 wage data for waiters and waitresses, the most current government figure available. The lowest 10% earned under $8.70; the highest 10% earned over $31.12, reflecting how much tipped income varies by market and venue type.

The more telling BLS number sits next to it: about 423,100 job openings for waiters and waitresses are projected every year, on average, over the coming decade — against a projected employment growth rate of just 2%. Almost none of that hiring volume is growth. It's replacement.

Why so many openings for so little growth? Turnover.

Restaurant industry turnover has run consistently in the 70-80% range annually for over a decade — Toast's own compilation of National Restaurant Association-attributed figures puts the 2013–2019 pre-pandemic average at 71.6%, rising to 79.6% as of 2024. Read plainly: in a typical year, a restaurant replaces somewhere between seven and eight of every ten positions it has. This is a structural feature of the industry, not a bad year at any one restaurant — the workforce skews young, part-time, and often transitional, and the front-of-house and back-of-house hourly roles turn over fastest of all.

What does replacing a server actually cost?

Here the numbers genuinely disagree, and it's worth knowing why.

The figure you'll see everywhere — $5,864 per employee — traces back to hospitality research from Cornell's School of Hotel Administration: Timothy Hinkin and J. Bruce Tracey's foundational 2000 study, The Cost of Turnover: Putting a Price on the Learning Curve (Cornell Hotel and Restaurant Administration Quarterly), which gathered detailed turnover-cost data from 33 U.S. hotels — not restaurants — and later extended in a 2006 Cornell Center for Hospitality Research web-tool report. Their underlying data ranged from $2,604 to $14,019 per position depending on job complexity, with $5,864 as the average. That figure has since been repeated across the restaurant-industry press for two decades as though it were a restaurant-specific number. It is a real, rigorously gathered figure — about hotel staff.

A more restaurant-native, if less rigorously documented, estimate circulates alongside it: industry sources commonly put the cost of replacing an hourly restaurant employee at $2,000 to $5,000, factoring recruiting, onboarding, and training time — lower than Cornell's hotel average, in the same order of magnitude. We could not find a published study that measured restaurant turnover cost with Cornell's level of methodological detail. Until one exists, the honest summary is: replacing a front-of-house employee costs somewhere in the low thousands of dollars, every time, and the industry does this to most of its workforce every year.

Is front-of-house actually understaffed right now?

Yes, by the industry's own most recent measurement. A National Restaurant Association and Workday report released in April 2026 found 62% of restaurants said recruiting and retaining staff was a very or fairly significant challenge, and warned that understaffing is a measurable drag on growth, service, and sales — being short just one employee, the report noted, can cost a restaurant thousands of dollars in annual sales. Older NRA-attributed figures (70% of operators reporting hard-to-fill openings, 45% saying they lack the staff to meet demand) circulate widely in 2026 coverage but we could not confirm a matching 2026 survey wave behind them; they may be recycled from an earlier NRA report. Where we can confirm the year, we cite the year.

So what actually reduces the load — hiring, or fewer unnecessary trips?

Put the pieces together and hiring your way out doesn't look like the available lever: the industry is already trying to hire at scale (423,100 openings a year for one role alone) against a workforce that leaves at 70-80% annually, and every replacement costs real money whether you trust the $2,000 estimate or the $5,864 one. The variable actually within a restaurant's control on a given shift is not how many people it has — it's how many trips each of them has to make.

That's the "extra walk" in this page's title: the round trip to ask what a guest needs, and the second round trip to bring it. The two studies above already show servers near or above 13,000 steps and 76% of a shift on their feet before accounting for a single avoidable trip — every question a server has to physically walk over to answer, and every request that takes two trips instead of one because the first trip only gathered the ask, is load stacked on an already fully loaded job, on top of a role the industry is struggling to keep staffed.

Where our own product fits, for what it's worth: Brief First's paid tier lets a guest attach what they need to the request itself — "more water, extra napkins, the check" — so the server's one walk to the table already carries the answer, instead of a first trip to ask and a second to fetch. The free tier's menu host answers a guest's question about a dish without a server being flagged down at all. Neither changes a server's wage or the industry's turnover rate; both are aimed narrowly at the one variable this page's research says is actually addressable shift to shift — the number of physically avoidable trips a server makes.