← Back to blog

The Cost of a Bad Hire, Priced in Dollars

SHRM says a bad hire costs 30 percent of first-year salary. In restaurants the number is wider. A line cook lands near 15 percent. A general manager lands closer to 90. Here is the line-item math, and the one case that cost my group $78K in seven months.

Kiyong (Eric) Lee·Operations Executive·August 2026·10 min read

The short version

SHRM benchmarks a bad hire at 30 percent of first-year salary. Restaurants sit above and below that band depending on the seat.
Bad line cook hire on a $42K role: $6,500 total, roughly 15 percent.
Bad general manager hire on an $85K role: $65K to $95K total, roughly 75 to 110 percent.
I once pushed hiring on a Zareen's GM slot to close a gap. It cost the group about $78K in seven months.
What actually reduces it: paid working interview, reference triangulation across two supervisors and one peer, and 30-60-90 milestone accountability signed before day one.

The Society for Human Resource Management publishes a number every operator has heard quoted at least once. A bad hire costs about 30 percent of the first year's salary. It is a useful anchor. It is also almost never right for restaurants, and the direction of the error changes depending on which seat you are filling.

On line and hourly roles, the actual cost usually lands below 30 percent because the ramp is short and the wage is low. On management roles, the cost usually lands well above 30 percent because a weak general manager takes a whole unit down with them for every week they sit in the chair. I have paid for both. I want to walk through the real line-item math, with the numbers I use when I brief owners on what a hiring decision actually costs.

The line cook: $6,500 on a $42K role

Start with a role most operators think of as low-cost to fill and low-cost to lose. A line cook on a $42K base. Six weeks in, it is clear this hire is not going to make it. The station is not clean, the tickets are stacking, and the shift lead is spending 40 percent of the rush covering the section. You separate at week eight. Here is what the bad hire actually cost the P&L:

Recruit and posting cost · $400. Job board fees, referral bonus paid to the person who brought them in, and the two hours of the sous chef's time on interviews.
Onboarding and initial training · $1,200. Payroll for the first two weeks while the person was shadowing at 40 percent output, plus the trainer time.
Drag on shift for six weeks · $2,800. The section covered by the shift lead, the slower tickets, the extra prep the next day because prep was rushed, the incremental waste from mistakes on the station.
Guest experience damage · $1,500. Two comps and one refund per week attributable to this station, plus the estimated repeat-visit loss on tables who got a plate below standard.
Separation and re-recruit · $600. Exit process, any unemployment claim exposure, and the cycle of running the job posting again.

Total: about $6,500. On a $42K role, that is 15 percent. Below the SHRM number, but real money, and the number goes up by roughly 2 to 5 percent per additional pay period you keep the person past the point where the answer is clear.

Fig. 1 · The stacks tell the story. On the GM role, the drag and the guest damage dominate.

The general manager: $65K to $95K on an $85K role

The math on a management seat is different because the drag line explodes. A weak general manager does not just fail at their own job. They pull the unit's operating rhythm with them. Here is what a bad GM hire on an $85K role costs in the seven-month range, which is roughly how long it usually takes to fully diagnose and separate:

Recruit and search cost · $6,000. Job posting, agency fee at the low end for a mid-size unit, plus the executive time on interviews and reference calls.
Onboarding and training · $8,500. First 60 days of salary while the person ramps, plus the area director's time and the shadowing hours across two other GMs.
Drag on unit for six months · $34,000. One to two points of prime cost drift on a $3M unit is $30K to $60K by itself. Schedule sloppiness, food cost creep, comp discipline slipping, and the extra visits the area director now has to make to hold the numbers up.
Guest experience damage · $18,000. Turnover on the floor and in the kitchen because good people do not work for a weak leader. Two or three key departures at $6K to $8K replacement cost each, plus the review scores drifting down by three tenths of a star.
Separation and re-recruit · $11,500. Severance if you offer it, unemployment claim exposure, the six-week interim while the area director covers the seat, and the search you now run again.

Total: roughly $78,000 in the mid case, and the range across the concepts I have seen is $65K to $95K depending on unit volume, market, and how long the wrong person stays in the seat. On an $85K role that is 75 to 110 percent of first-year salary. The SHRM number understates it by a factor of two to three.

The case I keep coming back to

At Zareen's, our $30M Michelin-recognized group with a workforce of 215, we had a GM slot open at one of the growth locations. The pipeline was thin. We had two internal candidates who were not ready and one external candidate who looked good on paper but had two open questions on the reference check that we did not close. The unit needed a manager on the floor by the following week. I pushed the hire.

The person was pleasant and knew the front of house well. They could not run a P&L. Over seven months prime cost drifted 2.4 points, two shift leads I did not want to lose walked, and the catering channel out of that unit missed its number in three of the seven months. When we finally separated and moved an internal candidate over, the total damage totalled about $78K by my estimate. Every dollar of it was a decision I made because I did not want to be short a GM for another three weeks.

The most expensive hire is the one you make because you do not want to sit with an open seat. The open seat is uncomfortable. It is also cheaper than the wrong person filling it.

The lesson I took from that engagement, and the lesson I now brief every owner on, is that an open seat is a known cost. Overtime, area director cover, some guest experience drag. Call it $4K to $8K a week on a full-service unit. A wrong hire in that seat can cost $70K to $95K over seven months. On the arithmetic alone, you can afford to leave a GM seat open for eight to twelve weeks and still be ahead of a bad hire.

What actually reduces the cost

The playbook is not fancy. It is three practices, in order of impact, and it cuts bad-hire frequency roughly in half in every operation I have installed it in.

1. The paid working interview

Four to six hours, paid at the role's hourly rate, on a real service shift. For a line role that is a Friday dinner or a Saturday brunch. For a management role it is a full opening or closing shift with the current GM. You are not testing whether they can pass a demo. You are watching whether they can find their own rhythm under real load, whether they check in when they should, and whether they treat the staff the way you would want them treated when nobody is watching.

One working interview tells you more than three panel rounds ever will. It also tells the candidate more about the job than any offer letter, which is why the ones who accept after a working interview stick at a higher rate.

2. Reference triangulation

Skip the check-the-box reference call. Ask two supervisors and one peer the same three specific questions about actual shift behavior. What did the candidate do the last time a station went down mid-service? Give me an example of a coaching moment they handled with a struggling team member. What would you fix about how they run their calendar?

You are looking for consistency across the three answers, and for the small divergences that reveal the parts of the story the candidate edited. It takes about 40 minutes total. It catches roughly one in five hires that would otherwise fail.

3. 30-60-90 milestone accountability

Write down the specific outcomes the new hire must own at day 30, day 60, and day 90. Both sides sign it before the start date. Review it on the day, in a 30-minute one-on-one, not in a hallway.

The document does two jobs. It gives the new hire a clear target that helps a good person succeed, and it gives you an early honest read that helps a bad-fit situation resolve fast. If the person misses two of three milestones at day 30, start the exit conversation the same week. The cost of a bad hire is roughly linear in days on the payroll. Every additional pay period is another 2 to 5 percent on the total.

The signals I read at day 14

Waiting until day 30 to know if a hire is going to land is too long on a management seat. The earliest honest read comes at day 14, and there are three signals I trust.

Do they ask about the numbers on their own? A GM who is going to own their P&L asks to see the weekly by day 10. A GM who is going to hide from it does not. In my experience the ones who wait to be shown the P&L are the ones who will not read it in month three either.
Do the team members closest to them tell you a consistent story? Talk to two line cooks and one shift lead in the first two weeks. Ask what the new person is like on the floor when the pressure hits. If the three answers agree, you have signal. If they diverge, you have a different problem to run down.
Do they take one small stand? A good manager, in the first two weeks, will push back on one thing they think is wrong. A weak manager will agree with everything. The pushback is the shape of the person who will hold the standard when you are not there.

None of these three are unambiguous by themselves. Two out of three lining up the wrong way at day 14 has been reliable enough that I trust it as a leading indicator. It has been wrong on me maybe once out of every eight or nine reads.

The point

Hiring is the most expensive decision an operator makes on a weekly basis, and it is the one most operators run on the least discipline. The reason is that the cost of a bad hire is not on the P&L in a single line. It is spread across five or six lines, and it never gets aggregated back into a single number that would make the pattern visible.

Run the aggregation yourself. Line cook: $6,500. Shift lead: about $12K. Sous chef: about $22K. GM: $65K to $95K. Multi-unit director: north of $150K. Put those numbers on the wall in the office where hiring decisions get made. Then run the paid working interview, the reference triangulation, and the 30-60-90 milestone doc on every hire above line level. The numbers are the case for the discipline. The discipline is what actually saves the money.

Sitting with an open manager seat?

I work with restaurant operators on hiring structure and bench development, and I have opinions about the tradeoff between an open seat and a wrong seat. If you are looking at that decision this quarter, I am happy to trade notes.

Get in touch

Questions people ask

How much does a bad hire cost in a restaurant?

SHRM benchmarks the cost of a bad hire at 30 percent of first-year salary across industries. In restaurants the range is wider. A bad line cook hire on a $42K role runs about $6,500, or 15 percent. A bad general manager hire on an $85K role runs $65K to $95K, or 75 to 110 percent, because the drag on the whole unit compounds every week the wrong person is in the seat.

What line items make up the cost of a bad hire?

Recruit and posting cost, onboarding and training labor, the drag on shift productivity while the person ramps and then fails, guest experience damage during the ramp, separation cost including any severance and unemployment claim exposure, and the rehiring cycle you now have to run again. On a line role those five buckets add to about $6,500. On a GM role the guest experience and team damage lines dominate.

Why is a bad general manager hire so much more expensive than a bad line hire?

A general manager sets the operating rhythm for the whole unit. A weak GM drops labor discipline, food cost tightness, service standards, and staff retention for every shift they are in the seat. On a $2M to $4M unit, one point of prime cost drift over six months is $10K to $24K by itself. Add the two or three good team members who leave because the GM cannot lead them, and the total lands in the $65K to $95K range.

What actually reduces the cost of a bad hire?

Three practices, in order of impact. A structured paid working interview on a real shift, not a conversation in an office. Reference triangulation across two supervisors and one peer, asking the same three questions. And a 30-60-90 milestone document signed before the start date so both sides know what good looks like at day 30. These three cut bad-hire frequency roughly in half in every operation I have installed them in.

How long should a working interview be?

Four to six hours, paid at the role's hourly rate, on a real service shift. Not a slow Tuesday afternoon. A Friday dinner or a Saturday brunch, whichever is closer to the load the person will be under. You are not testing whether they can pass a demo. You are watching whether they can find their own rhythm under real pressure.

When should you cut a bad hire loose?

If they miss two of three milestones at day 30, start the conversation the same week. The cost of a bad hire is roughly linear in days on the payroll. Every additional pay period you delay adds two to five percent to the total cost on a line role, and roughly the same on a manager role because the compounding damage keeps compounding. Speed on the exit is the second most important thing after speed on the hire itself.

What is reference triangulation and how is it different from reference checks?

A reference check asks the reference the questions the candidate wants asked. Reference triangulation asks two supervisors and one peer the same three specific questions about actual shift behavior, then compares the answers. You are looking for consistency across the three, and for the small divergences that reveal the parts of the story the candidate edited. It takes about 40 minutes total and it catches roughly one in five hires that would otherwise fail.