Someone resigns on a Monday. On a ten-person team, you feel it by Wednesday. There is a shift nobody wants to cover, an account with no owner, and the training you thought you had finished in the spring. Then payroll drops by one salary and, on paper, the month looks slightly cheaper.
That is the trap. The cost of losing that person is real, but almost none of it arrives as an invoice, so it never lands anywhere you would think to look.
Most owners I speak to have never worked the number out, and it is rarely carelessness. Nearly every published figure on turnover is either a percentage of salary, which explains nothing, or an average with no arithmetic underneath it.
The sum itself is not hard. Getting a straight answer out of it is, and that turns out to be the more interesting problem.
The four things you actually pay for
Recruitment is the obvious one. Advertising, screening, interview time, reference and background checks, and an agency fee if you use one.
Training, you probably half-count already. Trainer hours, materials, classroom or shadowing time, and the wage of somebody who is learning rather than producing.
Equipment gets forgotten because it feels like a one-off. Laptop, phone, licenses, a desk, provisioning, and the admin wrapped around all of it.
Then there is ramp, and ramp is the one that never shows up in your accounts at all. A new hire costs a full wage from day one and produces a fraction of full output. Somewhere between week one and week eight, they close that gap, and you pay for every hour in between. Nobody sends you a bill for it. That is precisely why it goes uncounted.
The formula
annual cost = turnover rate
x (recruitment + training + equipment + ramp drag)
x team size
ramp drag = loaded hourly wage x ramp weeks x 40 hours x 0.7
The 0.7 is an assumption: a new hire runs at roughly 30 percent of full productivity across the ramp period, so about 70 percent of what you paid for is lost. Move it if your onboarding is quicker than that.
The numbers
Ten seats. Thirty-five percent annual turnover. Eighteen dollars an hour loaded, meaning employer cost, not take-home pay. Recruitment $1,244, training $1,500, equipment $750, ramp six weeks.
Direct costs: 1,244 + 1,500 + 750 = $3,494.
Ramp drag: 18 x 6 x 40 x 0.7 = $3,024.
One departure: 3,494 + 3,024 = $6,518.
Departures in a year: 0.35 x 10 = 3.5.
Total: 3.5 x 6,518 = $22,813.
Twenty-two thousand dollars on a team of ten, and not one line of it sits in your accounts under a heading you would recognize. Notice also that the ramp is $3,024 of that $6,518, about 46 percent. It is the largest single component and the only one with no paper trail.
The part that should make you suspicious
That $1,244 is not my figure. It is the median cost per hire in the 2022 Talent Access Report from SHRM, the Society for Human Resource Management, across 472 organizations.
The mean in the same report is $4,683.
Same survey, same respondents, same year, and a gap of nearly four times. That gap is skewed. A small number of very expensive hires, executives, and hard-to-fill technical roles drag the average upward and leave the typical hire far below it.
Put $4,683 into the arithmetic, and the identical ten-person team costs $34,850 a year rather than $22,813. Nothing about the team changed. One statistical choice moved the answer by twelve thousand dollars. For what it is worth, SHRM’s 2025 Recruiting Benchmarking report puts the nonexecutive median at $1,200, slightly below the 2022 figure, so the median has not run away either.
When someone tells you turnover costs X per employee, the useful question is not whether X is correct. It is which end of a badly skewed distribution they reached for, and what that choice does for their argument. Use the median for a budget you have to defend. Use the mean if you are building the case for retention spend. Both are defensible. Say which one you used, and you are ahead of most of the people quoting numbers at you.
The other three inputs are worse than skewed. I know of no published figure at all for training cost per hire, equipment cost per hire, or ramp length in a small business. The $1,500, the $750, and the six weeks above are my own working assumptions, not measurements. Your payroll knows those three better than any survey ever will.
What to do with the number once you have it
The formula gives you two levers, and they multiply against each other: the rate at which people leave and the cost of each departure. The rate is the harder problem and the slower one. The cost per departure has one line you can move this quarter, and it is ramp.
Take the same team at the same wage and cut ramp from six weeks to four. Ramp drag falls from $3,024 to $2,016. Cost per departure falls to $5,510. The annual total falls to $19,285. That is $3,528 saved a year, without hiring anybody differently or paying anybody more.
That changes the argument about onboarding. Writing the procedures down so a new hire is not waiting for someone to be free is worth $3,528 a year on this team. It has a budget now instead of a place on a values poster.
Run it on your own figures rather than mine. The calculator I publish at https://callforce.global/tools/turnover-cost-calculator/ does the arithmetic live, prints the formula on the page, and lets you overwrite every default, including the ones I have just admitted are guesses. There is no email gate on it.
Then measure your own ramp, because six weeks is my number and not yours. Ask the last three people you hired when they stopped checking with someone before answering a question. Whatever date they give you is your ramp length, and it is very unlikely to be six weeks.
Miki Furman is the Founder and CEO of Call Force Global.