How to Run Headcount Planning on a Dollar Cost per Hire

Headcount planning decides who a company hires, when, and at what fully loaded cost. A common shortcut multiplies salary by 1.25 to 1.4. But most of the load is a fixed sum per head or a tax that stops at a cap, so one multiplier overprices senior hires, underprices junior ones and misdates the cash.

The shortcut is two decades old, the government’s own data appear to support 1.43, and for headcount planning both numbers are wrong in directions that can be measured. Exhibit 1 prices one hire line by line from the 2026 rules; the rest of the piece shows where the multiplier breaks and what each break costs.

Exhibit 1. The 2026 cost of a US hire has three shapes, and only some of it is a percentage of salary

Line	2026 rule	Shape	Employer cost on a $160,000 salary
Salary	Base pay, which already covers vacation and holidays	Salary	$160,000
Social Security	6.2% of wages up to $184,500	Capped percentage	$9,920
Medicare	1.45% of all wages; the extra 0.9% above $200,000 is the employee’s	Percentage	$2,320
Federal unemployment (FUTA)	0.6% of the first $7,000, more in credit-reduction states	Fixed per head	$42
State unemployment (California)	3.4% for new employers plus 0.1% training tax, on the first $7,000	Fixed per head	$245
Health insurance	Employer share of the average premium: $7,885 single, $20,143 family (2025)	Fixed per head, by tier	$7,885
Retirement match	Plan-specific; 3% assumed	Percentage	$4,800
Equipment and software	Company-specific; $4,000 a year assumed	Fixed per head	$4,000
Recruiting	SHRM average $5,475 per non-executive hire (2025)	One-off, before day one	Not included
Loaded cost			$189,212, or 1.18 times salary

Sources: IRS Publication 15 (2026); Social Security Administration; US Department of Labor; California EDD; KFF Employer Health Benefits Survey (2025); SHRM (2025). The match and equipment lines are hypothetical. Employer shares and totals are our arithmetic.

Loaded cost = head floor + salary × variable rate, with Social Security charged only up to $184,500

Headcount planning often runs on a rule of thumb from 2005

The range has an early, documented source. Joseph Hadzima, in a column long hosted on MIT’s website, put salary, employment taxes and benefits “typically in the 1.25 to 1.4 times base salary range” (Hadzima, 2005). The column carries a 1994-2005 copyright. Its inputs date it: Social Security at 6.2% on the first $90,000 of salary, which was the 2005 wage base, and single health coverage at $2,000 to $3,000 a year.

Those inputs have moved. The Social Security base for 2026 is $184,500 (SSA), 2.05 times the 2005 figure. KFF puts the average single premium for employer coverage at $9,325 in 2025 (KFF, 2025), 3.1 to 4.7 times Hadzima’s range. Only the $7,000 unemployment base has stayed where it was.

The stakes are large because people are most of what a startup spends. Among Kruze Consulting’s clients, payroll-related costs make up 76% of venture-backed startups’ operating expenses (Kruze Consulting, 2024), a figure from one accounting firm whose sample the firm describes in two different ways. At that share, a 10% error in the cost of a head becomes a 7.6% error in operating spending.

Applied to a salary, the government’s 30% benefit share counts paid leave twice

The obvious recalibration for headcount planning is the Bureau of Labor Statistics. Its Employer Costs for Employee Compensation (ECEC) survey put private-industry compensation at $46.89 per hour worked in June 2026, with wages and salaries at $32.82, or 70.0%, and benefits at 30.0% (BLS, 2026). Dividing one by the other gives a multiplier of 1.43, close to the top of Hadzima’s range.

That division mixes two definitions. ECEC measures cost per hour worked, and its hours exclude leave hours (BLS Handbook of Methods). Paid leave (vacation, holiday, sick and personal days) is one of its five benefit categories. An offer letter’s salary already pays for those days. Hadzima made the same point in a single line: “Vacation is another cost but is subsumed in the basic salary.”

Exhibit 2. The same BLS data give a multiplier of 1.47 or 1.29, depending on where paid leave sits

Full-time, professional and related occupations, June 2026	$ per hour worked	% of pay for time worked	% of a salary (time worked plus paid leave)
Wages and salaries for time worked	52.92	100.0	87.6
Paid leave	7.52	14.2	12.4
Supplemental pay (overtime, bonuses)	3.21	6.1	5.3
Insurance	6.17	11.7	10.2
Retirement and savings	3.02	5.7	5.0
Legally required (Social Security, Medicare, unemployment, workers’ compensation)	4.96	9.4	8.2
Total compensation	77.80	147.0	128.7

Source: BLS, ECEC June 2026, Table 5. Percentages are our arithmetic.

On a salary base, the load for full-time professional occupations is 1.29, and 1.23 if bonuses sit on their own budget line. Applying the naive 1.47 instead of 1.29 to ten engineers at $160,000 adds $292,667 to the plan, more than the $189,212 loaded cost of an eleventh engineer.

Smaller employers carry lighter benefits. In ECEC’s March 2026 data, benefits were 25.9% of compensation at private establishments with 1 to 49 workers and 35.2% at those with 500 or more (BLS, Table 6). Converted to a salary base, that is 1.24 against 1.35. Kruze’s client data put payroll taxes at 7.6% and benefits at 8.2% of wages, which adds up to a load of about 1.16. A multiplier moves with the line between wage and add-on, so the first question about any multiplier is where its author drew that line.

But no single multiplier fits a team, because the load is a fixed sum plus capped percentages

Even a correctly measured average misprices individual hires, and headcount planning is done hire by hire. Within one health plan and coverage tier, the employer pays the same premium whether the salary is $70,000 or $240,000 ($7,885 for single cover on KFF’s averages); unemployment taxes stop at the first $7,000, and Social Security stops at $184,500. Only Medicare and the retirement match scale all the way up.

Insurance pricing met this problem decades ago. Actuaries once loaded every expense as a constant percentage of premium. The Casualty Actuarial Society’s Basic Ratemaking records the result: “the All Variable Expense Method undercharges the risks with premium less than the average and overcharges the risks with premium more than the average” (Werner & Modlin, 2016). Insurers fixed it with an expense constant, a flat fee per policy, and in workers’ compensation with a variable provision charged only on the first $5,000 of premium. A payroll load has the same two parts. The employer has more say over its fixed part than an insurer does: an insurer prices costs it will incur on a customer’s behalf, while an employer chooses some per-head costs, such as equipment and the health plan’s tier.

Call it the head floor: the dollars a hire costs before the first dollar of salary. For a single-cover hire in the example below it is $12,172 a year; with family cover it is $24,430.

Exhibit 3. Four hypothetical California hires in 2026: the loaded multiple runs from 1.14 to 1.46

Hire	Salary	Health cover	Employer taxes	Health	Match and equipment	Loaded cost	Multiple	1.325 rule minus loaded cost
Support specialist	$70,000	Single	$5,642	$7,885	$6,100	$89,627	1.28	+$3,123 (+3.5%)
Support specialist	$70,000	Family	$5,642	$20,143	$6,100	$101,885	1.46	−$9,135 (−9.0%)
Engineer	$160,000	Single	$12,527	$7,885	$8,800	$189,212	1.18	+$22,788 (+12.0%)
Senior engineer	$240,000	Single	$15,206	$7,885	$11,200	$274,291	1.14	+$43,709 (+15.9%)

Hypothetical hires at a new California employer. Taxes: Social Security, Medicare, FUTA at 0.6%, and California unemployment and training tax, 2026 rules. Health: KFF’s 2025 average premium minus the average worker contribution. The 3% match and $4,000 a year of equipment and software are assumptions; workers’ compensation is excluded. The rule is the midpoint of Hadzima’s range; the loaded costs and differences are computed by us.

The health tier moves the multiple more than salary does: family cover adds $12,258, which is 17.5% of a $70,000 salary. The engineer’s taxes, health and match come to 15.8% of salary, the same proportion as Kruze’s client average. A plan priced at the rule’s midpoint overstates ten such engineers by $227,880, enough to fund another engineer at $189,212. Overstated cost leaves a hire unmade; understated cost, on a team weighted toward junior staff with families, shortens the runway.

The cost also has a calendar: employer taxes are front-loaded, and part arrives the next January

An annual figure is enough for a budget. Headcount planning for cash also needs the month in which each dollar leaves, and payroll taxes do not leave evenly. Social Security and both unemployment taxes start again from zero every 1 January, and the employer’s share runs only until the employee’s wages with that employer reach the cap (IRS Publication 15, 2026).

Exhibit 4. Employer payroll tax on a hypothetical $240,000 salary, paid twice a month from 1 January, $

Month	Social Security	Medicare	Unemployment (FUTA and California)	Total
January	1,240	290	287	1,817
February to September, each	1,240	290	0	1,530
October	279	290	0	569
November and December, each	0	290	0	290
Year	11,439	3,480	287	15,206
January 2027, potential			105 to 371	105 to 371

Hypothetical salary at a new California employer, 2026 rules. The Social Security cap is reached in the 19th of 24 pay periods. The last row is California’s potential 2026 FUTA credit reduction, 1.5% to 5.3% of $7,000, due with the fourth-quarter deposit. Our arithmetic.

The employer pays $1,817 of payroll tax on this salary in January and $290 in December, 6.3 times as much (our arithmetic), and the first quarter carries 4.2 times the tax of the fourth. A plan that spreads the $15,206 evenly, at $3,802 a quarter, understates the first quarter by $1,076 and overstates the fourth by $2,652 for each such hire.

Two rules make the calendar sharper. The cap applies per employer. A senior hire who joins in September, having already paid the maximum elsewhere, still costs the new employer its full 6.2%, because employees can reclaim excess Social Security tax and employers cannot (Kruze Consulting, 2024). And the federal unemployment tax is $42 a head only where the state has repaid its federal loans (US Department of Labor). California employers paid an extra 1.2% of the first $7,000 for 2025, and the Department of Labor lists a potential 2026 reduction of 1.5% to 5.3%, final on 10 November (US Department of Labor). That is $147 to $413 a head instead of $42, payable by 31 January 2027 (our arithmetic).

Recruiting is the small part of a hire’s cost; turnover is the rate that matters

The cost of hiring itself is smaller than its reputation. SHRM’s benchmarking report, with 2021 data from 472 member organisations, put the median cost per hire at $1,244 and the average at $4,683, an average that sits above the survey’s own 75th percentile of $4,375 (SHRM, 2022). SHRM recommends the median when the two diverge. Its 2025 survey of 2,371 members, unweighted, gives averages of $5,475 for non-executive and $35,879 for executive hires (SHRM, 2025). Against Exhibit 3’s engineer, $5,475 is 2.9% of a year’s loaded cost. The measure leaves out hiring managers’ time, which is one reason a consultant quoted by SHRM could estimate the total cost to hire at “three to four times the position’s salary” (SHRM, 2022). The article gives no method for that figure.

Economists have a cleaner way to put hiring into the cost of a seat. Daniel Raff and Lawrence Summers wrote the user cost of labour as C = w + (i + q)T, “where w is the wage, i the relevant interest rate, q the turnover rate, and T training costs” (Raff & Summers, 1986). The paper, published in the Journal of Labor Economics in 1987, tests the formula on Henry Ford.

In 1913, annual turnover at the Ford plant reached 370 percent. Ford hired 50,448 men that year to keep an average workforce of 13,623, about 3.7 hires for every job (our arithmetic). In January 1914 it introduced the $5 day, which doubled most workers’ pay from roughly $2.34. Turnover fell, but with T set at a week’s pay, $30, the saving came to about $0.16 a day, and the authors could not credit lower turnover with more than about 6% of the program’s cost.

For a knowledge worker T is far larger than a week’s pay. Take Exhibit 3’s engineer and assume three months at half output before full productivity, plus SHRM’s $5,475: T is $29,126. At an assumed turnover of 20% a year, the second term adds $5,825 a year, 3.1% of the loaded cost; at 30% it adds 4.6%. Turnover belongs in headcount planning as a rate on every seat, sized from the company’s own history. A multiple of three or four times salary does not.

For pricing billable hours, Hadzima’s multiplier is the right tool

Hadzima’s column makes the case for simplicity, telling founders that “instead of figuring out each cost component separately you can develop some simple metrics”. His example is an engineering-services entrepreneur who multiplies base salary by 1.25 for taxes and benefits, by 1.75 for rent and equipment and by 1.25 again for management and non-billable time. On her arithmetic, “the fully functioning managed employee costs about 2.7 times the base salary” in that business. For pricing billable hours, a compound multiplier is the right tool, because the question is what each hour must recover. At seed stage, when the plan holds five similar hires, the difference between 1.2 and 1.3 is also small next to the uncertainty in start dates and revenue.

For pricing and for sanity checks, Hadzima’s case stands. For a team of engineers paid like those in Exhibit 3, a rule near 1.3 also overstates cost, which is the safer error for cash.

It fails in three places. Overstatement has a price: in Exhibit 3, the rule withholds an engineer the company can afford. The error is not uniform, so a plan weighted toward junior hires with family cover runs short rather than long. And no annual multiplier, however well calibrated, says that the first quarter carries four times the tax of the fourth. The multiplier is a test of the dollar build, and it cannot replace it.

A dollar build for founders, three questions for boards

• The dollar build. Price each hire in dollars: the head floor (health tier, unemployment taxes, equipment) plus the percentages, with Social Security stopped at $184,500. Record each hire’s state and health tier; in Exhibit 3, family cover moves the loaded multiple further than a $90,000 difference in salary does.
• The tax calendar. Spread payroll taxes by month rather than evenly, and add any FUTA credit reduction to the following January.
• Recruiting and turnover. Put recruiting fees before the start date and turnover in as a rate on every seat, sized from your own history.
• The multiplier as a check. Keep 1.2 to 1.3 as a check on the build, and investigate when the two differ. The annual budget template and variance tracker below is a starting point.

A board reviewing the plan needs three answers from it. The first is the loaded cost per planned hire and the quarterly payroll-tax profile, in place of a multiplier. The second is which BLS figure, if any, sits behind the multiplier, and whether that figure counts paid leave twice. The third, for a California company, is whether the plan carries the 2026 FUTA credit reduction due in January 2027.

A hire costs a fixed sum before it costs a percentage

Health cover, unemployment taxes and equipment arrive per head; Social Security stops at $184,500; only Medicare and the match scale with pay. A single multiplier averages these shapes, so it overcharges senior salaries, undercharges junior hires with families and spreads the cash evenly across a year in which it is front-loaded. Headcount planning that prices each hire in dollars and dates its taxes by month gets the total and the timing right, and leaves the multiplier the job it does well, a quick check on the build.

Sources

• Bureau of Labor Statistics (2026). Employer Costs for Employee Compensation, June 2026 (USDL-26-1494), Tables 5 and 6, and Technical Note; Handbook of Methods: ECEC Concepts.
• California Employment Development Department. Tax Rates, Wage Limits, and Value of Meals and Lodging (2026 rates).
• Hadzima, J. G., Jr. (2005). How much does an employee cost? Column formerly hosted at web.mit.edu.
• Internal Revenue Service (2026). Publication 15 (Circular E), Employer’s Tax Guide.
• KFF (2025). 2025 Employer Health Benefits Survey.
• Kruze Consulting (2024). Startup Payroll Costs and Tax Budgeting for New Hires.
• Navarra, K. (2022). The real costs of recruitment. SHRM, 11 April.
• Raff, D. M. G. and Summers, L. H. (1987). Did Henry Ford pay efficiency wages? Journal of Labor Economics 5(4), S57–S86; NBER Working Paper 2101 (1986).
• Social Security Administration, Office of the Chief Actuary. Contribution and Benefit Base.
• Society for Human Resource Management (2022). SHRM Benchmarking: Talent Access Report (data collected April to November 2021); (2025) SHRM Releases 2025 Benchmarking Reports, press release, 15 October.
• US Department of Labor, Office of Unemployment Insurance. Unemployment Insurance Tax Topic; FUTA Credit Reductions, with the actual 2010–2025 and potential 2026 tables.
• Werner, G. and Modlin, C. (2016). Basic Ratemaking, 5th edition. Casualty Actuarial Society.

Tax rates and wage bases are the 2026 rules published by the IRS, the Social Security Administration, the Department of Labor and California’s EDD; KFF’s premiums are 2025 figures, and California’s 2026 FUTA credit reduction is the Department of Labor’s preliminary estimate. Any figure without a citation is computed from those rules and from the BLS, KFF, SHRM and Kruze figures cited in the text. The hires in Exhibits 3 and 4, the retirement match, equipment cost, ramp time and turnover rates are hypothetical.