River
Y CombinatorBacked by Y Combinator
FREE TEMPLATE

Headcount Planning and Cost Template

Every role costed against its own wage bases, premium tier and start month, because one blended rate over-costs the senior hires and under-costs the junior ones.

Free download  ·  No account needed

Fully-Loaded Cost
Aldergrove Systems Inc. · 2026 plan, in-year · the blended rate their own register produced was 18.05%
ReqRoleSalaryStTier In-year wagesReal load RateFlat est.Error
01Staff engineer218,000Feb family199,83337,332 18.68%36,079 under 1,253
02Account executive132,000Mar employee110,00019,012 17.28%19,860 over 848
03Senior engineer176,000Apr employee132,00020,420 15.47%23,832 over 3,412
04Product designer148,000May family98,66722,367 22.67%17,814 under 4,553
05Support specialist68,000Jul employee34,0007,350 21.62%6,138 under 1,212
06Implementation lead96,000Oct employee24,0003,627 15.11%4,333 over 706
Total838,000 598,500110,10818.40% 108,0552,053
Real fully loaded 708,608.22
Flat method 706,555.22
Agreement 0.29%
Gross line error 11,983.45

Three roles over-costed by 4,965.22, three under-costed by 7,018.23. The agreeing total conceals 9,930.44 of offsetting mispricing, and REQ-03 and REQ-04 rank the wrong way round because of it.

Every headcount template costs a hire as salary over twelve times a fully loaded multiplier, and the better ones tell you to derive that multiplier from your own payroll rather than guess at 1.3x. That is better advice and it still misprices every role, because employer payroll taxes are capped per employee per calendar year. Social security stops at 184,500.00 for 2026 and FUTA at the first 7,000.00 paid to each employee. Only medicare runs on every dollar, and medical premium is a flat sum per enrolled head.

So a senior hire's real load rate is lower than a junior hire's, and no single percentage can be both. Aldergrove Systems, twelve staff in Seattle, hiring six people in 2026. Their own register produced a blended load of 18.05 percent, which priced the plan at 706,555.22 against a real 708,608.22. Those agree to 0.29 percent, and the agreement is the problem: three roles were over-costed by 4,965.22 and three under-costed by 7,018.23.

The gross error across six lines was 11,983.45, so an agreeing total concealed 9,930.44 of offsetting mispricing. Two roles competing for one approval slot ranked the wrong way round, because a product designer on family cover carries a higher load rate than a support hire at half the salary. Nobody approves a plan total. They approve roles, by comparing two of them, which is the comparison a blended rate corrupts. Then the trigger to reforecast is a separate question.

Every document in the pack

The cost curve by month, the Payroll Burden Baseline, the Hiring Plan Review, and the Attrition Assumptions sheet.

Employer cost by month, and the rate that will not sit still
Costed per role, against real wage bases Costed at the register's blended 18.05%
Jan
 
 
Feb
1
9.21%
Mar
2
9.12%
Apr
3
9.08%
May
4
9.05%
Jun
4
8.68%
Jul
5
8.64%
Aug
5
8.60%
Sep
5
8.38%
Oct
6
8.23%
Nov
6
7.99%
Dec
6
6.55%
Month
Heads on payroll
Employer tax as a share of that month's wages

October, November and December carry the same six heads and the same 69,833.33 of wages. Employer tax falls from 5,746.01 to 4,573.35, a fifth of it, because the higher earners have run out of taxable social security and Washington unemployment base. No prorated annual figure produces that shape.

Payroll Burden Baseline
Twelve existing staff, 2025 actuals. The sheet exists to produce one number, and then to prove that number cannot be used on a role.
Role2025 wagesTierMatch Social securityMedical Total burdenLoad rate
Founder / CEO265,000family yes10,91819,320 45,69117.24%
VP engineering232,000family yes10,91819,320 43,89218.92%
Staff engineer205,000employee yes10,9188,580 31,68115.45%
Senior engineer172,000 familyyes10,664 19,32040,36823.47%
Head of sales190,000employee yes10,9188,580 30,86316.24%
Account executive128,000 waivedno7,936 010,8028.44%
Product manager158,000 familyyes9,796 19,32038,73724.52%
Support lead82,000employeeno 5,0848,58015,863 19.35%
four further rows: senior engineer, engineer, designer, controller
Total, twelve staff2,021,000 119,700158,340 364,87818.05%
Blended load rate 18.05%
Lowest per head 8.44%
Highest per head 24.52%
Spread 16.08 points

One waived medical election and one family election, 16.08 points apart on the same plan in the same state. The blend describes all twelve together and none of them individually, which is the whole reason it checks a plan total and never costs a role.

Hiring Plan Review
2026 plan · six requisitions · for the Founder, VP engineering and Head of sales
What changed

Nothing in the plan. Six roles, same salaries, same start months. What changed is how they were costed, and three of them were priced wrong enough to change how they rank against each other.

The draft used a blended load of 18.05 percent, which was the right thing to try: it came off our own register rather than a rule of thumb. It produced 706,555.22 against a real 708,608.22. Those agree to 0.29 percent, which is why the draft went three rounds without anyone questioning it. The agreement is the problem.

Finding 1 · REQ-03 is 3,411.67 cheaper than the draft says

The senior engineer at 176,000.00 carries a real load rate of 15.47 percent, second lowest in the plan. The salary exhausts the Washington unemployment base and most of the social security base before year end, and the role takes employee-only cover at 715.00 a month rather than family at 1,610.00.

REQ-03 and REQ-04 have been competing for the same approval slot all quarter. The draft had their load costs 6,018.11 apart with REQ-03 the more expensive. They are actually 1,946.67 apart with REQ-04 the more expensive. The comparison the plan was built to support was inverted.

Finding 2 · REQ-04 is 4,553.11 more expensive

The product designer at 148,000.00 carries the highest load rate in the plan at 22.67 percent, higher than the support specialist at less than half the salary. Family cover is the entire reason. Note what this defeats: a senior multiplier and a junior multiplier would not have caught it, because the driver is not seniority.

Recommendation

Approve all six as planned, with REQ-03 and REQ-04 re-ranked and both tier assumptions confirmed in writing before the offers go out. Add the 3,280.70 attrition reset to the Q4 forecast. Recost in January when the bases publish.

Attrition Assumptions
Why a replacement costs more employer tax than the person they replace, on identical wages across identical months
SeatSalaryOutBackfill in Wages, either wayTax if they stayed Tax across the splitReset
VP engineering232,000AugOct 212,66715,604.73 18,164.462,559.73
Engineer141,000MayJul 129,25010,969.68 11,690.65720.97
Total reset 3,280.70
Component detail, VP engineering
ComponentLeaver, Jan to Aug Backfill, Oct to DecCombined If they had stayedReset
Social security9,589.333,596.00 13,185.3311,439.00 1,746.33
FUTA42.0042.00 84.0042.0042.00
Washington unemployment1,040.06 771.401,811.461,040.06 771.40

A base is paid twice only when the seat's combined annual wages exceed it. FUTA resets on every backfill, because a 7,000.00 base is cleared by any full-time seat. Washington resets on almost every backfill. Social security resets only above 184,500.00 of combined wages, which is why this seat carries 1,746.33 of it and the engineer seat carries none.

Which means the cost of a departure depends on which seat and which month, not on a company-wide attrition rate. The vacancy months here save 31,083.33 of wages and add 3,280.70 of tax, so the gap is real and 10.6 percent smaller than the wage line suggests.

What is in the pack

01

Fully-Loaded Cost sheet

One row per requisition, six components built up separately against their own wage bases and eligibility waits, and the variance against your blended rate stated in words rather than as a minus sign.

02

Payroll Burden Baseline sheet

Your own last full year rebuilt per head, producing the blended rate and the per-head spread around it. Aldergrove's twelve staff ran 8.44 percent to 24.52 percent on the same plan in the same state.

03

Role Register sheet

The six fields that drive cost, captured before anything is costed: salary, start month rather than start quarter, assumed medical tier, match participation, backfill or net new, and the approver.

04

Start-Date Phasing sheet

Twelve months, each role as its own column, with the employer tax percentage that falls through the year as the wage bases run out. The phased column, not the annual run rate, is what an annual budget bridges on. Feeds a cash forecast that a reforecast trigger can actually be tested against.

05

Attrition Assumptions sheet

Departure and backfill per named seat, with the wage-base reset each replacement carries, because a new employee starts every base at zero and the plan pays some of them twice.

06

Headcount Policy Note

What fully loaded means here, what is excluded and why, the six fields a requisition has to carry, the approval thresholds, and the five events that force a recost.

07

Load Rate Method

Each component, its base, its waiting period, and where its rate is read from. Two of the six are company-specific and on a notice in your own file rather than in any benchmark.

08

Hiring Plan Review

The memo for whoever signs off headcount, naming at most three findings largest first and the role comparisons a blended rate was getting backwards. Once approved, each role's monthly cost is the number a manager carries into next cycle's budget request, not a guess.

09

Attrition and Backfill Note

Why replacing someone costs more employer tax than keeping them, component by component, and how to net a vacancy's wage saving against the tax it adds before calling a freeze a saving.

How it works

  1. 1

    Take the files, or open the pack

    Download the blank Word and CSV files with no account and no signup, or open the same pack in River and hand it your payroll register and the roles you plan to hire.

  2. 2

    Derive your blend, then set it aside

    Rebuild the register per head to get the rate your payroll actually carries. It checks a plan total. It never costs a role, and the spread it exposes is why.

  3. 3

    Cost each role against its own bases

    Six components per head, month by month, each stopping at its own wage base and starting after its own eligibility wait. Tier and start month move a role further than its level does.

  4. 4

    Find the rankings that invert

    The output is not the plan total, which will look fine. It is the pairs of roles whose cost ranking flips between the two methods, plus what a capital request does to the same budget.

Frequently asked questions

Is this free?

The download is, with no account and no email. Every document and sheet named above comes down complete, filled in for the sample company so the structure is visible. Editing with AI creates a free account, which is what lets the agent read your payroll register and cost the plan for you.

What format are the downloaded files?

Word documents and CSV sheets. The four documents arrive as .docx with their headings and tables intact, and the five sheets as .csv. Nothing needs converting: they open in Word, Pages, Google Docs, Excel, Numbers and Sheets as they are.

Why not just use a fully loaded multiplier?

Because it is an average of per-head rates, so it reproduces a total and cannot reproduce any member of it. Aldergrove's twelve staff ran 8.44 to 24.52 percent. Use the blend to check your plan total, then cost each role on its own.

Would two multipliers, one senior and one junior, fix it?

No, because seniority is not the main driver. A product designer at 148,000.00 on family cover carried a higher load rate than a support hire at 68,000.00, since premium is a fixed sum per head. Tier and start month both move a role further than its level does.

Does it work outside Washington?

Yes. The federal components are identical everywhere and the mechanism is the same in every state: a state unemployment rate and a taxable wage base, both on your own annual notice. Only the two numbers change, and the pack asks for them by name.

Why does a backfill cost more than the person who left?

Because a replacement's wage bases start at zero. FUTA applies to the first 7,000.00 paid to each employee, and the successor employer exception covers acquiring a business, not replacing a person. Aldergrove's two 2026 backfills add 3,280.70 on identical wages.

What if I do not have last year's register?

You can still cost every role, because that arithmetic needs only statutory rates plus your own state rate notice and carrier invoice. What you lose is the check on the plan total, so one input error stays invisible. Get the register when you can.

Cost the plan you are about to approve

Send last year's payroll register and the roles you plan to hire. The per-role variance is what changes a decision.

Edit with AI