River
Y CombinatorBacked by Y Combinator
FREE TEMPLATE

Recruiting Agency Agreement Template

Four documents and three sheets, including the one number that decides which agency you keep and that no invoice will ever show you.

Free download  ·  No account needed

One metric, four decisions

Fees over surviving salary, not fees over placements

Effective rate = fees paid, minus cash actually recovered, over the combined salaries of the seats that agency filled which are still filled at twelve months.

Cash Actually Recovered

Not cash owed. A guarantee clause that fired and produced a credit nobody used is worth zero and goes in as zero. Three of six departures in the worked example landed inside a guarantee window; one produced money.

Seats Still Filled

Not placements made. A free replacement who stayed keeps the seat filled and counts. Somebody who left with nothing coming back bought an introduction, not a seat.

Salaries, Not Headcount

The correction that makes the comparison honest. An agency working general manager roles places at 128,000 and one working banquet managers places at 64,000. Cost per seat punishes the first for reasons that have nothing to do with performance. Dividing by salary removes the mix and lands in the same units as the headline rate.

Twelve Months, Fixed

Not the guarantee window and not the fiscal year. Long enough that a bad hire has surfaced, short enough that you are still measuring the placement rather than the job. Anything younger is shown separately as in-flight, because counting it flatters whoever placed most recently.

Agencies get chosen, compared and renewed on a fee percentage. Fifteen is cheap, twenty is normal, twenty-two is expensive, and that number is on the front page of every agreement. It is the price of an introduction. What you are buying is a filled seat, and those are the same thing only if nobody leaves. So this pack computes one number instead, and the agreement template exists to make that number negotiable rather than to be filled in and filed.

Brightmoor Hotels used three agencies across eighteen months. Fifteen placements, 1,313,000 of base salary, 263,820 in fees at a blended 20.1 percent. Six people left inside twelve months. Once the departure dates land next to the invoices, the effective rate reads 22.0 percent for the agency charging 22, 28.3 for the one charging 20, and 67.0 for the one charging 15. The ranking inverts, and the cheapest agency is three times the price.

Nobody at Brightmoor was wrong about the percentages. Fees live in accounts payable, departure dates live in the human resources system, and no report in either one joins them. Two smaller findings fall out of the same join. One agency's 15 percent was charged on base plus target bonus, so 16.5 percent on base, and of the 89,370 in fees sitting on people who left, 4,640 ever came back. This sits after the interview loop and beside the offer.

The agency charging 15 percent is three times the price

Three agreements, fifteen placements, six departures, and the join that nobody's applicant tracking system will run.

Agency Register

Illustrative for a fictional hotel group, Brightmoor Hotels. Eighteen months, three agencies, thirty requisitions released.

AgencyHeadline
rate
Fee baseTrue rate
on base
GuaranteeRemedyElectionEffective
rate
Ardenway Partners22%base only22.0%60 daysfree replacementagency22.0%
Kelbrook Search20%base only20.0%90 dayspro-rata cash refundcompany28.3%
Vellmont Talent15%base plus target bonus16.5%30 dayscredit against a future placementagency67.0%
All three: 15 placements, 1,313,000 base salary placed, 263,820 fees, 4,640 recovered, 259,180 net, 10 seats still filled at twelve months on 943,000 of salary. Blended headline 20.1%, blended effective 27.5%.

Read the first and last columns together. The rate order and the outcome order are exact opposites. Vellmont's own headline is also not what it charges: 15 percent of base plus a ten percent target bonus is 16.5 percent of base, worth 4,020 across four placements, and it is invisible in procurement because both agreements say percent of first-year salary and only one of them means base.

Placement and Fallout Tracker

The departure date is the field nobody keeps. Every figure on the register is computed from this sheet.

IDAgencyRoleStartBaseFeeDepartedDayIn
window
Cash
back
Seat at
12 mo
K-02KelbrookExecutive Housekeeper04 Nov 2472,00014,40004 Jan 2561yes, of 904,640empty
K-04KelbrookFood and Beverage Director10 Feb 2596,00019,20008 Jun 25118no0empty
A-02ArdenwayDirector of Sales20 Jan 25102,00022,44002 Mar 2541yes, of 600filled by replacement
V-01VellmontRestaurant Manager03 Feb 2568,00011,22001 Mar 2526yes, of 300, credit expiredempty
V-02VellmontBanquet Manager10 Mar 2564,00010,56018 Apr 2539no0empty
V-03VellmontRestaurant Manager07 Apr 2570,00011,55009 Sep 25155no0empty
Six departures of fifteen placements, a 40% fallout rate. Three landed inside a guarantee window. One produced cash. 89,370 of fees sat on people who left; 4,640 came back, which is 5.2%. The other nine placements are still filled and are not shown.

Fallout is not evidence of a bad agency. Monthly total separations in accommodation and food services [ran 5.2 to 6.1 percent](https://data.bls.gov/timeseries/JTS720000000000000TSR) through the first half of 2026, against [3.2 to 3.4 percent](https://data.bls.gov/timeseries/JTS000000000000000TSR) across all nonfarm employment. A hotel group losing 40 percent of agency placements inside a year is seeing its own labour market. Which is precisely why the guarantee terms decide the comparison: you cannot make fallout rare, only decide in advance what happens when it occurs.

Submission Quality by Agency

The second dataset, and the one that carries a cost appearing on no invoice.

AgencyReqsSubsReturned
unread
1st
interviews
OnsitesOffersPlacedSub to
1st
Sub to
placed
Review
hours
Ardenway Partners92201496563.6%22.7%4.4
Kelbrook Search114731987640.4%12.8%9.4
Vellmont Talent1096142175421.9%4.2%19.2
Total30165175424181532.7%9.1%33.0

Two ratios, because they answer different questions. Submission to first interview says whether the agency read the requirement register, and it ran from 21.9 to 63.6 percent on the same requisitions. Submission to placement is the outcome. Then the line nobody bills: at twelve minutes a submission, Vellmont consumed 19.2 of the 33 review hours, 58.2 percent of all submissions, for 26.7 percent of the placements and one surviving seat. Before concluding anything, check your own median days to a decision. Silence from the company is the most common reason an agency goes wide.

What is in the pack

01

Agency Register

One row per agency with the five clauses that decide the effective rate pulled out and made comparable. Headline rate, fee base, guarantee window and the date it runs from, the remedy classified rather than quoted, who holds the election, and the exclusions listed individually rather than summarised as standard.

02

Placement and Fallout Tracker

Every external placement with start date, base salary, the fee base actually used, and the departure date. Plus a last-verified date, so a blank departure field means somebody still employed rather than somebody nobody checked on. Those are different facts.

03

Submission Quality by Agency

Submissions, returned unread, first interviews, onsites, offers and placements, with two ratios kept side by side. Then review hours at twelve minutes a submission, and each agency's share of submissions beside its share of placements.

04

How the Effective Rate Is Built

The definition, the four decisions inside it, and the three variables that decide what a guarantee is worth: window, remedy, exclusions. Also what the metric does not measure, which is quality of hire, speed, and whether an agency is worth using at all.

05

Engagement Terms

Nine clauses with the reasoning under each, and the three worth negotiating before fee percentage named as such. Clause three carries the election, the no-exclusions position, and the refusal to accept a credit against a future placement as a remedy.

06

Submission Rules

Seven things a submission must contain, what gets one returned unread, and a section on what you owe the agency in return. Three business days to a decision, a specific reason on every decline, feedback within two days.

07

Performance Review Note

Seven sections, sent before the conversation rather than during it, with the group range included so the argument is about the requisitions rather than the metric. Notes cover leading with the ask that is not about money.

How it works

  1. 1

    Send the agreements, all of them

    Including the ones you think are identical. Five fields get pulled from each: fee base, window, remedy, who elects, and the exclusions. That is why documents rather than a summary.

  2. 2

    Then eighteen months of placements

    Agency, role, start date, base salary, fee invoiced. And the departure date for anybody who has left, which is the field that will be missing.

  3. 3

    The rate gets computed, not estimated

    Fees net of cash actually recovered, over the salaries of surviving seats. Anything younger than twelve months is excluded and reported separately as in-flight.

  4. 4

    The review note goes out before the call

    Their number, the blended number, and the full range across every agency on the same requisitions. Plus your own turnaround times, whether or not they flatter you.

Frequently asked questions

Why divide by salary instead of counting seats?

Because cost per surviving seat punishes whoever places your expensive roles. One agency here worked general manager and director roles at 128,000 and 102,000; another worked banquet managers at 64,000. Dividing by salary removes the mix and lands the answer in percent, directly comparable to the headline rate it is replacing.

Is a 40 percent fallout rate unusually bad?

No, and reading it that way leads to the wrong conversation. Separations run high everywhere and higher in hospitality: monthly rates ran 5.2 to 6.1 percent through the first half of 2026 against 3.2 to 3.4 percent across all nonfarm employment. You cannot make fallout rare. You can decide in writing what happens when it does.

Which clause should I actually negotiate?

The guarantee remedy, then who holds the election, then the fee base. Each is worth more than a point or two of fee, and each is easier for an agency to concede because none of them costs anything unless they place badly. Fee percentage is fourth, and it is what the whole meeting is usually about.

What is wrong with a credit against a future placement?

It is worth zero unless you were going to use that agency again, which converts a refund into a retention device. In the worked example a departure at day 26 fell inside a thirty-day window, the clause fired exactly as written, and the resulting credit of 11,220 expired unused. The tracker records that as zero.

Our departure dates are not in the recruiting system. Now what?

They rarely are, and that is why nobody has this number. They sit in the human resources system or in payroll, keyed on employee rather than on candidate. Join them on name and start date, then confirm the still-employed names explicitly rather than reading a blank as a survivor. Those are different facts.

Where does this sit against the rest of hiring?

After it. The interview loop decides who gets hired, pipeline analysis covers your own funnel rather than an agency's, and the offer prices the number an agency fee is calculated on. Four documents and three CSV sheets, free.

Put the departure dates next to the invoices

Send the agreements and eighteen months of placements. The first thing back is whether your rate ranking and your outcome ranking agree, and which clause is worth more than the percentage you have been negotiating.

Edit with AI