The five clauses your client has been coached to redline

This article describes contract language, and in one section two state court decisions, and it is not legal advice: whether any clause here holds up depends on your own state's law and on how your own agreement is written. Acting on it has a price of its own, and the price is the relationship. Reopening signed terms with a live client can cost you the account, and a contingency recruiter who refuses a term can lose the role to the next agency, so decide what you are willing to lose before you send a redline.

Search recruitment agency contract clauses from a US browser and Google answers in your client's voice. Your counterparty has read that answer. Their checklist and your agreement are the same five clauses seen from opposite ends of the table, and only one end has been written up.

What your client is being told to ask for

The AI Overview on that query, read on 4 September 2026, ran three numbered sections with three bullets each: Fee Structure and Calculation, Candidate Ownership and Non-Circumvention, and Replacement and Refund Guarantees. Three of those bullets, verbatim:

> "Time Windows: Limit the candidate ownership window (e.g., 6 to 12 months) so an agency cannot claim a fee if a candidate applies directly much later."

> "Exclusion of Prior Applicants: Exclude candidates the company has already interviewed or rejected independently."

> "Full Refund Protection: Negotiate for a full cash refund rather than a replacement-"

The third is truncated in our capture, so read it as unfinished. The synthesis cites Dover, BountyJobs, a Baltimore law firm and a YouTube channel inline, opens on a LinkedIn post, and page one of that query returned nine results with not one written from the agency's side. These summaries are regenerated per query, so the wording above is a snapshot of that date.

Two of those sources publish the whole checklist, and both have a commercial interest in it. Dover sells in-house and embedded recruiting, which competes with the agency model. Its Recruitment Fee Agreements: 5 Clauses, January 2026, asks buyers to take a full refund over a replacement-only guarantee or a hybrid refunding in full for 30 days then prorating, and to request net 60 or net 90 terms. The same piece asks them to cap the ownership window at "6 months maximum instead of 12", exclude candidates who applied directly or came by referral after a rejection, and require a formal submission rather than LinkedIn outreach to trigger ownership.

BountyJobs, a marketplace connecting employers to agencies, adds the rest in Top 5 Terms for Every Recruiting Agency Contract, July 2023. It asks for a full refund "if a candidate leaves the job within the first 60-90 days", and a no-poaching clause that "prevents the agency from recruiting your employees away". It also asks for OFCCP compliance, and indemnification so the client is "protected from liability for the agency's actions".

Neither page is dishonest. Neither is written for you.

Clause 1: the payment trigger

The fee can attach at the invoice, at offer acceptance, at the start date, or past a probation period. Gordon Feinblatt LLC, a Baltimore law firm with nothing to sell here, tells employers to move it: its December 2017 piece Rethinking Recruitment Agency Agreements argues a trigger set at offer acceptance should be rejected and payment attach later. That the advice was worth writing tells you an offer-acceptance trigger was on paper its employer clients were reading in 2017, and nothing here measures how common it is on agency paper now. If your client's paper puts the fee behind probation, that is where it came from.

One signed agreement beats a template here. A contingency search fee agreement between Wilshire Energy Partners, LLC and Foothills Petroleum, Inc., dated August 2016, was filed as an 8-K exhibit and is public on SEC EDGAR. Its trigger: "The Fee shall become 100% due and payable before or on the starting date of employment." One agreement, one search, two parties. Nothing in it says the market drafts the same way.

Keep the trigger and the payment term separate in the conversation, because only one of them costs you fee. Net 30 moving to net 60 costs you float. A trigger moving to past probation hands you the whole risk of a fall-off before you have been paid anything, with your guarantee clause on top. Define "starting date" while you are in there, and what happens if the person never arrives at all.

The fee models this trigger sits inside are in recruitment agency fees, and which engagement model is worth taking at all is in contingency versus retained recruiting.

Clause 2: the guarantee, and why stacking is the expensive part

Four structures circulate: replacement at no cost, a full refund, a pro-rated refund scaled to how long the person stayed, and a pro-rated replacement priced against a new fee. Which one you signed changes what the clause costs you to carry, and the model for that, with the accounting treatment, is in what a recruitment agency replacement guarantee actually costs.

Gordon Feinblatt describes the agency-side remedy its clients saw in 2017 as "no refund, but the agency will attempt to find a replacement", and tells employers to push for "a full refund if the employee resigns or is terminated within the refund window" instead. Foothills shows a negotiated version: no fee is owed where the candidate "leaves under his/her own volition or is terminated for cause within ninety (90) days after commencing employment". Where the fee was paid, "Advisor shall refund 100% of any Fee received for such Candidate, minus $2,500." Cash back with a fixed carve-out on the front, in one agreement, where the $2,500 is those parties' number rather than a convention.

The expense comes from two clocks running at once. The guarantee window and the ownership window are separate, and a client's paper can leave both running on one candidate: a ninety-day guarantee inside a one-year ownership window means your exposure spans both. The checklists ask for the longest version of one and the shortest of the other. Read them together, and if you concede cash back, price it as a longer obligation rather than a wording change.

Clause 3: candidate ownership, and how to draft it

Three moving parts decide whether a fee attaches to a hire at all: what counts as a submission, how long that submission lives, and who counts as the client.

Dover's ask on the first is one you should probably give. A formal, written, dated submission requirement protects you as much as the buyer, because a dated introduction the client acknowledged is the first thing anyone will ask you to produce if the fee is disputed. What that record is worth, and the three things that decide whether an unpaid fee is recoverable, are in the backdoor hire.

On the window, nobody has published a number. The only figure in circulation is a BountyJobs post from November 2015 stating that "six months ownership for both the agency and employer is the new market standard". No survey, sample or method is cited for it anywhere on the page, it is a decade old, and the outlet serves the buyer side.

Dover asks buyers to cap the window at six months, so the unsourced assertion and today's opening ask are the same number. The rationale attached to that ask in the AI Overview, that an agency should not be able to claim a fee on a direct application made much later, is Google's synthesis of Dover's checklist rather than a reason Dover states. Every window is negotiated, and none is standard, including yours.

Two filed agreements show range rather than norm. Foothills: no fee is owed if the "Company does not employ Candidate within one (1) year of Advisor's first submission", with carve-outs where the candidate "contacted Company prior to Advisor's introduction" or was "identified... from another source" first.

A 2001 agreement between AlphaNet Solutions, Inc. and EDS Information Services L.L.C., also filed with the SEC, runs its clock three months from termination instead. The two clocks are not measuring the same thing. AlphaNet's is structurally a subcontract and termination-transition clause rather than a new-business fee agreement, so its window governs contractors already placed with a client relationship that is winding down, not whether a fee attaches to a new hire.

Two instruments fifteen years apart, and of different kinds, are not a sample.

The prior-applicant carve-out is reasonable in principle and the drafting is what goes wrong. Gordon Feinblatt tells employers to carve out candidates "with whom the employer had prior contact".

Concede the principle and attach evidence to it: the exclusion applies where the client can produce a dated record in its own system predating your submission. Without that it is an assertion made after the hire, and the cheapest way to lose a fee you earned. The definition of "client", and the hole where the hire lands in a sister entity instead, is covered on the same page.

Clause 4: off-limits, the clause that sterilises your own desk

An off-limits clause stops you recruiting out of the client you just placed into, and BountyJobs recommends buyers ask for it in those terms. It is the only clause here priced in candidates rather than dollars. If you work a vertical, every client you sign under a broad version removes a pool from the desk that fills your next search, so you pay for it on roles you have not won yet.

The Foothills agreement carries a version running against the recruiter: the advisor may not solicit the client's employees "during the Term of this Agreement or for a period of thirty-six (36) months after termination". Three years past the relationship, in one signed agreement.

The 2001 AlphaNet agreement carries a restraint pointing the other way, and a much shorter one. For a 30-day window after termination the agency "shall not re-assign, re-deploy, offer future assignments or in any other way interfere or compete with EDS' right to hire". That stops the agency pulling back the contractors it had placed rather than protecting the client's payroll from the recruiter.

Four things decide what it costs you, and all four are drafting. Scope is the entity that signed, or every affiliate in the group; population is employees you placed there, or anyone on the payroll. Duration is the term, or years past it. And conduct is whether an inbound application counts as soliciting, which is the one that turns a narrow clause broad without changing its length.

The fallbacks come off in that order: your own placements, then the signing entity, then an end date, then inbound applicants carved out. That order is a negotiating sequence rather than a statement of law. Whether any version of this clause binds you at all is a state question, and the two decisions people reach for, neither of them about a client-facing off-limits clause, are two sections down.

Clause 5: indemnification creep

Indemnification is the only clause here that can cost more than the fee.

Gordon Feinblatt's 2017 piece describes agency agreements that run the indemnity one way, in the agency's favour, and tells employers to "seek indemnification from the agency or make mutual any existing indemnification provision". On a client's own paper that advice has already been taken and the direction is reversed: you indemnify them. BountyJobs asks for the same outcome in plainer words, and pairs it with an OFCCP compliance obligation, a second route to the same exposure.

Creep describes how it arrives. In the AlphaNet agreement the indemnity in Section 7 is mutual on its face, a cross-indemnity for either party's acts, and Section 8 then adds a one-way indemnity running from the agency. One instrument, two sections, and a clause you would have called mutual after the first is not mutual by the end of the second.

So read your own for four things: whether you cover claims from your own acts or any claim connected to the hire, and whether there is a cap and whether that cap relates to the fee. The other two: whether it survives termination, and whether compliance obligations elsewhere are wired into it. Then ask your broker whether your insurance would respond to what you promised.

What is actually enforceable, and why that is a state answer

No general US rule exists on whether off-limits, no-hire or candidate-ownership restraints are enforceable. There is no federal test to borrow, and the two decisions this article leans on are a Wisconsin case and a California case that bind Wisconsin and California.

Take the federal question first, because it is the one that has moved most recently: the FTC's 2024 non-compete rule is not in effect. The FTC's own rule page says so plainly: "The Noncompete Rule is not in effect and it is not enforceable". Finalized 7 May 2024, blocked nationwide by a Texas federal district court on 20 August 2024, with the FTC appealing on 18 October 2024 and then moving to dismiss its own appeal on 5 September 2025.

The rule was then formally removed from the Code of Federal Regulations effective 12 February 2026, after, in the Federal Register's words, "federal district courts in three jurisdictions issued opinions" against it. One concluded that the agency "promulgated the Non-Compete Rule in excess of its statutory authority". Not pending, not on appeal, out of the CFR.

In Wisconsin, a no-hire clause between two healthcare staffing companies was held unenforceable. The Wisconsin Supreme Court decided Heyde Companies, Inc. v. Dove Healthcare, LLC, 2002 WI 131, and its published opinion treats the clause as an unreasonable restraint of trade under Wis. Stat. § 103.465, because the employees whose hiring was restricted had not known of or consented to the restriction. Wisconsin law, Wisconsin statute.

In California, employee non-solicitation provisions were held void as applied to recruiters. AMN Healthcare, Inc. v. Aya Healthcare Services, Inc., 28 Cal. App. 5th 923, was decided by California's Court of Appeal in 2018 under Cal. Bus. & Prof. Code § 16600, because the provision restrained the individual defendants "from practicing with [their new employer] their chosen profession". We read it through summaries by Proskauer and Crowell & Moring rather than the opinion, so that fragment is theirs.

Now the limits, because this is where a piece like this overreaches. Both cases concern restraints on hiring or soliciting employees. A client-facing off-limits clause is adjacent to that, not identical, and nothing in this research says § 16600 reaches one the way it reached the non-solicit in AMN Healthcare. Persuasive by analogy at best, binding nowhere outside their own states.

What is genuinely general is thinner: restraints in this family are tested for reasonableness rather than enforced as written, a test Gordon Feinblatt set out in 2003 while recording that most jurisdictions, Maryland included, had not directly ruled at the time. So one question goes to a lawyer in your state: given how my clauses are drafted, and that the people they restrain are not parties to my agreement, what would a court here enforce.

The fallback ladder: what to concede, and what to hold

Negotiating clause by clause without deciding the order first is how agencies give away the expensive things to keep the cheap ones. The ladder below is built from the published checklists, so it is the ask to expect rather than a prediction about your client.

One combination is worth refusing rather than trading: cash back stacked on a long ownership window with no change to the fee, because you have then underwritten one candidate twice.

Then the part no checklist carries. Every redline above is a conversation with a client who is buying, and the account is the stake: reopening signed terms mid-relationship can cost you the relationship, and a solo desk feels that far more than a procurement team feels a redline. The cheapest moment to fix these five clauses is the next agreement, not this one.

Two gaps, stated rather than buried. The American Staffing Association's Search & Placement Model Contract is member-gated, so none of its clause text is in here. And every piece of contract language above comes from two filed agreements, neither randomly sampled: real signed instruments, not a survey of what agreements say. Send us one that contradicts anything here and we will update the page and say what changed.