Book the fee, then keep it through the guarantee window

A perm search takes weeks. Sourcing, screening, submittal, first interview, second interview, offer, acceptance, notice period, start. By the time a desk learns that a decision made in week one was wrong, it is week nine and the quarter is already shaped. This is why gut-feel forecasting fails on perm desks specifically: the loop is too slow for intuition to calibrate against.

The only reliable forecast is arithmetic on your own conversion rates. If your desk converts a known proportion of first interviews into offers, and a known proportion of offers into starts, then the number of candidates sitting at each stage today implies a range for next quarter. Surhires computes those conversion rates per recruiter, per client and per role type from your actual stage history, and shows the forecast as a range with the assumptions visible rather than as a single confident number.

Where there is not yet enough history for the numbers to mean anything, the forecast says so. A conversion rate computed from six placements is not a forecast, and presenting it as one is worse than presenting nothing.

The riskiest period of a permanent placement begins the moment it looks safe. The candidate accepts, everybody stops working the deal, and then they hand in notice to an employer who would rather keep them. Four weeks of notice is four weeks for a counter-offer, a competing process to conclude, or a slow drift back toward the familiar.

Surhires treats that period as an active stage rather than a waiting room. The placement carries the acceptance date, the notice period, the agreed start date and the resignation date once it happens. Structured check-ins run through the gap on a cadence you set, and the Placement Risk Agent watches the signals that precede a drop-out: notice not yet handed in when it should have been, a check-in unanswered, a start date pushed, references stalling.

None of this makes a candidate honest. It makes the desk aware in week two of the notice period rather than on the morning the candidate does not appear, which is the difference between placing a backup and refunding a fee.

Most perm terms carry a guarantee: if the placement leaves inside a stated period, the client gets a free replacement or a sliding rebate. The commercial consequence is that a fee invoiced in March is not fully earned until the window closes in June, and a desk that treats invoiced revenue as banked revenue is overstating itself by whatever is still at risk.

Every placement carries its guarantee terms: the window length, whether the remedy is a rebate or a replacement, and the sliding scale where one applies. Placements inside the window stay visible on the pipeline after they close, with the days remaining and the exposure attached. If a placement fails inside the window, it moves to a drop-out state, the obligation is recorded against the client, and the replacement search is linked back to the original so the rebate or replacement can be tracked to its conclusion.

That linkage matters at review time. A desk with a high gross fee and a high rebate rate is a different business from one with the same gross and none, and only one of them is worth expanding.

Because feedback arrives so late on a perm desk, the only way to learn from it is to record it in a form that can be counted. Surhires requires a coded disposition at every drop-out: failed technical screen, salary above budget, a better-qualified candidate was selected, candidate declined, candidate accepted a counter-offer, position withdrawn. A free-text note can sit alongside it, but the code is mandatory.

The patterns that emerge are usually about the brief rather than the candidates. A client rejecting on salary in most cases is telling you the band is wrong, and that is a conversation to have in week two rather than week seven. A desk losing candidates to counter-offers repeatedly has a qualification problem at the start of the process, not a closing problem at the end.

A growing number of US states and other jurisdictions require a salary range on a posted role, with different rules about what has to be disclosed and when. For a perm desk advertising across several states, that is a per-posting question rather than a policy question.

Surhires can check a job posting against the salary-range disclosure rules of the places it is being published to and flag it before it goes out. This is a capability, not a compliance guarantee: the product surfaces the rule that appears to apply and shows what the posting is missing. Whether a given advert satisfies the law where it runs remains a decision for you and your counsel.

The Placement Pipeline Command Centre shows offers live with acceptance risk, placements closed, revenue booked against revenue invoiced, and at-risk placements inside the guarantee window. On a perm desk that last panel is the one that turns a management report into a working tool.

Recruiter performance follows the same logic. Placements, average fee, interview-to-offer conversion, offer-to-start conversion, drop-out rate inside guarantee, and net fee after rebates. A desk optimised on gross placements will eventually book fees it has to return, and the only defence is measuring the number that survives the guarantee window.

Surhires is recruitment software. It is not an HRIS, a payroll system, a background-check provider or an assessment platform. It raises an offer document and a placement invoice; it does not onboard the hire, run their probation review, or hold their employment record after day one.

It also does not decide anything. AI agents draft, score, summarise and flag, and a recruiter takes every decision that matters. Where scoring is used and an automated employment decision tool is in scope, the product records which candidates were scored and can export the data an independent bias audit needs. Commissioning that audit and issuing candidate notice remain yours.