What changes when screening moves?
A recruitment agency does two things a company pays for: it finds candidates who were not looking, and it screens the ones who apply. For roles with steady inbound applications, the second is most of the fee, and it is the part a Recruiter Digital FTE on ftes.ai does as evidenced shortlists against a written scorecard, with every decision staying human. The migration moves inbound screening and its administration; it does not move search, persuasion, or the roles where the agency's network is the point.
The comparison piece on a Recruiter FTE against an agency recruiter sets out the division; this guide is about the move.
What should you keep with the agency?
Passive candidate search, senior and specialist roles, roles where inbound is thin, salary benchmarking and market knowledge, and the persuasion of a candidate who has other offers. Keep the agency on those roles under a contract that reflects the narrowed scope; most agencies would rather do search than screen inbound. Keep the hiring managers as the deciders, because they always were. And keep the agency through the first migrated round as a comparison, if the contract allows, by running one role each way.
The head-of-talent persona piece describes what the in-house side owns after the move.
What moves to the FTE, and in what order?
First, record the baseline from the last two rounds of comparable inbound roles: time from application close to shortlist, the share of interviewees who met the hard requirements, and the agency's fee per role. Second, make the data protection decision for the applicant tracking system with the privacy lead, in writing, and connect it with a dedicated account scoped to read and draft. Third, write the scorecard for one inbound role with the hiring manager, versioned, and let the FTE screen the next round while the agency screens a comparable role in parallel. Fourth, at the first review, compare the two, revise the scorecard from the criteria send-backs, and move the next inbound role. Fifth, narrow the agency contract to search and senior roles once two rounds are clean.
The thirty-day recruiting playbook runs the first role step by step.
- Baseline from the last two rounds and the agency's fees
- Data protection decision, then the ATS connection
- One inbound role with the FTE, one with the agency, compared
What breaks in the first round?
The scorecard. The agency screened against criteria that were never written down, and the hiring manager's first written scorecard will produce shortlists the manager disagrees with, because the manager's real criteria were also never written down. The criteria send-backs in the first round are the scorecard being made explicit, and the second version is usually the one that works. The data protection review also breaks in the sense that it may never have been done for the agency either, and doing it for the FTE surfaces questions about the ATS itself. And candidate communication breaks if the templates are not approved before the round; write them first.
The fairness audit, run for the first time on the FTE's shortlists, may reveal a criterion the agency had been applying too, which is the point.
What does the record show after two rounds?
Time to shortlist per role against the baseline and against the agency's parallel role. The hard-requirement match rate for interviews. Communication within window. The criteria approval rate with the hiring managers' reasons, and the scorecard's version history. The fairness audit's findings and the criteria changes they produced, which the agency's process could not have given you because it produced no per-score record. Cost per shortlist from the role's usage against the agency's fee. And a screening process the company can describe in one sentence to a candidate or a regulator.
The metric explainer on time to shortlist describes the companions that keep the speed honest.
What should you refuse during the migration?
Moving senior or thin-inbound roles, where the agency's search is the value. Connecting the ATS before the data protection decision. Letting the FTE communicate a rejection or an offer; both are on the never-automated list. Editing shortlists by hand instead of revising the scorecard, because edits leave no record. Promoting anything but interview scheduling. And ending the agency contract on the strength of one round; two rounds and a fairness audit are the evidence.
The myth piece on AI screening removing bias sets out why the audit is the guarantee, not the tool.