Step 1: List the frozen roles and the work they would have owned
For each frozen requisition, write down the work the role was meant to do, from the job description and from the manager who requested it, as a list of task classes rather than a title. A frozen finance assistant is reconciliation, invoice processing, and supplier queries; a frozen support agent is first-line replies in named categories, escalations, and knowledge base upkeep; a frozen SDR is research, qualification, outreach, and CRM hygiene. Add the planned cost of each role from the budget, which is the baseline the plan will be measured against.
The work did not freeze. The list is what still has to happen.
Step 2: Sort each role's work into two columns
Repetitive with a checkable right answer in a system you run, or judgement, relationship, and decisions about people. Use the First-Function Matrix's clarity axis and the Three-List Remit to do it honestly. Most frozen junior roles are mostly the first column, which is why they were junior; most frozen senior roles are mostly the second, and a Digital FTE does not replace them. Be precise: reconciliation is first column, the supplier dispute is second; first-line replies are first, the complaint is second; qualification is first, the reply to the prospect is second.
The myth piece on AI employees replacing people describes why the sort is the honest step and the replacement claim is not.
- Column one: what a Digital FTE proposes with evidence
- Column two: what the existing team decides
- The approval step: minutes a day, between them
Step 3: Give the first column an owner
For each frozen role whose first column is large and whose evidence is clear, hire the matching Digital FTE from the workspace: a Finance Controller FTE for the finance assistant's reconciliation, a Customer Support FTE for the agent's categories, an SDR FTE for the SDR's research and qualification. Write the remit's three lists from the sorted work, with the always-escalate list carrying everything from column two. Record the baseline, which here is the frozen role's planned cost plus the hours the team currently spends covering the gap. Run the first thirty days under the role's playbook.
One role at a time, under the Second-Role Test, even in a freeze; three unsettled roles are worse than one settled one.
Step 4: Redeploy the team to the second column
The existing team was covering the frozen role's work in evenings and gaps; now the first column arrives as proposals they approve in minutes, and the second column, the disputes, the complaints, the prospect conversations, is theirs. Say this early and in each person's terms: the finance manager reviews exceptions instead of matching, the senior agent handles escalations from summaries instead of password resets. Some will like it; some need help with the change, and that is the manager's conversation, early. Route each role's routine approvals to the person closest to the function.
The whitepaper on hybrid team design describes the team shape this produces.
Step 5: Measure against the frozen role's planned cost
After two cycles, compute cost per outcome for each role from its usage, its outcomes with evidence, and the team's approval and review hours, and compare with the frozen role's planned cost for the same outcomes, and with the cost of the team's gap-covering hours before. Apply the Two-Cycle Go/No-Go. Where the role's outcomes meet the targets at a lower cost, the freeze produced a function with an owner; where they do not, the honest finding is that the role is still needed when the freeze lifts, and the record says why.
The metric explainer on cost per outcome gives the arithmetic with the inputs blank.
Step 6: Say plainly what the freeze changed
To the team: which functions now have a Digital FTE owner, what each person's role is now, and that decisions about people were never on the table for the roles. To the board: which frozen hires the record shows are absorbed by a role at a measured cost, which are still needed when the freeze lifts, and which were never candidates because the work was judgement. The honest version of this conversation is allocation, not replacement, and the record makes it defensible. Hiding the absorbed hire inside redeployment language is how trust with a team is lost.
The headcount-versus-AI-budget piece sets the framing for the board line.
What does done look like?
A list of frozen roles with their work sorted into two columns and their planned cost as a baseline; one or more Digital FTEs with remits built from the first column and escalate lists from the second, each settled under its playbook; a team redeployed to judgement with routine approvals routed; cost per outcome per role against the frozen role's planned cost after two cycles, with hours counted; and a written statement, to the team and the board, of which hires the freeze turned into roles, which are still needed, and which were never candidates.
When the freeze lifts, the hires that are still needed are the ones the record could not absorb, and the requisitions are better written for it.