What decision does the Two-Cycle Go/No-Go frame?
Whether a Digital FTE pilot continues, and in what form. Pilots are usually judged too early, on the first cycle, which contains the setup week, the close-reading weeks, and the calibration, and are judged on impressions rather than the record. The frame fixes the timing at the second monthly review and fixes the evidence at five things a manager can open. The verdict is then defensible to a CFO, a board, or a sceptical team, because every line of it traces to the workspace or to the baseline sheet the team recorded before the pilot began.
The frame applies to any role, any vertical, and any size of pilot, including one role in one function.
What are the five criteria?
Outcomes: did the role meet its targets in the Targets tab over the second cycle, with the evidence attached, and were the targets the ones agreed at the first review? Cost: what is cost per outcome for the second cycle, from the role's usage at the plan rate divided by the outcomes, against the baseline cost of the same outcome recorded before the pilot, with human hours added? Time: did the manager's approval and review minutes in the second cycle fall against the first, and against the baseline hours the function consumed? Record: did at least one routine class meet the Promotion Standard, or is there a specific, named reason it did not? Incidents: were there any escalations that failed the summary standard, any send-backs for a substantive reason on a promoted class, or any incident, and is each resolved with a remit or memory change recorded?
Each criterion is a yes, a no, or a no with a named fix.
- Outcomes: targets met with evidence
- Cost: per outcome against the baseline, hours included
- Time: manager minutes falling
- Record: a class promoted, or a named reason
- Incidents: none open, or each resolved with a recorded change
How does the worked example read, with the inputs left blank?
Role: [role], hired [date], reviewed [date] and [date]. Outcomes: targets agreed at the first review were [target] and [target]; in the second cycle the record shows [result] and [result], with evidence in the Targets tab. Cost: usage in the second cycle was [tokens] at [rate], outcomes were [count], so cost per outcome was [figure]; baseline cost per outcome from the sheet was [hours] at [rate] plus [external], so [figure]; manager hours in the cycle were [hours] at [rate], added. Time: approval minutes per day fell from [minutes] in cycle one to [minutes] in cycle two; the function's baseline was [hours] per week. Record: [class] was promoted on [date] under the Promotion Standard; or, [class] was not promoted because [reason]. Incidents: [count], each resolved by [change] recorded on [date]. Verdict: [continue and add, continue and fix, stop], because [criteria].
If the first cycle's numbers are used anywhere, they are labelled first-cycle and are not the comparison.
What are the three verdicts?
Continue and add: all five criteria are yes; the role continues, its block is sized from usage, and the Second-Role Test is run for the next function. Continue and fix: outcomes or cost are close but one criterion is a no with a named fix, a remit too wide, a memory gap, an unsettled approver, a target set wrong; the role continues for a third cycle with the fix recorded, and the frame is re-run. Stop: outcomes or cost are clearly short with no named fix, or an incident is unresolved, or the function turned out to be judgement work; the role is paused, the audit trail and memory are exported, the credentials are revoked, and the note records why, so the next attempt, if any, starts from the record.
Continue-and-fix is the most common honest verdict, and it is not a failure; it is the frame naming the change.
Why the second review and not the first?
Because the first cycle is the calibration period and measures the setup, not the function. Its approval rate is low because every proposal was read closely; its cost per outcome is high because setup hours are in it and no class had earned autonomy; its send-backs are about facts the memory lacked. Judging on it produces a no-go for roles that would have worked and, occasionally, a go for roles that were only impressive. The second cycle runs with a tuned remit, a corrected memory, and a manager who has the habit, and its numbers are the ones a CFO can sign.
The myth piece on working perfectly from day one describes what the first cycle looks like and why it is misread.
How is the frame applied?
Before the pilot: record the baseline with the template, and agree with the sponsor that the verdict comes at the second review, in writing. At the first review: set targets, tune the remit, promote nothing or one class, and label the cycle's numbers first-cycle in the note. At the second review: fill the five criteria from the Targets, Performance, and Approvals tabs and the baseline sheet, write the verdict with its evidence in the review note, and only then tell the sponsor. For continue-and-fix, name the fix and the date of the third review. For stop, run the offboarding checklist the same week.
The whitepaper on the economics of digital labor gives the cost arithmetic in full, and the monthly review note template gives the fields.