What are the universal KPIs?
Three KPIs apply to every Digital FTE regardless of role. Outcomes against targets: the explicit contract in the Targets tab, each item with a status and evidence. Approval rate: the share of proposals approved without edits, with send-backs and their reasons, which shows whether the work is right and how much human time it costs. Cost per outcome: tokens consumed by the role over a period divided by the outcomes it delivered.
These three together tell a manager whether the FTE is delivering, whether the team agrees with its work, and what each result costs. Activity metrics such as tasks completed are not on the list, because activity is not value.
What are the role-specific KPIs?
Each role adds a short list tied to the real outcome of its function. A Finance Controller FTE: transactions reconciled, exceptions open at day one of close, days to close. A Customer Support FTE: first-line resolution rate, median first response, escalation quality. A Recruiter FTE: time to shortlist, share of interviewees meeting hard requirements. A Search Visibility FTE: citation presence across sampled prompts, search console impressions and clicks where connected, pages published and their visibility.
Keep each list to three or four. A role with fifteen KPIs has none, because nobody watches fifteen.
- Finance: reconciled, day-one exceptions, days to close
- Support: first-line resolution, first response, escalation quality
- Recruiting: time to shortlist, hard-requirement match rate
- Sales: qualified pipeline created, CRM fields current, selling hours recovered
- Search visibility: citation presence, console metrics, pages published
- Social and content: approved posts published, engagement where measured, fact-gate pass rate
Why are targets the contract?
Targets are the contract because they state, per item, what the Digital FTE is accountable for, with a status and the evidence behind it. They are set at hiring, proposed by the FTE and approved by the manager, and FTE-suggested additions wait for approval. The Targets tab is where the KPIs get their numerator.
A target without evidence is a claim. On ftes.ai each target item shows what the FTE did toward it and what the measurement says, so the review is a reading of the record rather than a conversation about impressions.
Why report the misses?
Report the misses because a scorecard that shows only wins is a marketing document, and because the misses are where the management decisions are. A missed target with the reason attached points to a remit change, a connection problem, or a target that was wrong. A missed target hidden points to nothing.
The same applies to unmeasured items. If a KPI cannot be computed from the data available, the honest report says unmeasured rather than estimating, and the ftes.ai performance view is built on that rule. An estimated KPI is a number nobody can defend.
How should the review work?
Review monthly from the Digital FTE's own first-person report in its Performance tab: what it delivered against targets, what it missed and why, what is unmeasured, and what it recommends. Read the send-back reasons alongside, because they explain the approval rate, and the usage per role, because it explains the cost.
Decide at each review whether the targets are right, whether the remit needs changing, and whether any class deserves promotion on the autonomy ladder. The KPIs are inputs to those three decisions, which is what KPIs are for.
What does a labelled KPI set look like for one role?
As a labelled example for a Customer Support Digital FTE, with the levels left for your own baseline: the universal three are outcomes against targets, with the targets being first-line resolution rate for the categories in its remit and median first response time, each with evidence per ticket in the Targets tab; approval rate, meaning replies approved without edits, with send-back reasons grouped by category; and cost per outcome, meaning tokens consumed for the role divided by first-line tickets resolved in the period.
The role-specific list adds escalation quality, judged by whether escalations arrived with the four-part summary, and knowledge base gap closure, meaning proposed articles approved and the resolution rate of their categories afterwards. Every item is reported with its misses and its reasons, and anything the platform cannot measure for the period, such as satisfaction where no survey is connected, is marked unmeasured rather than estimated. Five items, read from the record in thirty minutes a month, are enough to decide the three things the review exists to decide.