A four-week pilot plan for an AI-in-HR tool
A four-week pilot plan sets one deliverable per week: baseline and scope, guided first use, independent use with light support, and a decision point with evidence. Each week has a named owner and a checkpoint, so the pilot produces a rollout decision rather than open-ended enthusiasm. This structure works for scheduling, screening, or workforce planning tools alike.
- 01Capture the baselineRecord current time-on-task, error rate, or cost for the process the tool will touch.
- 02Fix the scopeName the exact team, volume, and decisions in and out of scope for the pilot.
A pilot without a plan becomes an extended free trial. This checklist gives each of four weeks one deliverable and one owner.
Week 2: guided first use
- 01Run supervised sessionsHave a lead sit with each pilot user for their first real task.
- 02Log friction pointsKeep a running list of confusion points and workarounds as they appear.
Week 3: independent use with light support
- 01Remove direct supervisionMove to a daily async check-in instead of sitting alongside users.
- 02Track usage, not just sentimentCompare actual usage against the baseline volume set in week one.
Week 4: decision point
- 01Compile the evidencePull baseline, usage, and friction logs into a one-page summary.
- 02Make the callDecide go, extend, or stop, with named next steps either way.
Empley's ROI calculators for hours saved and optimised labour cost can turn the week one baseline into a comparable figure once week four usage data is in.
Underlag
- Our assessment
Pilots with a named weekly deliverable are more likely to reach a clear go or no-go decision than open-ended trials.
Common questions
- What if the pilot needs more than four weeks?
- Extend deliberately with a new deliverable and end date rather than letting the pilot run open-ended.
- Who should own the pilot overall?
- One named HR ops or programme lead, not a committee, to keep weekly decisions moving.