# Ldg Plan The Pilot

> Plans a pilot run of a training programme before rollout, producing the questions the pilot must answer, a pass line written in advance, the pilot group by role and working conditions, an observation sheet, the debrief, a change list and a launch, fix or stop call. Use for "run ldg-plan-the-pilot", "training pilot", "pilot program", "pilot test a training", "pilot session", "test run before rollout", "beta test the course", "what should a pilot measure", part of The Complete Claude Guide for L&D Pack by 90n.

- Skill: `polar-bear-org/ldg-plan-the-pilot` (Agent Skill)
- Install (CLI): `npx get-skill add polar-bear-org/ldg-plan-the-pilot`
- Raw SKILL.md: https://api.skillmd.com/api/skills/polar-bear-org/ldg-plan-the-pilot/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: polar-bear-org (https://skillmd.com/u/polar-bear-org)
- Updated: 2026-11-14
- Page: https://skillmd.com/skills/polar-bear-org/ldg-plan-the-pilot

---


# Plan the Pilot
## When To Use
The first full run is also the launch, and every problem is found by the whole audience. Use this when the programme is near-final, with its session plan, materials and checks, and there is still time to change it. It answers: what must this pilot tell us, what counts as passing, and who makes the call afterwards?

## When Not To Use
If the material is still a draft and you want to see where one learner gets lost, Run the Learner Walkthrough is earlier and cheaper. If the question is whether people use it at work weeks after launch, that is Check Use on the Job.

## Inputs
- The programme as it will run: session plan, facilitator guide, materials, knowledge checks.
- The real audience by role and working conditions (shift, site, experience mix), and the launch date.
If you have none of this, I start from the programme outline, the audience by role and the launch date and mark the output as a first draft.

## Approach
A pilot test before rollout: decide what the pilot must answer and what passing looks like before it runs, use a group like the real audience, and try the near-final version in the real delivery setting. The judgement sits in the pass line, because a pilot with no pass line confirms whatever was built. The failure it prevents: the pilot runs with friendly colleagues from the next desk, overruns by a whole segment, everyone says it was useful, and the same overrun meets the night shift at the launch.

## Workflow
1. Ask at most three questions: what are you least sure about in this programme, who is the real audience and in what conditions do they work, and who makes the launch call and by when?
2. Write three to five questions the pilot must answer: does it fit the time, can people do the practice, do the checks work, does the technology hold, would they use it. Drop any question whose answer would change nothing.
3. Write the pass line per question before the pilot, with thresholds you set, and what happens if it is missed. Record who agreed it and the date, so it cannot move after the results are in.
4. Describe the pilot group by role and working conditions, chosen to resemble the audience, not the friendliest volunteers. Write the note participants get in advance: the programme is being tested, not you, and nothing about you is kept. Then set the scope: which modules or segments run, in the real setting and format, and what is deliberately left out and so stays untested.
5. Build the observation sheet about the programme: planned against actual timings, stuck points, questions asked, check items most people miss (by item, in aggregate), technology faults. No names, no notes on individuals.
6. Plan the debrief for the same day (what worked, what got in the way, what to change), then the change list with an owner and a date, and the call: launch, fix and re-run, or stop.

## Output Format
```markdown
# Pilot Plan
Programme: [name] | Version piloted: [version] | Pilot date: [date] | Launch planned: [date]
## What the pilot must answer
| No. | Question | Pass line (set before the pilot) | If missed | Agreed by, on |
|---|---|---|---|---|
| [1] | [does it fit the time?] | [threshold you set] | [fix / re-run / stop] | [name, date] |
## Pilot group and scope
Roles: [roles] | Working conditions: [shift, site, experience mix] | Number: [set by you] | Where it differs from the audience: [note]
Told in advance, in writing: the programme is being tested, not you; nothing about you is kept.
Tested: [modules or segments, setting, format] | Left out, so untested: [parts]
## Observation sheet (about the programme)
| Segment | Planned time | Actual time | Where people got stuck | Questions asked | Technology faults |
|---|---|---|---|---|---|
| [segment] | [minutes] | [minutes] | [what happened, no names] | [question] | [fault] |

Check items most people missed (by item, group of [minimum size you set] or more): [item] / [item]
## Same-day debrief
With participants, then with the trainer and observer: what worked / what got in the way / what to change / what the pass lines show
## Change list
| Change | From which finding | Owner | By |
|---|---|---|---|
| [change] | [question no. or observation] | [name] | [date] |
## Decision
[Named decider] makes the call (launch, fix and re-run, or stop) against the pass lines by [date].
```

## Done When
- Every question has a pass line with a date that is earlier than the pilot.
- The pilot group is described by role and working conditions, and its differences from the audience are stated.
- Participants have been told in writing that the programme is being tested, not them, and the observation sheet holds no names.
- The change list has owners and dates, and one named person holds the call.

## Quality Bar
- Thresholds, group size and segment lengths are yours; the plan invents no benchmark for a good pilot.
- Results are read against the pass line as written, including the ones that were missed.
- Check results are read by item and in aggregate, and what the pilot did not test is stated, so nobody reads a pass as cover for it.
- The pilot tests the programme, never the participants; the pass line is set before it runs and a named person makes the launch, fix or stop call.

## Next
Run ldg-train-the-trainers (Train the Trainers) once the pilot's changes are in, so the facilitators learn the version that will launch.

## About the makers

This pack is made by 90n ([90n.ai](https://90n.ai)): AI-native ways of working and tools for Product, Marketing and HR teams, built by ex-McKinsey founders. We map where time and money leak, redesign the work around AI and stay until your people run it themselves, and we run our own company on Claude. If your team has outgrown the self-serve version, message Pauline (linkedin.com/in/paulinebertry).

