# Impact Programme Shortfall Note

> Drafts a plain account of a programme that missed its measure, producing what was expected beside what the evidence shows, the likely reasons sorted into five kinds, what stops now and what was learned for the next request. Use for "run impact-programme-shortfall-note", "programme shortfall note", "the training did not work", "programme missed its target", "how to report a failed training", "tell the sponsor it did not work", "lessons from a programme that missed", "blameless programme review", part of the Claude Guide for Training Impact Pack by 90n.

- Skill: `polar-bear-org/impact-programme-shortfall-note` (Agent Skill)
- Install (CLI): `npx get-skill add polar-bear-org/impact-programme-shortfall-note`
- Raw SKILL.md: https://api.skillmd.com/api/skills/polar-bear-org/impact-programme-shortfall-note/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Docs & Writing
- Author: polar-bear-org (https://skillmd.com/u/polar-bear-org)
- Updated: 2026-11-14
- Page: https://skillmd.com/skills/polar-bear-org/impact-programme-shortfall-note

---


# Programme Shortfall Note

## When To Use
The programme did not move the measure, and you would rather say so yourself than be found out. Use this in the week the evidence is clear, when the verdict against the agreement is "did not" or a "partly" that nobody believes. It answers: what did we expect, what do we actually know, why did it most likely fall short, and what changes because of it?

## When Not To Use
While the programme is still running and can be turned, send Programme Early Warning Note instead. For any verdict on one page, including a good one, use Sponsor Impact One-Pager; this note is the fuller account of a miss. To state what was decided as a result, use Programme Decision Record.

## Inputs
- The Success Measures Agreement: the result expected and what "worked", "partly" and "did not" were to mean.
- The results table with claim strengths, and whatever else exists: comment themes, case write-ups, the observation tally, the contribution analysis, and what you already plan to stop.
If you have none of this, I start from what was expected and the one figure you have, write "no agreement on file: first draft", and mark every reason "assumed".

## Approach
The four review questions (what was meant to happen, what happened, why the difference, what we do next), a common review practice, used here without blame. Around them sits one rule from the UK Government Analysis Function's "Communicating quality, uncertainty and change": state what can and cannot be concluded, in place of vague cautions. The judgement is that a miss explained by the L&D lead is a finding, while a miss discovered by finance is a scandal. The failure it prevents: a soft paragraph about "mixed results and valuable learnings", after which the same programme is requested again and built again.

## Workflow
1. Ask at most three questions: what the agreement said the programme would change, what evidence you hold and who vouches for it, and who receives this note and by when.
2. Set expected beside observed. The expected result is quoted from the agreement; the observed result is copied from the results table with its claim strength. A measure nobody captured is written "unknown", not "no change".
3. Sort the likely reasons into five kinds: wrong problem (training was not the fix), design, delivery, work environment (no time, tool or support to use it), measurement (the change may exist but was not captured). Each reason carries its evidence or the word "assumed". An empty kind is left empty.
4. Run the blameless pass. Any reason that points at a trainer, a manager, a team or the learners is rewritten as the condition behind it: "managers did not follow up" becomes "no time was set aside for follow-up, and nobody owned it".
5. Write two short lists: what we can conclude, and what we cannot. "The measure did not move" and "the programme had no effect" are different sentences; only the evidence says which one you may write.
6. Write what stops now and what that frees (days, budget, people's time, in your figures or blank), then one or two changes to how the next request is questioned or how measures are agreed. The note is a draft for a named person to send; I do not send it, and I do not soften the first line.

## Output Format
```markdown
# Programme Shortfall Note
Programme: [programme] | To: [sponsor role] | From: [named sender] | Date: [date]
In one line: [the programme did not move measure X; what stops; what is asked]
## Expected and observed
| Agreed measure | Expected (from the agreement) | Observed | Claim strength | Source and voucher |
|---|---|---|---|---|
| [measure] | [quoted] | [your figure, or "unknown"] | [shown, suggested, unknown] | [source, name] |
## Likely reasons
| Kind | Reason, stated as a condition | Evidence or "assumed" |
|---|---|---|
| Wrong problem | [reason] | [evidence] |
| Design | [reason] | [evidence] |
| Delivery | [reason] | [evidence] |
| Work environment | [reason] | [evidence] |
| Measurement | [reason] | [evidence] |
## What we can and cannot conclude
- We can conclude: [statement the evidence supports]
- We cannot conclude: [statement it does not support, and what it would take to know]
## What stops now, and what we learned
- Stops: [activity] from [date], which frees [your figure, or blank]
- Learned for the next request: [change to how requests are questioned or measures agreed]
## Decision
[Sponsor] confirms what stops and whether anything replaces it, by [date]; [named sender] sends this note by [date].
```

## Done When
- The first line says the programme fell short, with no softening clause in front of it; expected figures are quoted from the agreement, and observed figures have a source and a voucher.
- Every reason sits in one of the five kinds with its evidence or "assumed", and both lists exist: what can be concluded and what cannot.
- No sentence names or points at a person, a team or a cohort as the reason.

## Quality Bar
- A reason with no evidence stays "assumed" and is never promoted to a finding by confident wording.
- "Measurement" is a real possibility, not an escape: it needs a line on what was not captured and how it could be.
- If you ask for a version that assigns the miss to named people, I decline and offer the conditions behind it.
- The note says plainly what did not work and why, from the evidence only; it is about the programme and never about the people in it.

## Next
Run impact-needs-assessment-update (Training Needs Assessment Update) to carry what was learned into next period's needs.

## 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).

