AI Toolkit
All projects
Learning & impactadvanced

Programme evaluation report

Turn a term or year of a programme into a short, honest evaluation report with what worked, what did not, and what next.

60-90 minutes
A 3-page evaluation: outcomes vs plan, evidence, participant voice, learning, and 3 recommendations.

When to use this project

Use this at the end of a programme cycle, funded project, or academic term when you need to show funders and internal stakeholders what worked, what didn't, and what changes next.

How to use AI in this project

AI compares planned outcomes against actual results, weaves in participant voice, and drafts honest lessons and recommendations. It works best when you paste in real data (outputs, outcomes, quotes with consent) rather than asking it to imagine what a good report would say.

AI tips for this project

  • Only include participant quotes you have explicit consent to use.
  • Ask the model to name what didn't work — evaluation reports that only celebrate are not credible.
  • Recommendations should be specific enough to assign to a person and a date.
  • Cross-check any statistic in the AI output against your monitoring spreadsheet.

The 4 steps

  1. 01

    Outcomes vs plan

    Compare honestly, no gloss.

    How to use AI here · Paste planned outcomes and the actual results (numbers and stories). Ask AI for a clear "planned vs actual" table, plus commentary that names variances honestly.
    • Off-track outcomes are named clearly
    • Reasons focus on system not blame
  2. 02

    Participant voice

    Young people first, in their own words.

    How to use AI here · Paste consented participant quotes and any survey free-text (anonymised). Ask for 3–4 themes grouped by outcome area, with quotes attributed to a role rather than a name.
    • Every quote is approved for use
    • No identifiable details included
  3. 03

    What we learned

    Learning, not marketing.

    How to use AI here · Give AI the highlights, variances, and participant voice. Ask for 5 lessons — including at least 2 things that didn't work. Reject any lesson too vague to change practice.
    • Anything not evidenced is caveated
    • The surprise is genuine, not manufactured
  4. 04

    Recommendations

    Three, not thirty.

    How to use AI here · Turn each lesson into a recommendation with an owner (role), a timeline, and a resource implication. Add a "what we chose not to do" line to show you thought about trade-offs.
    • Each recommendation is genuinely actionable
    • Resource implications are named
YMCA AI Adoption Toolkit

A practical library, guided context and safeguards so YMCA staff, volunteers and young leaders can use AI safely, confidently and locally.

Aligned with YMCA Vision 2030. Human review, always.

Toolkit

Your work

Safeguarding

    Never enter identifying information about children, health or safeguarding cases into any AI tool. AI drafts, humans decide.

© 2026 YMCA AI Adoption Toolkit

Adapt every prompt to your local YMCA.