Evidence Review Meeting
When To Use
Evaluation reports pile up, and next year's plan looks exactly like this year's. Use this once a period, when the one-pagers, results tables and shortfall notes are in and nobody has sat down to ask what they change. It answers one question: for each programme running now, do we continue, change, stop or scale, and who decides by when?
When Not To Use
If there is one programme and its decision is already taken, write it down with Programme Decision Record. To rank what runs next period against capacity, use L&D Priority Plan: this meeting settles the fate of what is running now. If the reports have not been tested yet, run Overclaim Check before you put them in front of a room.
Inputs
- The report for each programme: the one-pager, the results table with its claim strengths, and any shortfall note or early warning already sent. Paste them, or keep them in Projects (beta, on select plans) so the pre-read draws on one set of files.
- The Success Measures Agreement for each programme, or "no agreement on file".
- Who attends (roles), who chairs, and who has the authority to decide on each programme.
If you have none of this, I start from your list of programmes and what each was meant to change, and mark the output as a first draft.
Approach
Four review questions, asked without blame: what was meant to happen, what happened, why the difference, what do we do next. The fourth is where most reviews stop short, so the meeting is built around "act on findings", the closing step of the CDC Program Evaluation Framework, 2024. The failure it prevents: an hour spent touring six reports, every programme leaves with "going well", the one nobody measured carries on by default, and the plan is copied across with new dates.
Workflow
- Ask at most three questions: which programmes are in scope and who can decide on each, how many minutes you give each block of the hour, and which reports you can share (and which programmes have none).
- Write the pre-read, one line per programme in three parts: what was meant to happen, quoted from the agreement; what happened, copied from the results table with its claim strength (shown, suggested, unknown); why the difference, from the reports or "not yet explained". I copy what the reports say and pass no verdict on a programme.
- Draft a proposal per programme: continue, change, stop or scale, with the evidence behind it and its strength. Unknown evidence leads to "decide what to collect", never to "continue by default". Each proposal is marked as a starting point for the room.
- Order the agenda by what needs deciding. Programmes with a decision to take go first, the hardest one while the room is fresh. Programmes with nothing to decide are read in the pre-read and not discussed. Then fit the hour to your minutes per block: the ground rule and questions on the pre-read, the decisions, the items left open, a read-back of owners and dates. If the decisions do not fit, I say which programmes move to a second meeting instead of squeezing them.
- Print the ground rule on the agenda: this meeting reviews programmes, not people, and no one's performance is discussed. A reason that points at a trainer, a manager or a cohort's ability is restated as the condition behind it.
- Build the decisions table for the chair to fill in the room: programme, decision, owner, date. An item with no decision gets an owner and a date for deciding. A person chairs and records; I never fill the decision column.
Output Format
# Evidence Review Meeting
Period: [period] | Date: [date] | Chair: [role] | Attending: [roles]
Ground rule: this meeting reviews programmes, not people. No one's performance is discussed.
## Agenda (one hour)
| Block | Minutes | What happens |
|---|---|---|
| Ground rule and questions on the pre-read | [you set] | [chair] |
| Programmes needing a decision | [you set] | [programme, programme] |
| Items left open, then read-back | [you set] | [owner and date for deciding; decisions, owners, dates read aloud] |
## Pre-read
| Programme | Meant to happen | What happened | Claim strength | Why the difference | Source |
|---|---|---|---|---|---|
| [programme] | [from the agreement] | [from the results table] | [shown, suggested, unknown] | [from the reports, or "not yet explained"] | [report, voucher] |
## Proposals
| Programme | Proposal | Evidence behind it | Strength | If unknown, what to collect |
|---|---|---|---|---|
| [programme] | [continue, change, stop, scale] | [evidence] | [strength] | [measure, owner] |
Read, not discussed: [programmes with nothing to decide this period]
## Decision
| Programme | Decision | Owner | Date |
|---|---|---|---|
| [programme] | [filled in the room by the chair, or "to decide"] | [named owner] | [date] |
Done When
- Every programme in scope has a pre-read line with all three parts, a claim strength and its source, or a part marked "not yet explained".
- No programme with unknown evidence is proposed as "continue" without a line on what to collect.
- The ground rule is printed, the minutes add up to the hour, and every row of the decisions table has an owner and a date, even when the decision is still open.
Quality Bar
- Figures in the pre-read are copied from the reports, to the digit, with nothing added from memory.
- A programme with no report is listed as "no evidence on file", never left off the page.
- No result is attributed to a trainer, a manager or a cohort's ability.
- The pre-read says only what the evidence says, with its strength; the meeting reviews programmes, never people, and named owners decide.
Next
Run impact-programme-decision-record (Programme Decision Record) to write down each decision with its reasons.
About the makers
This pack is made by 90n (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).
1---2name: impact-evidence-review-meeting3description: Prepares a one-hour review of your evaluation evidence, producing an agenda, a pre-read of one line per programme, a proposal per programme and a decisions table with owners and dates. Use for "run impact-evidence-review-meeting", "evidence review meeting", "review our training evaluations", "what do we do with the evaluation reports", "continue change stop or scale", "programme review agenda", "training portfolio review", "act on evaluation findings", part of the Claude Guide for Training Impact Pack by 90n.4metadata:5 source: https://skillmd.com/skills/polar-bear-org/impact-evidence-review-meeting6---78# Evidence Review Meeting910## When To Use11Evaluation reports pile up, and next year's plan looks exactly like this year's. Use this once a period, when the one-pagers, results tables and shortfall notes are in and nobody has sat down to ask what they change. It answers one question: for each programme running now, do we continue, change, stop or scale, and who decides by when?1213## When Not To Use14If there is one programme and its decision is already taken, write it down with Programme Decision Record. To rank what runs next period against capacity, use L&D Priority Plan: this meeting settles the fate of what is running now. If the reports have not been tested yet, run Overclaim Check before you put them in front of a room.1516## Inputs17- The report for each programme: the one-pager, the results table with its claim strengths, and any shortfall note or early warning already sent. Paste them, or keep them in Projects (beta, on select plans) so the pre-read draws on one set of files.18- The Success Measures Agreement for each programme, or "no agreement on file".19- Who attends (roles), who chairs, and who has the authority to decide on each programme.20If you have none of this, I start from your list of programmes and what each was meant to change, and mark the output as a first draft.2122## Approach23Four review questions, asked without blame: what was meant to happen, what happened, why the difference, what do we do next. The fourth is where most reviews stop short, so the meeting is built around "act on findings", the closing step of the CDC Program Evaluation Framework, 2024. The failure it prevents: an hour spent touring six reports, every programme leaves with "going well", the one nobody measured carries on by default, and the plan is copied across with new dates.2425## Workflow261. Ask at most three questions: which programmes are in scope and who can decide on each, how many minutes you give each block of the hour, and which reports you can share (and which programmes have none).272. Write the pre-read, one line per programme in three parts: what was meant to happen, quoted from the agreement; what happened, copied from the results table with its claim strength (shown, suggested, unknown); why the difference, from the reports or "not yet explained". I copy what the reports say and pass no verdict on a programme.283. Draft a proposal per programme: continue, change, stop or scale, with the evidence behind it and its strength. Unknown evidence leads to "decide what to collect", never to "continue by default". Each proposal is marked as a starting point for the room.294. Order the agenda by what needs deciding. Programmes with a decision to take go first, the hardest one while the room is fresh. Programmes with nothing to decide are read in the pre-read and not discussed. Then fit the hour to your minutes per block: the ground rule and questions on the pre-read, the decisions, the items left open, a read-back of owners and dates. If the decisions do not fit, I say which programmes move to a second meeting instead of squeezing them.305. Print the ground rule on the agenda: this meeting reviews programmes, not people, and no one's performance is discussed. A reason that points at a trainer, a manager or a cohort's ability is restated as the condition behind it.316. Build the decisions table for the chair to fill in the room: programme, decision, owner, date. An item with no decision gets an owner and a date for deciding. A person chairs and records; I never fill the decision column.3233## Output Format34```markdown35# Evidence Review Meeting36Period: [period] | Date: [date] | Chair: [role] | Attending: [roles]37Ground rule: this meeting reviews programmes, not people. No one's performance is discussed.38## Agenda (one hour)39| Block | Minutes | What happens |40|---|---|---|41| Ground rule and questions on the pre-read | [you set] | [chair] |42| Programmes needing a decision | [you set] | [programme, programme] |43| Items left open, then read-back | [you set] | [owner and date for deciding; decisions, owners, dates read aloud] |44## Pre-read45| Programme | Meant to happen | What happened | Claim strength | Why the difference | Source |46|---|---|---|---|---|---|47| [programme] | [from the agreement] | [from the results table] | [shown, suggested, unknown] | [from the reports, or "not yet explained"] | [report, voucher] |48## Proposals49| Programme | Proposal | Evidence behind it | Strength | If unknown, what to collect |50|---|---|---|---|---|51| [programme] | [continue, change, stop, scale] | [evidence] | [strength] | [measure, owner] |5253Read, not discussed: [programmes with nothing to decide this period]54## Decision55| Programme | Decision | Owner | Date |56|---|---|---|---|57| [programme] | [filled in the room by the chair, or "to decide"] | [named owner] | [date] |58```5960## Done When61- Every programme in scope has a pre-read line with all three parts, a claim strength and its source, or a part marked "not yet explained".62- No programme with unknown evidence is proposed as "continue" without a line on what to collect.63- The ground rule is printed, the minutes add up to the hour, and every row of the decisions table has an owner and a date, even when the decision is still open.6465## Quality Bar66- Figures in the pre-read are copied from the reports, to the digit, with nothing added from memory.67- A programme with no report is listed as "no evidence on file", never left off the page.68- No result is attributed to a trainer, a manager or a cohort's ability.69- The pre-read says only what the evidence says, with its strength; the meeting reviews programmes, never people, and named owners decide.7071## Next72Run impact-programme-decision-record (Programme Decision Record) to write down each decision with its reasons.7374## About the makers7576This 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).