For nonprofit and public sector teamsTemplates

Run a lessons-learned review, for nonprofit and public sector teams

Turn your notes on a finished project or event into a blame-free review, with what happened, what went well, what got in the way and why, one change to try for each problem, and the questions the notes cannot answer.

This is the version for nonprofit and public sector teams. The steps and checks are the same as in the general template; each prompt carries one added line about your setting, and the examples are from your kind of work.

Sample results: runs and scores on this page are illustrations while testing is set up. How we test

  • Who it is for: Program managers, volunteer coordinators and agency staff who look back on a finished event, program launch or project and report what to change to a team or a board.
  • Time: 20 minutes

Download it, in the format you use

  • Prompt chain

    Copy each step into ChatGPT or Claude, one after another.

  • Checklist

    Follow the steps and check off the checks by hand.

  • Claude skill

    A folder Claude loads, so it follows the steps by itself.

  • n8n file

    Import into n8n and connect your own AI account.

Using a download:How to use this in n8nHow to install a Claude skill

How it runs

Select a part to see its prompt, input, output and check.

The same job, with and without the template

Sample

One plain prompt

Run a lessons-learned review on these notes: what went well, what did not, and what to change next time. (The same goal and notes are pasted below.)

Lessons learned: free tax-help day

What went well

  • 52 households were helped, and 29 of the 31 people who returned the survey said they would come back.
  • 14 volunteers signed up, including 6 certified tax preparers.
  • Library staff said the room was left tidy.

What went wrong

  • Omar gave the wrong opening time in the announcement, so clients waited outside in the cold.
  • Grace should have asked the library about wifi when she booked the room.
  • Two laptops had no charger because nobody checked them properly.
  • The team missed its goal of 80 households.

Recommendations

  • Communicate more clearly and plan ahead.
  • Check every announcement at least 7 days before it goes out.
  • Book rooms 12 weeks ahead and visit them first.
  • Everyone should double-check their tasks.
Problem:It blames Omar and Grace by name, ends on advice with no owner ('communicate more clearly'), and adds two numbers the notes never gave (7 days, 12 weeks).

With this template

The same input, run through the steps.

Summary: The free tax-help day ran on Saturday, February 14 and helped 52 households against a goal of 80. An opening time in the announcement that differed from the plan, a room with only 3 outlets and no wifi in the back half, and two laptops with no charger made the day harder.

What went well

  • 14 volunteers signed up, 6 of them certified tax preparers.
  • 29 of the 31 people who returned the client survey said they would come back.
  • Library staff said the room was left tidy.

What got in the way

Swipe sideways to see all columns

What got in the wayWhy it happenedChange to tryOwner (a role)How we will know
The first clients arrived at 9 am and waited outside in the cold for an hour.The announcement said the doors open at 9 am, and the plan was 10 am to 4 pm.Compare the announcement with the agreed plan before it is sent, and correct any difference.Outreach leadNo clients are waiting outside before the doors open.
Wifi dropped twice in the morning, in a room with only 3 outlets.The room was booked by email without asking what it offers, so the missing wifi and outlets were learned on February 9.Ask the library in writing what the room offers, including outlets and wifi, before booking.Volunteer coordinatorA written list of what the room offers is on file before the day.
Two of the 8 laptops had no charger.Not in the notes.Find out why two laptops had no charger by asking the lender what was handed over on February 13.Not decidedThe reason is written down before the next event.

Questions the notes do not answer

  • Why did the day help 52 households and not 80?
  • How did a 9 am opening time get into the announcement when the plan said 10 am?
  • What did the booking email say about what the room offered?

Check this draft with the people who were there before you share it.

  • Check:Every quote is word for word from the notes: Twelve events quoted word for word from the notes.
  • Check:Each effect is went well or got in the way: Six rows, each Effect one of the two allowed words: three went well and three got in the way.
  • Check:Every change has an owner and a way to know: Three rows, each with a change, an owner role and a way to know filled in.
  • Check:The review ends with the reminder to check it with the people involved: The closing line is there: Check this draft with the people who were there before you share it.
  • Check:Every number in the review is in the notes: Every number in the review (14, 52, 80, 3, 6, 29, 31, 9, 10, 4, 8 and 13) is in the notes.
  • Check:No line puts the fault on a person: No line names a person as a cause or says anyone failed; the rows name roles and what the situation lacked.
  • Check:Each change fits its cause and names an action: Each of the three changes answers the cause in its own row; the laptop row says to find the reason first, as the cause is not in the notes.
  • For your read:The people involved read it before it is shared: Will you ask the outreach lead and the volunteer coordinator whether the rows match what they remember?

Runs and scores

Sample

Claude, Sonnet model

October 4, 2026, easy sample

  • Check:Every quote is word for word from the notes

    Six events quoted word for word from the notes.

  • Check:Each effect is went well or got in the way

    Four rows, each Effect one of the two allowed words: all four went well.

  • Check:Every change has an owner and a way to know

    One row reading None found in every cell, because nothing got in the way.

  • Check:The review ends with the reminder to check it with the people involved

    The closing line is there: Check this draft with the people who were there before you share it.

  • Check:Every number in the review is in the notes

    Every number in the review (17, 64, 60, 4, 6 and 20) is in the notes.

  • Check:No line puts the fault on a person

    No line names a person as the cause or says anyone failed.

  • Check:Each change fits its cause and names an action

    The only change row is None found, so no change was written and none was needed.

  • For your read:The people involved read it before it is shared

    Will you ask the people who were there whether anything went wrong that the notes leave out?

Fixed: Nothing needed fixing: every check passed on the first pass.

Met 8 of 8

1 of these is for your read: the tool hands it to you.

ChatGPT, GPT model

October 5, 2026, tricky sample

  • First pass failed:Every quote is word for word from the notes

    The first version quoted the second account of the message as "Nasir never saw the message", which is not in the notes word for word.

  • Check:Each effect is went well or got in the way

    Four rows, each Effect one of the two allowed words: one went well and three got in the way.

  • Check:Every change has an owner and a way to know

    Three rows, each with a change, an owner role and a way to know filled in.

  • Check:The review ends with the reminder to check it with the people involved

    The closing line is there: Check this draft with the people who were there before you share it.

  • Check:Every number in the review is in the notes

    Every number in the review (1, 2, 10, 11, 29 and 40) is in the notes.

  • Check:No line puts the fault on a person

    No line puts the fault on a person; the two accounts of the van time and the two of the message are each given in full.

  • Check:Each change fits its cause and names an action

    Each of the three changes answers the cause in its own row and says what a person would do.

  • For your read:The people involved read it before it is shared

    Two accounts differ, so will you hear from the driver and the project lead before the review goes to the board?

Fixed: The events step ran again once with the rule to quote word for word, and the row now quotes "says he never saw that message". Nothing else changed.

Met 8 of 8

1 of these is for your read: the tool hands it to you.

n8n, import format 2.20

October 6, 2026, typical sample

  • Check:Every quote is word for word from the notes

    Twelve events quoted word for word from the notes.

  • Check:Each effect is went well or got in the way

    Six rows, each Effect one of the two allowed words: three went well and three got in the way.

  • Check:Every change has an owner and a way to know

    Three rows, each with a change, an owner role and a way to know filled in.

  • Check:The review ends with the reminder to check it with the people involved

    The closing line is there: Check this draft with the people who were there before you share it.

  • First pass failed:Every number in the review is in the notes

    The first report gave the survey result as 94 percent, a number the notes never give.

  • Check:No line puts the fault on a person

    No line names a person as the cause or says anyone failed; the rows name roles and what the situation lacked.

  • Check:Each change fits its cause and names an action

    Each of the three changes answers the cause in its own row and says what a person would do.

  • For your read:The people involved read it before it is shared

    Will you ask the outreach lead and the volunteer coordinator whether the rows match what they remember?

Fixed: The report step ran again once with the rule to add no number the notes do not give, and the line now reads 29 of the 31 people who returned the survey said they would come back.

Met 8 of 8

1 of these is for your read: the tool hands it to you.

The steps, and the check on each

  1. Step 1: Lay out what happened

    List the events in the order they happened, each quoted from the notes, with no judgment yet.

    Hands back: A table of the events in order, each quoted from the notes.

    Check:Every quote is word for word from the notes

    A plain rule. If it fails, run the step again (the n8n file does this for you).

  2. Step 2: Sort what went well from what got in the way

    Pick out the events that mattered, mark each one, and say what in the situation made it happen.

    Hands back: A table that marks each event that mattered as went well or got in the way, with the cause from the notes.

    Check:Each effect is went well or got in the way

    A plain rule. If it fails, run the step again (the n8n file does this for you).

  3. Step 3: Turn each problem into a change

    For every event that got in the way, write one change to try, a role to own it and a way to tell it worked.

    Hands back: A table with one change, one owner role and one way to know for each problem.

    Check:Every change has an owner and a way to know

    A plain rule. If it fails, run the step again (the n8n file does this for you).

  4. Step 4: Write the review

    Put the summary, what went well, the changes and the open questions in one short review a team can read together.

    Hands back: A short review with a summary, what went well, a table of problems and changes, and the open questions.

    Check:The review ends with the reminder to check it with the people involved

    A plain rule. If it fails, run the step again (the n8n file does this for you).

After the last step

  • Check:Every number in the review is in the notes

    A plain rule. If it fails, run the step again (the n8n file does this for you).

  • Check:No line puts the fault on a person

    A second look by the AI. If it fails, run the step again (the n8n file does this for you).

  • Check:Each change fits its cause and names an action

    A second look by the AI. If it fails, run the step again (the n8n file does this for you).

  • Check:The people involved read it before it is shared

    You check this one. If it fails, it is flagged for you.

What you need first

  • What the project set out to do(needed)

    One line on what you were trying to achieve, with any target.

  • What happened(needed)

    Paste your notes, a timeline or the messages, in any order. Include what went well and what did not.

Ways to adapt it

  • The template looks for causes in the situation, such as a missing step or a date that slipped, and never in a person. It cannot know what the notes leave out, so it ends with questions and a reminder to read the review with the people who were there.
  • Paste only what you are allowed to share with an AI tool. If the notes name people you would rather not share, change the names to roles first.
  • To change how many open questions it writes, edit the number in the last step's rules.
  • For a long project, run it one phase at a time and join the tables.
  • A figure that is only in the goal line is not in the notes, so the number check will not accept it in the review. Repeat the figure in your notes if you want it used.
  • A change is a suggestion to try, not a decision. The people involved decide which ones to adopt and who owns them.

Change any step or check in the Editor, then download it again.

Words that fit nonprofit and public sector teams

  • After-action review, debrief and lessons learned name the same thing; use the term your group uses. The review is about what the situation lacked, not who failed.
  • Name roles (the volunteer coordinator, the program manager, the outreach lead) and not people, so a board can read the review without anyone feeling blamed.
  • A change is a proposal. The team or board decides which changes to adopt and who owns each one.
Try it with a sampleSample

Seven dated notes on an event that helped fewer households than its goal; three things got in the way, three went well, and two named people appear in the notes in ways that invite blame.

What the project set out to do
Help 80 households file their taxes for free at a one-day tax-help event on Saturday, February 14
What happened
Review notes for the free tax-help day, written the week after. The goal was to help 80 households file their taxes for free. January 5: plan agreed. Tax-help day on Saturday, February 14, 10 am to 4 pm, in the meeting room at the Eastside library. January 19: Omar, the outreach lead, sent the announcement to the newsletter list. It said the doors open at 9 am. February 2: Grace, the volunteer coordinator, had 14 volunteers signed up, 6 of them certified tax preparers. February 9: the library told Grace the meeting room has only 3 outlets and no wifi in the back half. She had booked the room by email and had not asked about wifi. February 13: the laptops were collected from the lender. Two of the 8 laptops had no charger. February 14: the first clients arrived at 9 am and waited outside in the cold for an hour. Wifi dropped twice in the morning. 5 of the 6 certified preparers came, and 1 was ill. The day helped 52 households. 31 of the 52 returned the client survey, and 29 of those 31 said they would come back. Library staff said the room was left tidy.

The scoring guide

Sample

One criterion for each check, marked met or not met. (The guides in Quality rate a whole piece of work on three levels.) Under each one is how the three sample runs scored.

Download the scoring guide (Markdown)

  • Every quote is word for word from the notes

    A plain rule

    Meets:Every quote in the table is found word for word in the notes.
    Does not meet:A quote has been reworded, or is not found word for word in the notes.
    Claude:Met
    ChatGPT:Met
    n8n:Met

  • Each effect is went well or got in the way

    A plain rule

    Meets:Every row says exactly went well or got in the way in the Effect column.
    Does not meet:A row has another word, such as mixed or neutral.
    Claude:Met
    ChatGPT:Met
    n8n:Met

  • Every change has an owner and a way to know

    A plain rule

    Meets:Every row has a change, an owner role and a way to know filled in, or a single row of None found when nothing got in the way.
    Does not meet:A row has a blank or a dash in the change, the owner role or the way to know.
    Claude:Met
    ChatGPT:Met
    n8n:Met

  • The review ends with the reminder to check it with the people involved

    A plain rule

    Meets:The review ends with the line Check this draft with the people who were there before you share it.
    Does not meet:The line is missing or reworded.
    Claude:Met
    ChatGPT:Met
    n8n:Met

  • Every number in the review is in the notes

    A plain rule

    Meets:Every number in the review is also in the notes.
    Does not meet:A number is in the review that the notes do not give, such as a number of weeks.
    Claude:Met
    ChatGPT:Met
    n8n:Met

  • No line puts the fault on a person

    A second look by the AI

    Meets:No line names a person as the cause, says anyone failed, forgot or should have done something, or guesses at a reason.
    Does not meet:A line puts the fault on a person or says what someone should have done.
    Claude:Met
    ChatGPT:Met
    n8n:Met

  • Each change fits its cause and names an action

    A second look by the AI

    Meets:Each change answers the cause in its own row and says what a person would do.
    Does not meet:A change is advice such as be more careful or plan better, or it does not fit the cause in its row.
    Claude:Met
    ChatGPT:Met
    n8n:Met

  • The people involved read it before it is shared

    You check this one

    Meets:You read the review with the people who were there and changed any line they disputed before sharing it.
    Does not meet:The review is shared without the people involved reading it, or a disputed line is left in.
    Claude:For your read
    ChatGPT:For your read
    n8n:For your read

Claude: met 8 of 8 (including 1 for your read)

ChatGPT: met 8 of 8 (including 1 for your read)

n8n: met 8 of 8 (including 1 for your read)

Learn how it works:Where a person must lookAdapting a template

Practice:Watch a check runScoring guides

Which one fits?

Related templates

All for Nonprofit and public sector

Analyze

Analyze

Write

Last run
October 6, 2026: n8n, import format 2.20 (sample run)
Tools and versions
Claude, Sonnet model; ChatGPT, GPT model; n8n, import format 2.20
What was checked
The files, against each tool’s import format, and 3 sample runs, each check’s finding recorded.
Version
1.0
What changed
First release.

Report a problem with this templateHow we test