Mission goal: understand what the AI-enhanced version of your role is and why it is emerging.
Step 1 of 5Where are you starting from?
Choose your starting point+10 XP
Where are you today with AI in your monitoring work?
Show answer
Every starting point is valid. Follow the mission suggested for your situation.
Step 2 of 5Meet your new role
Today's role
M&E Officer
AI-enhanced role
AI-Assisted M&E Officer
Level
Officer · 1–3 years
How the role uses AI
Uses AI tools inside monitoring work and checks every output before it is used
"AI-Assisted M&E Officer" is a proposed career title from the EvalCommunity framework, not a standardised occupational classification or a guaranteed vacancy.
Step 3 of 5What changes most in your job?
Your guess+10 XP
When AI drafts more of your routine work, where does your value go?
Show answer
In the profile, first data checks, standard tables and routine report text are now drafted by AI tools or simple automations. The officer's value shifts to checking those outputs, following up flagged records with the field and knowing the data well enough to notice when something looks wrong.
Step 4 of 5AI-Assisted, and still in charge
The goal is not to hand your work to AI. It is to become a stronger M&E officer who uses AI and stays in charge of it. Here are the five questions, asked about your own work:
1QualityDo the AI-written formulas and drafts you use make your data more accurate, and have you tested them?Test every formula on rows where you already know the answer before you use it on real submissions.
2AccountabilityWhen a figure in your monthly report is wrong, can you show where it came from?Every figure ticked against the IPTT, with your name on the check.
3IndependenceDo you still notice when something looks wrong, before the AI flags it?You know your sites well enough to question a flag, or a figure nobody flagged.
4EthicsDo names, phone numbers and sensitive answers stay out of AI tools?You remove personal data before you paste anything, and follow up with people respectfully.
5Professional developmentAre you learning the Excel and data skills behind what the AI does, or relying on what you can't check?You learn the formulas the AI writes, so you can tell when they are wrong.
The five questions as the series states them
Quality: Does AI improve the quality and credibility of evaluation findings?
Accountability: Who is responsible when AI produces an incorrect interpretation?
Independence: Are evaluators maintaining their ability to think critically and make independent judgments?
Ethics: Are confidentiality, human rights and the interests of affected populations protected?
Professional development: Are we developing stronger evaluators who understand AI, or creating professionals who increasingly depend on it?
Step 5 of 5Your goal for this tutorial
By the end, you will know how to use AI in six of your real monitoring tasks and check every output, which skills to build, and you will have a 12-week plan.
Next: see which of your tasks change, and how.
End of Mission 01
MISSION02
Traditional role vs AI-enhanced role
Before and after · Badge: Role Shifter
8 challenges
Mission goal: see which responsibilities become AI-assisted, which need new skills and which stay human.
Step 1 of 4Meet your case study
Case study · Learning and Livelihoods Project Illustrative
A fictional project you will follow through missions 2, 3, 4, 8 and 9.
The project
Teacher training, school visits and cash transfers in four districts (North, South, East and West), run with three partners.
The data
About 350 KoboToolbox submissions a day in collection weeks, a monthly IPTT with 18 indicators, partner returns, visit notes and a Power BI dashboard.
The tools
KoboToolbox, ODK, Excel, Power BI, Microsoft 365 Copilot, ChatGPT, Claude and SharePoint.
Your objective
Produce accurate monthly monitoring data and reports with help from AI, and check every output before it is shared.
Step 2 of 4Flip your tasks: today and with AI
AI drafts more of the routine work. You check it and follow up with the field. Tap each card to see how it changes.
Step 3 of 4Mini-game: sort the work
For each task, choose: AI-assisted, a new skill for you, or stays human.
Sort it+10 XP
Writing formulas that flag duplicate beneficiary IDs
Show answer
AI-assisted. AI writes them fast; you test them on rows where you know the answer.
Sort it+10 XP
First draft of the monthly monitoring summary
Show answer
AI-assisted. Copilot drafts from the IPTT; you check each figure (the profile's example).
Sort it+10 XP
Filling the visit template from anonymised notes
Show answer
AI-assisted. AI fills the template; you correct counts and add follow-up actions.
Sort it+10 XP
Writing prompts with clear instructions, examples and an output format
Show answer
New skill. Prompt design is one of the profile's learning priorities.
Sort it+10 XP
Checking AI-drafted figures against the IPTT
Show answer
New skill. Validating AI outputs: the officer is the first line of defence against AI errors entering the IPTT.
Sort it+10 XP
Deciding whether a flagged value is an entry error or real
Show answer
Human. Listed in the profile as human judgement.
Sort it+10 XP
Following up sensitively with enumerators and community members
Show answer
Human. Listed in the profile as human judgement.
Sort it+10 XP
Deciding what personal data to remove before using AI
Show answer
Human. Listed in the profile as human judgement.
Step 4 of 4Your turn: two minutes
Think about your last two weeks of work. Which tasks would go in each pile?
Think about it for a moment, then reveal.
A quick way to do it: three columns on a sheet of paper, AI-assisted, new skill, stays human. Keep the list: you will use it in mission 4 (to pick a task to try) and mission 6 (to rate your gaps).
End of Mission 02
MISSION03
What AI actually changes in your daily work
Checks, risks and red flags · Badge: Red Flag Spotter
2 challenges
Mission goal: learn the three checks that keep AI output safe to use, and practise spotting problems in an AI draft.
Step 1 of 4What can AI really do for you?
Before you look: which of your tasks could AI help with, and which must stay with you?
Think about it for a moment, then reveal.
AI can assist with
Formulas for data checks
First drafts of monthly reports
Filling visit templates from notes
Translating short responses
Can be partly automated
Daily duplicate and range flags
Dashboard refreshes
Lists of unusual changes by site
New responsibilities
Testing formulas on rows you know
Checking every AI figure against the IPTT
Removing personal data before using AI
Recording an explanation for each flag
Stays with you
Entry error, or a real situation in the field?
Sensitive follow-up with people
What personal data to remove
Figures match source before submission
Step 2 of 4The three checks you must always run
AI output is only usable when you check it. Here are the checks, one at a time.
Test it. Try every AI-written formula on rows where you already know the answer, before you use it on real data.
Trace it. Check every figure in an AI draft against the IPTT or the source register.
Protect it. Remove names, phone numbers and other personal data before anything goes into an AI tool.
Step 3 of 4Mini-game: spot the red flags
Case study · Learning and Livelihoods Project Illustrative
Copilot drafted the training paragraph for the monthly report from the partner returns. Here is the source and the draft.
Click every sentence you would not send to your manager.
What the partner returns show for March
Partner A: people trained
120
Partner B: people trained
95
Partner C
Return not received by the deadline
Last month (all partners)
198 people trained
Red flags found: 0 of 3+30 XP
Show answer
Three sentences should not be sent: the 80 people for Partner C (its return has not arrived), the "sharp rise thanks to the new outreach approach" (no cause in the data), and "all returns received on time" (Partner C is late).
Step 4 of 4Mission check
Your call+20 XP
Copilot's draft says all 18 indicators are "on track". What do you do before you send the report?
Show answer
Check each of the 18 indicators against the IPTT, then correct the draft.
End of Mission 03
MISSION04
Your daily tasks, step by step with AI
Six practical guides · Badge: Workflow Builder
6 challenges
Mission goal: learn how to use AI in six of your real monitoring tasks, and check every output, step by step.
Case study · Learning and Livelihoods Project Illustrative
Every worked example below uses this project's data, so you can follow one system from data collection to the monthly report.
Six real tasks from your working week. Pick one and go through it in six short steps: a task prompt, then a QA prompt that checks the AI's work.
YouAIYour teamAutomated shows who does each step. Examples are illustrative.
Choose a task
Guide 1 of 6 From the career profile
Check yesterday's submissions with AI-written formulas
From the profile: each morning the officer exports new KoboToolbox submissions and asks ChatGPT to write formulas flagging duplicate beneficiary IDs and ages over 100; flagged rows go back to the team leader for correction (350 submissions in the daily task).
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
You scroll through the export looking for problems.
Errors are found at the end of the month, when enumerators have moved on.
Values are sometimes corrected in the office without asking the field.
How it works with AI
ChatGPT writes the formulas; it never sees the data.
You test the formulas on rows where you know the answer.
Flagged rows go back to the team leader the same day.
Step 2 of 6Step by step
YouExport yesterday's submissionsKoboToolbox
Download the new submissions as an Excel file (350 in the profile's example).
YouGive the AI column names only
The AI does not need the data to write a formula. Share the column letters and names, never the rows.
AIAsk for the formulasChatGPT
One formula per check, with a one-sentence explanation of each (prompt below).
YouTest on rows you knowExcel
Add two test rows: a duplicate ID and an age of 104. Both must be flagged before you trust the formulas.
AutomatedApply to all submissionsExcel
Copy the formulas down and filter the flagged rows.
Your teamSend the flags to the team leader
The enumerator checks each flagged record. Record the reply and the decision. Never change a value without a reply.
Step 3 of 6Worked example
What you give the AI Illustrative
Column
Name
A
Submission ID
B
Date
C
Enumerator
D
Beneficiary ID
E
Age
F
Site
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
I have an Excel sheet of survey submissions with these columns: [paste column letters and names].
Data starts in row 2 and there are up to 400 rows.
Write two formulas for new columns:
1) flag a beneficiary ID that appears more than once;
2) flag an age below 0 or above 100.
Explain each formula in one sentence. Do not suggest deleting or changing any data.
What comes back
G2:=IF(COUNTIF($D$2:$D$400,D2)>1,"Duplicate","") marks any beneficiary ID that appears more than once.
H2:=IF(OR(E2<0,E2>100),"Check age","") marks ages below 0 or above 100.
Before you trust them, test both formulas on rows where you know the answer.
Step 4 of 6Check the AI output
Independence first: read the draft and note the problems you see yourself, before you run the QA prompt. Then compare.
Another draft came back like this Illustrative
G2:=IF(COUNTIF(D2:D400,D2)>1,"Duplicate","")
H2:=IF(E2>=100,"Check age","")
Then delete the duplicate rows to clean the data.
The QA prompt (run it on every draft)
Act as an independent reviewer of Excel formulas used for data checks.
Check the formulas below against the rule.
List: every range that will shift when the formula is copied down, every condition that differs from the rule, and every step that changes or deletes data.
Do not rewrite the formulas. List the problems only.
Rule: [paste the rule]
Formulas: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
The range D2:D400 has no $ signs, so it shrinks as the formula is copied down and misses earlier duplicates.
">=100" flags an age of exactly 100, which the rule allows, and it misses negative ages.
Deleting duplicates destroys evidence: the enumerator must check which record is right.
Corrected output
G2:=IF(COUNTIF($D$2:$D$400,D2)>1,"Duplicate","")
H2:=IF(OR(E2<0,E2>100),"Check age","")
Flagged rows go to the team leader. Nothing is deleted or changed without a reply from the field.
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Quality: test each AI-written formula on rows where you know the answer before you use it on real submissions.
Step 6 of 6Quick check
Quick check+25 XP
The formulas flag 12 duplicate IDs in yesterday's 350 submissions. What do you do next?
Show answer
Send them to the team leader for checking with enumerators, and record each decision.
Try it this week
Add a test row with a duplicate ID and one with an age of 104 to your next export. Check that your formulas flag both before you use them on the real rows.
From the profile: Copilot drafts the monthly report from the updated IPTT and last month's text; the officer checks each of the 18 indicator figures against the tracker and adds reasons for delays from field notes.
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
You copy last month's text and update the numbers by hand.
Figures are typed twice, so mistakes creep in.
Reasons for delays are vague or missing.
How it works with AI
Copilot drafts from the final IPTT and last month's report.
You check each figure against the tracker.
You add the reasons for delays from field notes.
Step 2 of 6Step by step
YouFinalise the IPTT firstExcel
Draft only from final figures. Note any partner return that is late.
AIAsk Copilot for a draftMicrosoft 365 Copilot
Give it the IPTT and last month's report. Ask it to write [REASON] where a delay needs explaining (prompt below).
YouCheck every figureExcel
All 18 indicators, one by one, against the IPTT. Tick each in the Report Figure Check.
YouAdd the reasons
From field notes and partner emails, not from the AI.
YouRemove what the data doesn't say
Causes, predictions and words like "significant" or "on track" that the figures don't support.
Your teamSend it to your supervisor
Attach the figure check so they can see what you verified.
Step 3 of 6Worked example
What you give the AI Illustrative
Indicator (IPTT, March)
Result
Target
Teachers trained
214
300
Women among trainees
96 of 214
50%
Households receiving cash
1,850
2,000
Youth enrolled in livelihoods training
140 (Partner C return late)
200
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
Using only the IPTT figures below, draft the monthly monitoring summary: one sentence per indicator with the result, the target and the percentage of target.
Where a result is behind target, write [REASON] instead of explaining it.
Do not add causes, predictions or figures that are not in the table.
[paste IPTT rows]
What comes back
214 teachers were trained against a target of 300 (71%). [REASON]
Women were 96 of the 214 trainees (45%), against a target of 50%. [REASON]
Cash transfers reached 1,850 of 2,000 households (93%).
Youth enrolment is 140 of 200 (70%); Partner C's return is late, so this figure is provisional. [REASON]
The [REASON] gaps are yours to fill from field notes.
Step 4 of 6Check the AI output
Independence first: read the draft and note the problems you see yourself, before you run the QA prompt. Then compare.
Another draft came back like this Illustrative
Teacher training reached 81% of target (214 of 300), thanks to strong partner engagement. Cash transfers reached 1,850 households (93% of target). Youth enrolment is on track to meet its target.
The QA prompt (run it on every draft)
Act as an independent reviewer of a monitoring report.
Compare the draft below with the IPTT rows.
List: every figure or percentage that differs from the IPTT, every cause the IPTT does not state, every prediction, and every indicator in the IPTT that the draft leaves out.
Do not rewrite the draft. List the problems only.
IPTT: [paste]
Draft: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
214 of 300 is 71%, not 81%.
"Thanks to strong partner engagement" is a cause the IPTT does not state.
"On track" is a prediction: youth enrolment is 140 of 200 (70%), and Partner C's return is late.
Women among trainees (96 of 214, 45%) is missing.
Corrected output
214 teachers were trained against a target of 300 (71%). Women were 96 of the 214 trainees (45%). Cash transfers reached 1,850 of 2,000 households (93%). Youth enrolment is 140 of 200 (70%); Partner C's return is late, so this figure is provisional.
Reasons for delays: added by the officer from field notes and partner emails.
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Accountability: the report goes out under your name: every figure ticked against the IPTT, every delay explained from field notes.
Step 6 of 6Quick check
Quick check+25 XP
Copilot's draft says "214 teachers trained (81% of target)". The target is 300. What do you do?
Show answer
Correct it to 71% (214 of 300), then check every other figure against the IPTT.
Try it this week
Next month, tick each figure in your report against the IPTT in the Report Figure Check. Count how many the AI draft got wrong.
From the profile: after visiting six schools, the officer pastes anonymised handwritten-note transcripts into an AI assistant, which fills the visit template; the officer corrects attendance counts and adds follow-up actions.
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
Handwritten notes are typed up days later.
The template is filled from memory.
Follow-up actions get lost between visits.
How it works with AI
You type and anonymise your notes the same day.
The AI fills the visit template from those notes.
You correct the counts and decide the follow-up actions.
Step 2 of 6Step by step
YouType up your notes the same day
While you still remember what each note means.
YouRemove names and identifying details
Replace names with roles (Teacher 1, the head teacher). Remove phone numbers.
AIAsk the AI to fill the templateClaude
Give it the template headings and your anonymised notes. Ask it to write [NOT IN NOTES] for empty headings (prompt below).
YouCorrect the counts against the registers
Attendance and other counts come from the register you copied, not from the AI.
YouAdd the follow-up actions
Who does what, and by when. These are your decisions.
Your teamSave and shareSharePoint
Save the note in the visit folder. File the original notes as your data protection procedures require.
Step 3 of 6Worked example
What you give the AI Illustrative
Notes, School 4 (anonymised): Visited 9:30. Class 5 register: 38 enrolled, 29 present. Teacher 2 says the latrine door has been broken since last month. Head teacher asks when the reading books will arrive. Two girls absent for the harvest. Meal served at 11:00.
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
Fill the monitoring visit template below using only these anonymised notes.
Headings: Date and site; Attendance; Observations; Issues raised; Follow-up needed.
Copy numbers exactly as written. Where the notes say nothing for a heading, write [NOT IN NOTES].
Do not add names, opinions or recommendations.
[paste the template headings and your notes]
What comes back
Heading
AI draft
Attendance
Class 5: 29 of 38 present; two girls absent for the harvest
Observations
Meal served at 11:00
Issues raised
Latrine door broken since last month; head teacher asks when reading books will arrive
Follow-up needed
[NOT IN NOTES]
The follow-up actions are yours to decide.
Step 4 of 6Check the AI output
Independence first: read the draft and note the problems you see yourself, before you run the QA prompt. Then compare.
Another draft came back like this Illustrative
Heading
AI draft
Attendance
32 of 38 pupils present (84%)
Issues raised
Head teacher Mrs Amina asked when reading books will arrive
Follow-up needed
Partner to repair the latrine door within one week
The QA prompt (run it on every draft)
Act as an independent reviewer of a monitoring visit note.
Compare the note below with the original anonymised notes.
List: every number that differs from the notes, every name or other detail that identifies a person, and every action or opinion that is not in the notes.
Do not rewrite the note. List the problems only.
Notes: [paste]
Visit note: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
The notes say 29 of 38 present (76%), not 32 (84%).
A name appears that is not in the anonymised notes. Remove it.
The repair deadline is an action the notes do not contain. Follow-up actions are yours to decide.
Corrected output
Heading
Corrected
Attendance
29 of 38 pupils present in Class 5 (76%), as in the register
Issues raised
The head teacher asked when reading books will arrive
Follow-up needed
Decided by the officer: report the latrine door to the partner; ask the education officer about the book delivery date
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Ethics: names and phone numbers come out of your notes before they go into any AI tool.
Step 6 of 6Quick check
Quick check+25 XP
The AI's note says 32 pupils were present. The register you copied says 29. Which goes in the report?
Show answer
29, from the register. Then check every other count in the note.
Try it this week
After your next visit, anonymise your notes, ask an AI assistant to fill the visit template, then compare every number with your notes.
Refresh the dashboard and check it before it goes out
From the profile: the officer refreshes a Power BI dashboard on training reach by district and uses Copilot to suggest a chart for sex-disaggregated completion rates, checking totals against the IPTT.
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
You refresh the dashboard and publish it without looking.
Charts are chosen by habit.
Nobody compares the totals with the IPTT.
How it works with AI
You refresh, then check the totals against the IPTT first.
Copilot suggests a chart for the question you ask.
You check that the chart shows rates, not just counts.
Step 2 of 6Step by step
AutomatedRefresh the dataPower BI
Run the refresh and note the date of the data.
YouCheck the totals firstPower BI · Excel
Women and men trained, by district, must add up to the IPTT total.
AIAsk Copilot for a chartMicrosoft 365 Copilot
For example: compare completion rates for women and men by district (prompt below).
YouCheck the chart
Rates, not counts; the axis starts at zero; labels are readable.
YouWatch small numbers
A district with a handful of trainees gives unstable rates. Show the counts too.
Your teamShare it, with what changed
Tell your manager what changed since last month and anything still provisional.
Step 3 of 6Worked example
What you give the AI Illustrative
District
Women: trained / completed
Men: trained / completed
North
30 / 27
26 / 22
South
20 / 15
34 / 30
East
24 / 21
30 / 24
West
22 / 19
28 / 25
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
Here is a table of teacher training by district and sex: trained and completed.
1) Calculate the completion rate for women and for men in each district.
2) Suggest one chart that compares those rates, and say why.
3) Check that women and men add up to [214] trained in total. Show the sums.
Do not comment on causes.
[paste table]
What comes back
District
Women
Men
North
90%
85%
South
75%
88%
East
88%
80%
West
86%
89%
Chart: clustered bars, one pair per district, axis 0–100%.
Sums: 96 women + 118 men = 214 trained, matching the IPTT.
South stands out (women 75%, men 88%). Ask the field why. Don't explain it yet.
Step 4 of 6Check the AI output
Independence first: read the draft and note the problems you see yourself, before you run the QA prompt. Then compare.
Another draft came back like this Illustrative
Suggested chart: stacked bars of trainees by district. Total trained: 224. Women complete the training more often than men in every district.
The QA prompt (run it on every draft)
Act as an independent reviewer of a dashboard update.
Compare the summary below with the training table and the IPTT total.
List: every total that differs from the IPTT, every claim the table contradicts, and whether the chart shows what was asked (completion rates for women and men by district).
Do not rewrite the summary. List the problems only.
Table: [paste]
IPTT total: [paste]
Summary: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
The total is 214 in the IPTT, not 224.
"In every district" is wrong: in South, women complete at 75% and men at 88%; in West, 86% and 89%.
Stacked bars of trainees show counts, not completion rates by sex.
Corrected output
Clustered bars, one pair per district, showing completion rates for women and men (axis 0–100%), with the counts in the labels.
Totals: 96 women + 118 men = 214 trained, matching the IPTT. Women complete at a lower rate than men in South and West.
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Quality: totals match the IPTT before the dashboard is shared, whatever the chart suggests.
Step 6 of 6Quick check
Quick check+25 XP
After the refresh, the dashboard shows 224 trainees. The IPTT says 214. What do you do?
Show answer
Hold the dashboard and find the difference, such as duplicate rows or a new partner file, before you share it.
Try it this week
Before your next dashboard refresh goes out, compare three totals with the IPTT and write down any difference.
From the profile: spotting unusual patterns in site-level data with AI prompts and recording the explanation for each; recognising when a flagged value is an entry error or a real situation in the field.
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
Unusual figures are noticed by chance, or by the donor.
Guesses about the cause go into the report.
Nobody records the explanation, so the question comes back.
How it works with AI
The AI lists the sites that changed more than your rule.
You ask the field what happened.
Each explanation and decision goes in the Anomaly Log.
Step 2 of 6Step by step
YouPrepare the site tableExcel
This month and last month for each site: aggregated, with no personal data.
YouSet the rule first
For example, more than 30% up or down. Write it down before you ask the AI.
AIAsk the AI to list, not explainChatGPT
It lists the sites over the rule and the change in percent. It does not guess causes (prompt below).
Your teamAsk the field
Call the site supervisor or partner. Ask open questions, politely.
YouRecord the explanation and decisionExcel
Entry error, real change or pending, and who told you, in the Anomaly Log.
YouCorrect only with evidence
Fix entry errors at the source with a note. Explain real changes in the report.
Step 3 of 6Worked example
What you give the AI Illustrative
Site
Households receiving cash, February
March
Site 1
410
420
Site 2
380
395
Site 3
150
450
Site 4
400
115
Site 5
465
470
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
Below are households receiving cash by site, last month and this month.
List every site where the change is more than 30% up or down, with the change as a percentage.
Do not explain why and do not suggest corrections.
[paste table]
What comes back
Site
Change
Site 3
150 → 450 (+200%)
Site 4
400 → 115 (−71%)
Two flags. Now ask the field: one may be a real change, the other a reporting gap.
Step 4 of 6Check the AI output
Independence first: read the draft and note the problems you see yourself, before you run the QA prompt. Then compare.
Another draft came back like this Illustrative
Site 3 rose by 300%, probably because of duplicate registrations. Site 4 fell by 71% because households left the area. Site 2 also changed a lot.
The QA prompt (run it on every draft)
Act as an independent reviewer of a data check.
Compare the list below with the table and the rule.
List: every percentage that is wrong, every explanation that the data does not contain, and every site listed that does not meet the rule.
Do not rewrite the list. List the problems only.
Rule: [paste]
Table: [paste]
List: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
Site 3 rose by 200% (150 to 450), not 300%.
"Duplicate registrations" and "households left the area" are guesses. Only the field can explain a change.
Site 2 changed by 4% (380 to 395): below the 30% rule.
Corrected output
Site
Change
Explanation (and who gave it)
Decision
Site 3
+200%
First full distribution round this month (site supervisor)
Real change
Site 4
−71%
Partner return covers 2 of 7 distribution days (partner)
Pending: full return requested
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Independence: the AI lists unusual figures. You and the field decide whether each one is an error or a real change.
Step 6 of 6Quick check
Quick check+25 XP
The AI says Site 3 rose "probably because of duplicate registrations". What do you do?
Show answer
Ask the field and record the explanation and its source. In this case, Site 3 had its first full distribution round.
Try it this week
Run the 30% rule on last month's site data. For each flagged site, write the explanation and who gave it in the Anomaly Log.
From the profile: using AI-assisted translation for short beneficiary responses, checking key terms with field colleagues; removing names and phone numbers before using any AI tool.
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
Short answers in local languages stay untranslated.
Names and phone numbers travel with the answers.
Key local terms are translated loosely.
How it works with AI
You remove names, phone numbers and places first.
An approved AI tool translates the short responses.
A field colleague checks the key terms.
Step 2 of 6Step by step
YouKeep only what you needExcel
The response ID and the answer. Drop the name, phone and GPS columns.
YouRead for hidden identifiers
Names of people or small villages inside the answer: replace them with [NAME] or [PLACE].
AITranslate with an approved toolClaude · ChatGPT
Keep the IDs, translate literally and mark unclear words (prompt below).
Your teamCheck key terms with a field colleague
Words for money, aid, complaints and local measures.
YouKeep a glossary
Note each corrected term so the next translation starts right.
YouStore safelySharePoint
Keep only the anonymised file in shared folders. Follow your organisation's rules for the original.
Step 3 of 6Worked example
What you give the AI Illustrative
Before
After (what goes into the AI tool)
R-104 · [full name] · [phone number] · [village] · "The money came late, so I bought maize on credit."
R-104 · "The money came late, so I bought maize on credit."
R-117 · [full name] · [phone number] · "The [local measure] was too small for my family."
R-117 · "The [local measure] was too small for my family."
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
Translate the short survey responses below from [language] into English.
Keep each response ID.
Translate as literally as possible. Mark any word you are unsure of as [UNCLEAR: original word].
Do not summarise or interpret.
[paste anonymised responses]
What comes back
ID
Translation
R-104
The money came late, so I bought maize on credit.
R-117
The [UNCLEAR: local measure] was too small for my family.
A field colleague confirms the local measure is one bag of grain. It goes in the glossary.
Step 4 of 6Check the AI output
Independence first: read the draft and note the problems you see yourself, before you run the QA prompt. Then compare.
Another draft came back like this Illustrative
ID
AI translation
R-104
The beneficiary was unhappy about the late payment.
R-117
The amount was too small for my family.
R-121
Amina says the distribution point is too far.
The QA prompt (run it on every draft)
Act as an independent reviewer of a translation.
Compare the translation below with the anonymised originals.
List: every response that is summarised or interpreted instead of translated, every unclear word translated without being marked, and every name, phone number or place that appears.
Do not rewrite the translation. List the problems only.
Originals: [paste]
Translation: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
R-104 is a summary, not a translation, and it loses "so I bought maize on credit".
R-117 replaces the local measure with "the amount" without marking it as unclear.
R-121 contains a name, so personal data went into the AI tool. Stop, and report it as your organisation's rules require.
Corrected output
ID
Corrected translation
R-104
The money came late, so I bought maize on credit.
R-117
The [UNCLEAR: local measure, confirmed by a field colleague as one bag of grain] was too small for my family.
R-121
[NAME] says the distribution point is too far. (Name removed; incident reported.)
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Ethics: people's words are translated, not summarised, and nothing that identifies them goes into the AI tool.
Step 6 of 6Quick check
Quick check+25 XP
You notice a phone number in a response you already pasted into an AI tool. What now?
Show answer
Stop and report it as your organisation's rules require, then fix your anonymisation step.
Try it this week
Take ten short responses. Remove every name, phone number and place, then ask a field colleague to check the AI translation of three key terms.
Pick a recurring task that is not covered above, such as post-distribution monitoring interviews or updating the IPTT from partner returns, and use this prompt to draft your own step-by-step guide.
I am an M&E officer. Here is a recurring task I do: [describe the task, how often, who is involved, which tools and data].
Break it into numbered steps. For each step, say who should do it (me, a colleague, the field team, an automated step or an AI assistant), which tool, and what could go wrong.
Mark every step where a person must check the result before it is used, and say what they should check.
Do not suggest putting names, phone numbers or other personal data into an AI tool.
End of Mission 04
MISSION05
The skills you need to develop
Skill inventory · Badge: Skill Builder
1 challenge
Mission goal: find out which of the 11 profile skills you already have and which need new learning.
Step 1 of 5Guess first
Your guess+20 XP
Which is the first learning priority for this role in the career profile?
Show answer
Excel, followed by prompt design, validating AI outputs, Power BI and data protection.
Step 2 of 5Skills you bring with you
Tick the ones you already have. Open a skill to see how to build it.
Your skills match
0 of 11
A. Existing M&E and professional competencies
Transfers from your role
XLSForm design, ODK-based platforms, offline collection and form versioning. Knowing the form tells you which flags make sense.
Build it: Write down the valid range for every number question in one form.
Transfers from your role
Deduplication, recoding, merging and documenting cleaning decisions. AI writes formulas; you decide what counts as an error.
Build it: Keep a log of every cleaning decision for one month.
Transfers from your role
M&E plans, IPTTs, data flow maps and reporting calendars: the source every AI draft is checked against.
Build it: Mark in the IPTT which figures are final and which are provisional.
Transfers from your role
Validity, reliability, timeliness, precision and integrity, including tracing figures back to source registers.
Build it: Trace three reported figures back to their registers.
Transfers from your role
SMART indicators, baselines, targets and disaggregation. Clear definitions show when an AI draft gets a figure wrong.
Build it: Read the reference sheet for three indicators you report.
Transfers from your role
Choosing charts that answer a specific question, whoever suggests them.
Build it: Redraw one dashboard chart to show rates instead of counts.
Transfers from your role
Clear findings without overclaiming: the skill you use to correct AI drafts.
Build it: Remove every unsupported cause from one AI draft.
Transfers from your role
Interviews, observation and notes from field visits and post-distribution monitoring.
Build it: Type up and anonymise your notes on the day of your next visit.
Step 3 of 5New skills for the AI era
These are the ones most officers still need to learn.
Your skills match
0 of 11
B. New AI competencies
Needs new learning
Instructions, examples and output formats that give reliable results from AI assistants (a learning priority).
Build it: Rewrite one prompt with an example and a fixed output format.
Needs new learning
Checking AI results against source data, sampling for errors and keeping outputs traceable (a learning priority).
Build it: Use the Report Figure Check on your next monthly report (guide 2).
D. Ethics, governance and quality assurance
Needs new learning
Consent, anonymisation and safe handling of personal and sensitive data. Officers handle raw beneficiary data (a learning priority).
Build it: List what you remove before using AI, and agree it with your supervisor (guide 6).
Step 4 of 5Your five learning priorities
In this order, from the career profile:
Excel. Most monitoring data still passes through Excel, and AI tools write formulas faster than they can be checked by someone without Excel skills.
Prompt design. Clear prompts with examples give more reliable report drafts and data checks.
Validating AI outputs. The officer is the first line of defence against AI errors entering the IPTT.
Power BI. Dashboard maintenance is becoming a standard officer task.
Data protection and privacy. Officers handle raw beneficiary data and must know what may go into AI tools.
You don't need to code. An AI assistant can write the formula; your Excel skills tell you whether it is right. Learn each formula the AI gives you.
End of Mission 05
MISSION06
Interactive skills gap assessment
Know your gaps · Badge: Self-Aware Analyst
1 challenge
Mission goal: measure your readiness in four areas and find your top three gaps.
Step 1 of 3How it works
10 quick statements, about 3 minutes. Rate yourself honestly: this is a learning tool, not a test, and your answers stay in your browser.
Beginner: little or no experience
Developing: I understand it but need practice
Competent: I can apply it independently
Step 2 of 3Questions 1 to 5
Step 3 of 3Questions 6 to 10
Prefer to score it yourself?
Beginner = 0, Developing = 1, Competent = 2. Add them up (maximum 20): 0–7 Getting started · 8–14 Developing · 15–20 Ready to use AI in your daily monitoring work. For every Beginner answer, start with the matching action:
Use the routine in mission 4, guide 1.
Write and test three prompts for the mission 8 project.
Use the Report Figure Check in mission 4, guide 2.
Use the Anomaly Log in mission 4, guide 5.
Trace three figures from your last report.
Use the routine in mission 4, guide 4.
Practise on the red flags in mission 3.
Use the routine in mission 4, guide 6.
Start a glossary of key terms (guide 6).
Agree a short, polite way to ask about flagged records with your team leader.
Mission goal: commit to a realistic 12-week plan and track it week by week.
Step 1 of 5Weeks 1–2: Understand AI
About 2–3 hours a week (a planning suggestion). Tick steps as you do them: +5 XP each, saved in this browser. The mission completes when every step in all four stages is ticked.
Stage 1 · Weeks 1–20 / 5 done
Understand AI
Build AI literacy on low-risk tasks.
Loot: an error log, a no-paste list and a list of routine tasks.
Step 2 of 5Weeks 3–4: Apply AI
Stage 2 · Weeks 3–40 / 4 done
Apply AI
Practise on your own data, with every output checked.
Loot: tested formulas, a checked report section, a visit note and an anomaly log.
Step 3 of 5Weeks 5–8: Integrate AI
Stage 3 · Weeks 5–80 / 5 done
Integrate AI
Make the checks part of your routine.
Loot: a daily routine, a checked monthly report and an agreed way of working.
Step 4 of 5Weeks 9–12: Demonstrate competence
Stage 4 · Weeks 9–120 / 5 done
Demonstrate competence
Show you can use AI with checks, every month.
Loot: a portfolio project, a month of evidence and a development plan.
Step 5 of 5Check
Mission check+20 XP
In which stage do you first test AI-written formulas on rows where you know the answer?
Show answer
Stage 2 (weeks 3–4).
End of Mission 07
MISSION08
Your practical learning project
Boss level · Badge: Routine Builder
1 challenge + 8 deliverables
Mission goal: produce a portfolio piece that proves you can use AI in monitoring work responsibly.
Step 1 of 6Your boss-level project
Boss level project
Build an AI-assisted monthly monitoring routine, with checks
Objective: show you can use AI in your daily and monthly monitoring work, check every output against source, and keep personal data out of AI tools.
Case study · Learning and Livelihoods Project Illustrative
No programme data of your own to use? Build the routine for this project, using the examples from mission 4.
Context: one month of your own monitoring work: daily submissions, the IPTT, the monthly report and the dashboard. Anonymise anything you share.
Resources: a KoboToolbox or ODK export, the IPTT, last month's report, your visit notes, an approved AI assistant, Excel and Power BI.
Step 2 of 6Deliverables, one by one
Deliverables checklist0 / 8 done
8 steps to a portfolio piece
The mission and the certificate unlock only when all 8 deliverables are ticked.
Step 3 of 6Templates to start from
Four ready-made files for your deliverables. Download, adapt and keep them.
Before you share it: what should a person check in your routine?
Think about it for a moment, then reveal.
Every figure in the report check, against the IPTT.
The formulas, on test rows with known answers.
The anomaly explanations, against what the field said.
Any AI-generated text, reviewed and marked as AI-assisted.
No names, phone numbers or other personal data in anything you share.
Step 5 of 6How it's judged
Criterion
What good looks like
Tested tools
Every formula and prompt was tried on data where you knew the answer.
Traceable figures
Every figure in the report is ticked against the IPTT or a register.
Field follow-up
Flags are explained by the field, and the source is recorded.
Data protection
No personal data went into an AI tool.
Honest reporting
No cause, prediction or figure the data does not support.
Clear evidence
The note shows what the checks caught, with numbers.
Step 6 of 6Check
Mission check+20 XP
Which deliverable shows your supervisor that the figures in your monthly report are right?
Show answer
The Report Figure Check.
End of Mission 08
MISSION09
Responsible AI: what you must never delegate blindly
Guardrails · Badge: Responsible AI Guardian
10 challenges
Mission goal: know exactly which judgements stay with you, and the control for each AI risk.
Step 1 of 6Would you approve this report?
Click every sentence you would not send. There are five problems: an invented figure, an unsupported causal claim, an incorrect percentage, a missing indicator and a misleading interpretation.
Case study · Learning and Livelihoods Project Illustrative
Copilot drafted the monthly monitoring report from the IPTT. Your supervisor needs it today.
Source (IPTT, March)
Result
Teachers trained
214 (target 300)
Households receiving cash
1,850 (target 2,000)
Post-distribution monitoring: full amount received
82% of recipients
Attendance at the six schools visited
76% (last month 79%); visits took place in exam week
Youth enrolled in livelihoods training
140 (target 200); Partner C return late
Women's income
No indicator
Problems found: 0 of 5+40 XP
Show answer
Five problems: the 98% (it is 93%), the invented income figure, the unsupported claim about the registration system, the misleading reading of attendance (exam week), and the claim that all indicators are reported (youth enrolment is missing).
Step 2 of 6Your decision
Your decision+20 XP
You found the problems. Can you send this report to your supervisor?
Show answer
Correct it first. Fix each problem, recheck every figure against the IPTT, then send it.
Step 3 of 6Five questions before you send it
The report is corrected. Before you send it, answer the five questions. Tick each one you can honestly answer yes to.
Step 4 of 6Mini-game: delegate or never?
AI can write, draft and translate. It cannot be accountable. For each task: can AI help (with checks), or must you never delegate it?
Delegate?+10 XP
Writing Excel formulas that flag duplicate beneficiary IDs
Show answer
AI can help. AI writes them; you test them on rows where you know the answer.
Delegate?+10 XP
Deciding whether a flagged age is an entry error or real
Show answer
Never delegate. Listed in the profile as human judgement, with the field.
Delegate?+10 XP
Drafting the monthly report text from the IPTT
Show answer
AI can help. With every figure checked against the tracker, a good use of AI.
Delegate?+10 XP
Confirming that report figures match source records before submission
Show answer
Never delegate. Listed in the profile as human judgement.
Delegate?+10 XP
Filling the visit template from anonymised notes
Show answer
AI can help. AI fills it; you correct counts and add the actions.
Delegate?+10 XP
Deciding what personal data must be removed before using AI
Show answer
Never delegate. Listed in the profile as human judgement.
Delegate?+10 XP
Translating short anonymised responses, with key terms checked
Show answer
AI can help. Useful when a field colleague checks the key terms.
Delegate?+10 XP
Following up a sensitive issue with a community member
Show answer
Never delegate. Following up sensitively with people stays with you.
Step 5 of 6Six risks, six controls
What can go wrong when AI helps with monitoring data? Name three risks before you look.
Think about it for a moment, then reveal.
Risk
What it looks like
Your control
Wrong formulas
A range that shifts, or a condition that differs from the rule
Test on rows where you know the answer
Changed figures
A percentage or total that differs from the IPTT
Tick every figure against the IPTT
Invented causes
"Thanks to" or "because of" with no evidence
Delete causes the data does not show
Personal data
Names or phone numbers pasted into an AI tool
Remove them first; approved tools only
Mistranslation
A local term replaced by a loose word
Field colleagues check key terms
Silent corrections
A value changed in the office without asking the field
No change without a reply; log the decision
Step 6 of 6Tell readers how AI was used
Transparency is a control too. Add a short note to any report drafted with AI, and adapt it to what was actually done.
This report was drafted with Microsoft 365 Copilot from the [month] IPTT. Every figure was checked against the IPTT and source records by [name], M&E Officer. Reasons for delays come from field notes and partner reports. No personal data was entered into AI tools.
The AI-Enabled M&E Specialist is covered in Tutorial 02, the AI-Enabled M&E Coordinator in Tutorial 05, the AI-Powered Monitoring Specialist in Tutorial 07 and the AI-Assisted MEAL Officer in Tutorial 14.
Step 2 of 5What you need to show
Strengths you build on
Data collection with mobile forms
Data cleaning and the IPTT
Field monitoring and spot checks
Training enumerators and partner staff
What you need to show
Data checks with tested formulas
Monthly reports with every figure traced
Unusual figures explained by the field
Dashboards checked against the IPTT
Safe use of AI with personal data
Step 3 of 5Tool: build your LinkedIn headline
Pick the pieces that are true for you. Only claim what you have actually done.
Your headline
Step 4 of 5CV bullets and qualifications
CV bullet templates
Set up AI-assisted daily checks on [350] KoboToolbox submissions a day, with formulas tested on known cases; flagged records followed up with enumerators the same day.
Drafted monthly monitoring reports with Microsoft 365 Copilot and checked all [18] indicator figures against the IPTT before submission.
Kept an anomaly log for [N] sites, recording a field-confirmed explanation for every unusual figure.
Maintained a Power BI dashboard on training reach by district and sex, with totals checked against the IPTT.
Replace the brackets with what you have actually done.
Qualifications in the profile
Bachelor's degree in social sciences, statistics, development studies or a related field.
1–3 years in data collection, programme monitoring or data entry, ideally in an NGO or UN programme.
Useful certification: MEAL DPro (PM4NGOs).
Step 5 of 5Check
Mission check+20 XP
What's the best way to use "AI-Assisted M&E Officer" on your CV?
Show answer
Use it to describe your capabilities, backed by evidence, unless your employer actually uses the title.
End of Mission 10
MISSION11
Your next steps
Commit and celebrate · Badge: Transition Ready
1 challenge
Mission goal: turn what you learned into a two-week plan and unlock your certificate.
Step 1 of 2Commit to your next two weeks
Pick the actions you commit to for the next two weeks. Your plan builds itself below.
Start learning
Do this
Show this
My plan
Recommended first project: the monthly monitoring routine from mission 8, starting with the daily submission check.
Step 2 of 2Your completion certificate
Complete all 11 missions to unlock it. Add your name, take a screenshot and share your progress.
This is a learning-progress certificate. It records what you completed in this free tool, in your own browser; it is not independently verified and not a formal qualification. Your mission 8 project is the evidence to show an employer.
End of Mission 11
Questions and answers
What does an AI-Assisted M&E Officer do?
Collects, checks and reports monitoring data for one programme or a set of field sites, keeping indicator records accurate and up to date. Uses AI tools to automate routine monitoring reports, analyse programme data, flag anomalies and prepare dashboards, and checks every output before it is shared.
How is AI changing the M&E Officer role?
Much of the officer's repetitive work, such as first data checks, standard tables and routine report text, is now drafted by AI tools or simple automations. The officer's value shifts to checking those outputs, following up on flagged records with the field and knowing the data well enough to notice when something looks wrong.
Which judgements should an AI-Assisted M&E Officer never hand over to AI?
The career profile lists these as human judgement:
Recognising when a flagged value is an entry error or a real situation in the field.
Following up sensitively with enumerators, partners and community members.
Deciding what personal data must be removed before using AI tools.
Checking that report figures match source records before submission.
Which skills should I learn first?
The learning priorities in the career profile, in order:
Excel. Most monitoring data still passes through Excel, and AI tools write formulas faster than they can be checked by someone without Excel skills.
Prompt design. Clear prompts with examples give more reliable report drafts and data checks.
Validating AI outputs. The officer is the first line of defence against AI errors entering the IPTT.
Power BI. Dashboard maintenance is becoming a standard officer task.
Data protection and privacy. Officers handle raw beneficiary data and must know what may go into AI tools.
The profile lists 11 skills in all: Mobile data collection, Data cleaning and wrangling, M&E planning, Data quality assessment, Indicator design, Data visualization, Prompt design, Validating AI outputs, Data protection and privacy, Analytical report writing and Qualitative methods.
Which tools does this role use?
The career profile names: KoboToolbox, ODK, Excel, Power BI, Microsoft 365 Copilot, ChatGPT, Claude and SharePoint. The tutorial shows how to check every AI output, whichever tool you use.
What qualifications and experience does the profile list?
Bachelor's degree in social sciences, statistics, development studies or a related field.
1–3 years in data collection, programme monitoring or data entry, ideally in an NGO or UN programme.
Useful certification: MEAL DPro (PM4NGOs).
How long does the transition take?
This tutorial suggests a 12-week plan of about 3 hours a week, in four stages: understand AI (weeks 1–2), apply AI (weeks 3–4), integrate AI (weeks 5–8) and demonstrate competence (weeks 9–12). It is a planning suggestion, not a requirement.
Is "AI-Assisted M&E Officer" an official job title?
No. It is a proposed career title from the EvalCommunity framework, not a standardised occupational classification or a guaranteed vacancy. Use it to describe your capabilities, backed by evidence, unless your employer actually uses the title.
