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 M&E function?
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 Manager
AI-enhanced role
AI-Enabled M&E Manager
Level
Manager · 7–12 years
How the role uses AI
Integrates AI into workflows and is accountable for the results
"AI-Enabled M&E Manager" 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 routine reporting is automated, what becomes your main job?
Show answer
In the profile, much routine data processing and report drafting is automated. Your main job becomes setting controls, staffing and budgets around those workflows. You stay accountable for what the system reports, and you need to know where it fails.
Step 4 of 5AI-Enabled, not AI-Dependent
The goal is not to hand your work to AI. It is to become a stronger M&E manager who uses AI and stays in charge of it. Here are the five questions, asked about your own work:
1QualityDoes AI make your quarterly pack and donor reports more credible, or only faster?Sample checks and source tracing before anything is signed.
2AccountabilityWhen an AI-drafted figure is wrong, can you show who checked it, and when?A named checker for every automated step, and your signature on the report.
3IndependenceDo you and your specialists still judge results yourselves before reading the AI's view?Specialists interpret the indicators; AI drafts the text.
4EthicsIs participants' and partners' data protected in every tool your team uses?AI data rules agreed with the data protection focal point.
5Professional developmentIs your team building its skills, or handing them over to the tools?A reskilling plan and a training line in every AI budget.
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 move your own daily tasks to AI step by step, 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 · Integrated Food Security Programme Illustrative
A fictional programme you will follow through missions 2, 3, 4, 8 and 9.
The programme
Six projects, five implementing partners, 25,000 households targeted. Monthly indicator tracking and quarterly donor reports.
The tools
KoboToolbox, Excel, Power BI and SharePoint.
Your challenge
You spend too much time consolidating data, checking reports, finding discrepancies and preparing donor presentations.
Your objective
Move routine work to AI-assisted workflows while staying accountable for every figure.
Step 2 of 4Flip your tasks: today and with AI
Most of your work is not replaced. It gains a layer of control. 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
Consolidating monthly data from six projects and their partners
Show answer
AI-assisted. Routine consolidation is exactly what the profile suggests automating first.
Sort it+10 XP
Writing the first draft of the quarterly performance narrative
Show answer
AI-assisted. Copilot can draft it; your team corrects and adds field explanations.
Sort it+10 XP
Deciding which M&E processes to automate first
Show answer
New skill. A new management decision: weigh staff time saved against the risk of errors.
Sort it+10 XP
Setting the QA sampling rate for AI-coded feedback
Show answer
New skill. In the profile, the manager agrees the sampling rate with the senior specialist.
Sort it+10 XP
Budgeting for LLM API usage and AI licences
Show answer
New skill. AI licences and API costs are now regular M&E budget lines.
Sort it+10 XP
Signing off figures sent to the donor
Show answer
Human. You sign and you accept accountability. AI can help you check, not sign.
Sort it+10 XP
A difficult performance conversation with a partner
Show answer
Human. Listed in the profile as one of the things that stay human.
Sort it+10 XP
Splitting staff time and budget between automation and fieldwork
Show answer
Human. A judgement call the profile keeps with the manager.
Step 4 of 4Your turn: two minutes
Take last quarter's workplan. Which of your own 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
Controls, risks and red flags · Badge: Red Flag Spotter
2 challenges
Mission goal: learn the controls an AI-supported M&E system needs, and practise spotting problems in AI drafts.
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
Consolidating project and partner data
Formatting the quarterly performance pack
First drafts of narratives and briefs
Listing every quantitative claim in a long report
Can be partly automated
IPTT data into dashboards and templates
Classifying feedback messages, with sample checks
Transcript analysis with LLM APIs
New responsibilities
Deciding the order of automation
QA checkpoints and stop rules
Budget lines and data rules for AI
Tracking error rates and time saved
Stays with you
What is signed and sent to donors
Staff time vs automation vs fieldwork
How roles change and who is reskilled
Difficult performance conversations
Step 2 of 4The three controls you must defend
Automation is safe only with controls around it. Here they are, one at a time.
Sample checks. A specialist confirms a share of figures against source before sign-off (10% in the profile's example).
Source tracing. Every AI-assisted figure and claim in a donor report can be traced to source data.
Stop rules. A workflow is paused when quality drops below a threshold agreed in advance.
Step 3 of 4Mini-game: spot the red flags
Case study · Integrated Food Security Programme Illustrative
Copilot drafted the Q3 summary for the senior management team. Here is the source data and the draft.
Click every sentence you would not sign.
What the IPTT and data log show for Q3
Households reached
11,640
Indicator 2.1 (improved food consumption)
Up 40% on Q2, but reporting sites grew from 18 to 25
Partner submissions
Partner C did not submit Q3 data
Women among participants
52%
Red flags found: 0 of 3+30 XP
Show answer
Three sentences should not be signed: the "clear success" claim (7 new sites may explain the jump), "all five partners submitted" (Partner C did not) and the scale-up recommendation (not supported by evidence).
Step 4 of 4Mission check
Your call+20 XP
Your automated pack shows every indicator on track for the first time in a year. What do you do first?
Show answer
Treat it as a signal to check: sample figures against source and look for changes in the flow, then review the narrative.
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 move six of your real daily tasks to AI, step by step, with the checks that keep you in control.
Case study · Integrated Food Security Programme Illustrative
Every worked example below uses this programme's data, so you can follow one system from collection to sign-off.
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
Review the monthly performance pack for six projects
From the profile: reviewing the automated monthly performance pack for six projects before it goes to the senior management team; the automated quarterly performance pack.
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
Each project emails its Excel IPTT extract, often late and in its own format.
A data officer copies the figures into a master workbook and rebuilds the charts.
You write the narrative the night before the meeting.
Errors surface in the meeting, when it is too late to check them.
How it works with AI
Project data lands in one place on a schedule, and missing data triggers an alert.
The dashboard refreshes itself and flags indicators that need attention.
Copilot drafts the narrative from the figures.
Your time goes to the flagged indicators and to signing off.
Step 2 of 6Step by step
YouFix what the pack containsExcel
List the indicators, the owner and source file for each project, and the monthly deadline. This list is what the automation will follow.
Your teamBuild the collection flowPower Automate · SharePoint
When a project saves its file in the agreed SharePoint folder, a Power Automate flow copies the rows into one master table. If a project has not submitted by the deadline, the flow emails the officer.
AutomatedRefresh the dashboard and raise flagsPower BI
Power BI reads the master table and flags indicators that are off track, missing or changed by more than a threshold you set.
AIDraft the narrativeMicrosoft 365 Copilot
Copilot drafts a short narrative from the table, using the prompt below. It describes the figures; it does not explain them.
Your teamCheck a sample against sourceExcel
A specialist checks a share of figures against the source files. The profile's example uses 10% of figures.
YouReview the flags and sign off
Read every flagged indicator, remove any claim the data cannot support, then send the pack to the senior management team.
Step 3 of 6Worked example
What you give the AI Illustrative
Project
Indicator
Target
Actual
Flag
Project A
Households reached
2,000
1,960
Project B
Households reached
1,800
0
Missing data
Project C
Children enrolled
900
1,260
Up 40% on last month
Project D
Trainings held
12
11
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
You are drafting the monthly performance summary for our senior management team.
Use ONLY the table below. For each project, write one or two sentences describing the figures against target.
Rules: do not explain causes, do not call anything a success or failure, and mark every flagged row with [CHECK].
Table: [paste the dashboard export]
What comes back
Project A reached 1,960 households against a target of 2,000.
[CHECK] Project B reported no households for this month; data may be missing.
[CHECK] Project C reports 1,260 children enrolled against a target of 900, 40% higher than last month.
Notice that the draft stays factual and marks the rows you must look into. If a draft calls the Project C jump "a success", send it back. Guide 3 shows how to investigate it.
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
Project A nearly met its target with 1,960 households. Project B had a quiet month. Project C's enrolment jumped 40%, a strong success driven by the new outreach campaign.
The QA prompt (run it on every draft)
Act as an independent M&E quality reviewer.
Compare the draft summary below with the source table.
List: every figure that differs from the table, every statement of cause or success, every flagged row not marked [CHECK], and every project that is missing.
Do not rewrite the summary. List the corrections only.
Source table: [paste]
Draft: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
"A quiet month" hides missing data: Project B reported 0 and is flagged as missing.
"A strong success driven by the outreach campaign" states a cause the data does not show.
Project D is missing from the summary.
Flagged rows (B and C) are not marked [CHECK].
Corrected output
Project A: 1,960 households against a target of 2,000.
[CHECK] Project B: no data received this month; follow up before sign-off.
[CHECK] Project C: 1,260 children enrolled against a target of 900, 40% more than last month; cause under review.
Project D: 11 of 12 trainings held.
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Quality: a tidy AI pack can hide missing data. Check a sample of figures against source before it goes to senior management.
Step 6 of 6Quick check
Quick check+25 XP
The flow ran without errors and the pack is ready. When do you sign off?
Show answer
Sign off only after a sample check against source (10% in the profile's example) and your review of the flagged indicators.
Try it this week
Take last month's pack. Write down the indicators, the owner and the source file for each project (step 1). That list is the brief for whoever builds the flow.
From the profile: the manager logs staff hours on 14 recurring tasks for a month, then automates partner data consolidation and report formatting first, leaving indicator interpretation with staff.
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
Five partners send spreadsheets with different column names, date formats and totals.
An officer spends days fixing columns, removing duplicates and chasing late files.
Nobody knows how much time this really takes.
How it works with AI
You first measure where staff time goes, then automate the high-hours, low-risk work.
Partners submit through one template or form, and differences are mapped once.
Validation rules produce an exceptions list; people deal only with the exceptions.
Step 2 of 6Step by step
YouMeasure before you automateExcel
For one month, the team logs hours on each recurring task (the profile's example uses 14 tasks). Rank tasks by hours and by the damage an error would cause.
Your teamGive partners one way to submitKoboToolbox · ActivityInfo
Agree one template, or a form in KoboToolbox or ActivityInfo, with fixed column names and formats.
AIMap the old formats onceClaude · Copilot
Ask an AI assistant to match each partner's column headers to your template and list what does not match (prompt below).
AutomatedAppend and validate automaticallyPower Automate · Excel
A flow appends each new file to the master table and runs checks: duplicates, totals, values out of range, missing partners.
Your teamWork the exceptions list
The officer resolves only the rows the checks flagged, and follows up with partners.
YouHandle the people side
Late or poor partner data is a conversation, not a formula. Use the exceptions list in your one-to-ones with officers and in partner meetings.
Step 3 of 6Worked example
What you give the AI Illustrative
Partner
Their column headers
Partner 1
HH_reached · Date (dd/mm/yy) · District
Partner 2
No. of households · Month · Location
Partner 3
Households (total) · Reporting period · Dist.
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
Our reporting template has these columns: households_reached, reporting_month (YYYY-MM), district.
Below are the column headers used by three partners.
For each partner, map each of their columns to our template, show any format change needed (for example dates), and list any column you cannot map with certainty.
Do not guess: write UNSURE where the meaning is unclear.
[paste headers, no personal data]
What comes back
Partner
Their column
Our column
Change needed
Partner 1
HH_reached
households_reached
None
Partner 1
Date (dd/mm/yy)
reporting_month
Convert to YYYY-MM
Partner 3
Dist.
district
UNSURE: confirm with partner
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
Partner
Their column
Our column
Change needed
Partner 1
Date (dd/mm/yy)
reporting_month
None
Partner 2
Month
reporting_month
None
Partner 3
Dist.
district
None
The QA prompt (run it on every draft)
Act as an independent data reviewer.
Check this column mapping against the partners' original headers and our template.
List: any partner column left out, any format change that is missing, and any mapping stated as certain that could mean something else.
Do not fix the mapping. List the problems only.
Template: [paste]
Partner headers: [paste]
Mapping: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
Partner 1 dates need converting from dd/mm/yy to YYYY-MM; the draft says "None".
Partner 2's "Location" and "No. of households" columns are not mapped at all.
Partner 3's "Dist." is stated as certain, but the abbreviation could mean district or distribution point.
Corrected output
Partner
Their column
Our column
Change needed
Partner 1
Date (dd/mm/yy)
reporting_month
Convert to YYYY-MM
Partner 2
No. of households
households_reached
None
Partner 2
Location
district
Check spelling against district list
Partner 3
Dist.
UNSURE
Confirm with partner
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Accountability: name who checks each automated load, so any wrong total can be traced to a person and a date.
Step 6 of 6Quick check
Quick check+25 XP
You have budget for one automation this quarter. What do you automate first?
Show answer
Collect evidence first (a one-month time log), then automate high-hours, low-risk tasks such as data consolidation and formatting.
Try it this week
Start the time log on Monday: one row per recurring task, with who does it and hours per month. Review it with your team in four weeks.
From the profile: querying a 40% jump in an outcome indicator that the AI-drafted narrative described as a success.
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
The jump is noticed late, or reported as good news.
Someone scrolls through spreadsheets looking for the cause.
The explanation depends on who you happen to ask.
How it works with AI
The dashboard flags the jump automatically.
AI lists possible explanations and the check that would confirm each one.
You and the project team confirm the real cause before anything is reported.
Step 2 of 6Step by step
AutomatedSpot the jumpPower BI
Your dashboard flags any indicator that changes more than your agreed threshold between periods.
YouDrill down by site and partnerPower BI · Excel
Find which sites or partners drive the change.
AIList explanations to testClaude
Give the AI the indicator definition and an anonymised table by site. Ask for possible explanations and the check for each, without a conclusion (prompt below).
Your teamRun the checksExcel
Check for new sites, a changed denominator or definition, duplicates and data entry errors.
YouConfirm with the field
Call the project team and ask what happened on the ground.
YouCorrect the narrative
Report the jump with its explanation, or as "under review". Never as a success that nobody can explain.
Step 3 of 6Worked example
What you give the AI Illustrative
Site group
Last quarter
This quarter
Note
18 existing sites
1,800
1,870
Small rise
7 new sites
not reporting
650
Joined this quarter
Total
1,800
2,520
Up 40%
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
An outcome indicator rose 40% between two quarters.
Definition: [paste]. Figures by site: [paste anonymised table].
List the most likely data-related and programme-related explanations. For each, say which check in the data would confirm or rule it out.
Do not conclude which explanation is correct.
What comes back
New reporting sites. Check: compare the total for the 18 sites that reported in both quarters.
Changed definition or denominator. Check: compare this quarter's data dictionary with last quarter's.
Duplicate records. Check: look for repeated IDs in the new sites' data.
A real programme effect. Check: ask the project team what changed in activities.
Here the first check explains most of the jump: the 18 existing sites rose by about 4%, and the 7 new sites account for the rest.
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
Indicator 2.1 rose by 40% because the programme's nutrition training improved diets in all districts. This explanation is verified by the data.
The QA prompt (run it on every draft)
Act as an independent M&E quality assurance reviewer.
Review the analysis below against the original indicator tracking data.
Identify: incorrect calculations, unsupported interpretations, missing context, inconsistencies between figures, and claims that cannot be verified.
List the corrections required before management approval.
Data: [paste anonymised table by site]
Analysis: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
The 40% is correct (1,800 to 2,520), but it mixes in 7 new sites. The 18 existing sites rose by about 4%.
"Because of the nutrition training" is a cause the data cannot show.
"In all districts" and "verified" are not supported: nothing was checked with the field yet.
Corrected output
Indicator 2.1 rose by 40% (1,800 to 2,520), mostly because 7 new sites started reporting. At the 18 sites that reported in both quarters, it rose by about 4%. The project team is being asked what changed; the result is under review.
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Independence: AI can suggest explanations for the jump. The data and the field team decide which one is true.
Step 6 of 6Quick check
Quick check+25 XP
The AI-drafted narrative calls a 40% rise "a clear success". What do you do first?
Show answer
Treat the jump as a question, and use the data and the field team to explain it before the narrative goes anywhere.
Try it this week
Open last quarter's pack and find the indicator with the biggest change. Run steps 2 to 4 on it and write one sentence explaining it.
From the profile: before submitting an annual report, the manager asks Claude to list every quantitative claim in the 40-page draft, then checks each one against the IPTT and removes two unsupported statements.
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
You read the full draft and spot-check the figures you remember.
Numbers that slipped in unnoticed reach the donor.
There is no record of what was checked.
How it works with AI
AI builds the checklist: every quantitative claim, with its page.
You check each claim against the IPTT in a simple table.
The table becomes your sign-off record.
Step 2 of 6Step by step
YouPrepare the draft safely
Remove or mask personal data, and use a tool your organisation has approved for this kind of document.
AIList every quantitative claimClaude
Ask the AI to list every number, percentage and statement of change, with its page (prompt below).
Your teamPut the IPTT next to itExcel
Paste the list into Excel, next to the IPTT values.
YouMark each claimExcel
Matches, differs or no source. Fix every claim that differs.
YouRemove unsupported claims and sign
Remove or rewrite claims with no source. Keep the table as your record, then sign.
Step 3 of 6Worked example
What you give the AI Illustrative
"... Over 12,000 households were reached in 2025. Enrolment rose by 18% compared with baseline, and women's income doubled in all target districts ..."
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
You are helping me check a donor report before submission.
List every quantitative claim in the attached report: numbers, percentages, comparisons and statements of change (for example "increased", "doubled").
Return a table with: claim (quoted exactly), page, indicator it seems to refer to.
Do not judge whether the claims are correct, and do not add claims that are not in the text.
What comes back
Claim
Page
IPTT check
Action
"Over 12,000 households reached"
11
IPTT: 11,640
Correct the figure
"Enrolment rose by 18%"
7
IPTT: 18.2%
Keep
"Women's income doubled"
23
No indicator
Remove
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
Claim
Page
"About 11,640 households reached"
11
"Enrolment rose by 18%"
7
The QA prompt (run it on every draft)
Act as an independent reviewer.
Compare this claims table with the full report.
List every number, percentage or statement of change in the report that is missing from the table, and every claim in the table that is not quoted word for word.
Do not judge whether the claims are true.
Report: [attach]
Claims table: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
The claim on page 11 was changed: the report says "over 12,000", not "about 11,640". The AI quietly used the IPTT figure, so the error in the report would have been missed.
"Women's income doubled" (page 23) is missing from the table, so it would have gone unchecked.
Corrected output
Claim (exact quote)
Page
IPTT check
Action
"Over 12,000 households reached"
11
IPTT: 11,640
Correct the figure
"Enrolment rose by 18%"
7
IPTT: 18.2%
Keep
"Women's income doubled"
23
No indicator
Remove
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Accountability: you sign the report, so you trace every claim before you sign, whoever drafted it.
Step 6 of 6Quick check
Quick check+25 XP
The annual report goes to the donor tomorrow. How do you use AI to check it?
Show answer
Use AI to build the checklist (every quantitative claim), then check each claim against the IPTT yourself.
Try it this week
Run steps 2 to 4 on a report you already submitted. Count how many claims had no clear source: that number is your case for making this routine standard.
From the profile: an LLM classifies 3,500 community feedback messages each quarter; the team hand-checks 200 per round and the manager stops the workflow when agreement drops below the agreed threshold. Agreeing the QA sampling rate with the senior specialist is part of the manager's typical week.
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
Staff read and code thousands of messages by hand, months late.
Or AI is used informally, with nobody checking its labels.
How it works with AI
An AI model codes every message against your coding framework.
The team hand-checks a random sample each round.
Agreement is shown next to the results, and a stop rule protects quality.
Step 2 of 6Step by step
YouAgree the coding framework
Agree the categories and their definitions with the MEAL team before any AI is used.
YouAgree the sample and the stop rule
With the senior specialist, agree how many messages are hand-checked each round (200 in the profile) and the agreement level below which the workflow stops.
AICode the messagesLLM APIs
The model labels each message using the framework (prompt below). Sensitive complaints are routed to people, whatever the label.
Your teamHand-check a random samplePython (pandas) · Excel
Staff code the sample without seeing the AI labels, then compare.
AutomatedShow agreement on the dashboardPower BI
Agreement = matching codes ÷ messages checked, shown next to the results.
YouApply the stop rule
If agreement falls below the threshold, stop the workflow until the cause is fixed, even if the deadline slips.
Step 3 of 6Worked example
What you give the AI Illustrative
Message (anonymised)
AI label
Hand label
"The water point in our village has been broken for two weeks."
Service complaint
Service complaint
"Thank you for the training, it helped my business."
Positive feedback
Positive feedback
"The distribution list left out older people."
Service complaint
Exclusion concern
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
Classify each community feedback message into ONE of these categories: [paste your categories with definitions].
If a message mentions abuse, exploitation or a safety risk, label it SENSITIVE so a person reviews it.
If no category fits, write UNSURE. Return: message ID, category, one-line reason.
[paste anonymised messages with IDs]
What comes back
Sample check this round: 180 of 200 hand-checked messages match the AI label. Agreement = 180 ÷ 200 = 90%. If your agreed threshold were 92%, the workflow would stop until the cause is found. Look first at the "exclusion" messages the model labels as service complaints.
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
Sample check: 180 of 200 messages match. Agreement: 96%. The workflow can continue. Message 0412 ("a staff member asked me for money to stay on the list") is labelled Service complaint.
The QA prompt (run it on every draft)
Act as an independent QA reviewer for AI-coded feedback.
Recalculate agreement from the sample table (matching codes divided by messages checked).
List every message that mentions abuse, exploitation, fraud or safety but is not labelled SENSITIVE.
Say whether the workflow meets the agreed threshold of [X]%.
Sample table: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
Agreement is wrong: 180 ÷ 200 = 90%, not 96%.
Message 0412 describes possible exploitation. It must be labelled SENSITIVE and go to a person, not into the statistics.
"The workflow can continue" depends on your threshold. At 92%, it must stop.
Corrected output
Sample check: 180 of 200 messages match, agreement 90%, below our 92% threshold. Workflow paused while we review the codes that disagree. Message 0412 relabelled SENSITIVE and referred to the safeguarding focal point.
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Ethics: messages about abuse, exploitation or safety go to people, never only to the AI.
Step 6 of 6Quick check
Quick check+25 XP
Agreement fell below your threshold this round, and the report is due Friday. What do you do?
Show answer
Stop the workflow when agreement drops below the agreed threshold, even if the deadline slips.
Try it this week
The profile does not set a threshold. Agree one with your senior specialist this week and write it down, with the sample size per round.
From the profile: approving the cost of a SurveyCTO licence renewal and an LLM API budget for transcript analysis, and meeting the finance team to align M&E spend with the donor's revised budget.
Step 1 of 6Today, and with AI
How it is usually done today Illustrative
Licences are bought one by one, when someone asks.
Usage-based AI costs are not in the budget, so they come as a surprise.
Training is not costed at all.
How it works with AI
Every AI workflow has a budget line: licences, usage and training.
Usage costs are estimated from volume and the vendor's current price.
Costs are reviewed against time saved and error rates.
Step 2 of 6Step by step
YouList your AI workflowsExcel
For each workflow, note what it uses: a licence per user, or usage billed by volume (for example API calls for transcript analysis).
Your teamGet current prices
Collect current quotes or price pages from vendors. Prices change, so do not reuse old figures.
AIDraft the budget tableMicrosoft 365 Copilot · Excel
Ask Copilot to turn your list into a budget table with formulas (prompt below).
YouAdd training and change costs
Budget staff training for every new tool. The profile's first learning priority is change management.
YouAlign with finance and the donor budget
Meet finance to fit the AI lines into the donor's revised budget.
YouReview every quarter
Compare each workflow's cost with the hours saved and its error rate, then continue, change or stop it.
Step 3 of 6Worked example
What you give the AI Illustrative
Workflow
What it uses
Volume per year
Data collection
SurveyCTO licence
1 renewal
Transcript analysis
LLM API, billed by usage
120 interview transcripts
Report drafting
Copilot licences
4 staff
What would you ask the AI to do with this?
Think about it for a moment, then reveal.
The task prompt (copy and adapt)
Turn the list below into an Excel budget table with columns: budget line, unit, quantity, unit cost, total, notes.
Leave unit cost as [VENDOR QUOTE] for every line, and write the total as a formula (quantity x unit cost).
Add a line for staff training on each tool. Do not invent any prices.
[paste your workflow list]
What comes back
Budget line
Unit
Qty
Unit cost
Total
SurveyCTO licence renewal
year
1
[VENDOR QUOTE]
=C2*D2
LLM API: transcript analysis
transcript
120
[VENDOR QUOTE]
=C3*D3
Copilot licences
user/year
4
[VENDOR QUOTE]
=C4*D4
Staff training on AI tools
session
2
[QUOTE]
=C5*D5
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
Budget line
Qty
Unit cost
Total
Copilot licences
4
30 USD per user per month
=C2+D2
LLM API: transcript analysis
120
[VENDOR QUOTE]
=B3*C3
The QA prompt (run it on every draft)
Act as an independent budget reviewer.
Check this AI budget table.
List: any price not marked as coming from a current vendor quote, any total that is not quantity × unit cost, any workflow without a staff training line, and any unit that does not match the quantity.
Do not suggest prices.
Table: [paste]
What should the check catch?
Think about it for a moment, then reveal.
What the check finds
"30 USD per user per month" has no source: the AI made it up.
The first total adds instead of multiplying.
The unit is per month but the quantity counts users per year.
No training line for either tool.
Corrected output
Budget line
Unit
Qty
Unit cost
Total
Copilot licences
user/year
4
[VENDOR QUOTE]
=C2*D2
LLM API: transcript analysis
transcript
120
[VENDOR QUOTE]
=C3*D3
Staff training on AI tools
session
2
[QUOTE]
=C4*D4
Step 5 of 6Your checks
Tick each check as you do it on your own work. Each one earns +5 XP.
Professional development: budget for training, not only licences, so your team grows with the tools.
Step 6 of 6Quick check
Quick check+25 XP
How do you budget an LLM API used for transcript analysis?
Show answer
Usage-based AI costs need a volume estimate and a current vendor price, and training belongs in the budget too.
Try it this week
List every AI tool your team uses today, including informal use. Anything without a budget line or data rule goes on your next team meeting agenda.
Pick a recurring task that is not covered above and use this prompt to draft your own step-by-step guide. Then review it with your team.
I am an M&E manager. 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, my team, an automated flow 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 personal or sensitive 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 13 profile skills you already have and which need new learning.
Step 1 of 4Guess first
Your guess+20 XP
Which skill is the first learning priority for this role in the career profile?
Show answer
Change management, followed by workflow automation, validating AI outputs, Power Automate, and budget and resource management.
Step 2 of 4Skills you bring with you
Tick the ones you already have. Open a skill to see how to build it.
Your skills match
0 of 13
A. Existing M&E and professional competencies
Needs new learning
Bringing teams through new tools and ways of working. The profile's first learning priority: adoption fails without clear communication and training.
Build it: Write a one-page change plan for one automated workflow.
Transfers from your role
Supervising and coaching staff and consultants as their roles shift.
Build it: Add AI topics to one-to-ones: which task would each person automate first?
Transfers from your role
AI licences and API costs are now regular M&E budget lines.
Build it: Add AI lines to your next budget revision with finance.
Transfers from your role
M&E plans, IPTTs, data flow maps and reporting calendars: the base for automation.
Build it: Redraw your data flow map with automated steps and QA points.
Transfers from your role
KPIs and the RBM cycle, now also applied to the AI workflows themselves.
Build it: Set two KPIs for one workflow: error rate and hours saved.
Transfers from your role
DQAs against validity, reliability, timeliness, precision and integrity.
Build it: Include one automated workflow in your next DQA.
Transfers from your role
Logframes and results chains: the test for AI narratives' causal claims.
Build it: Check one AI-drafted narrative against your results chain.
Transfers from your role
Explaining to donors and partners how AI is used and checked.
Build it: Draft a short note to your donor on AI use in reporting.
Transfers from your role
Workplans, timelines and risk logs for automation projects.
Build it: Add AI workflows to your risk log.
Transfers from your role
Editing AI drafts for accuracy, tone and overclaiming.
Build it: Edit one AI draft and list every change you made, and why.
Step 3 of 4New skills for the AI era
These are the ones most managers still need to learn.
Your skills match
0 of 13
B. New AI competencies
Needs new learning
Checking AI results against source data, sampling for errors, keeping outputs traceable. Sign-off on automated reports needs a defensible QA process.
Build it: Run claim tracing on a past report; write your QA rule for one report.
Needs new learning
Triggers, scheduled flows and integrations. You must judge what automation can do well enough to approve and budget for it.
Build it: Learn Power Automate basics: triggers, actions, failure alerts.
D. Ethics, governance and quality assurance
Needs new learning
Consent, anonymisation, DPIAs and safe handling of personal data, applied to AI tools.
Build it: Agree AI data rules with your data protection focal point.
Step 4 of 4Tools: know them, don't build them
The profile lists no separate technical skill for this role. You need to know these tools well enough to manage them:
You don't need to code. For LLM APIs, you approve the budget (usage is paid per use) and the data rules (data sent to an API leaves your system).
End of Mission 05
MISSION06
Interactive skills gap assessment
Know your gaps · Badge: Self-Aware Leader
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 lead the transition. For every Beginner answer, start with the matching action:
Start a one-month time log of recurring tasks and score them (mission 4, guide 2).
Write a QA rule for one report: what share is checked, by whom, against what (roadmap stage 3).
Run claim tracing on a past donor report (mission 4, guide 4).
Use an AI assistant on low-risk tasks for two weeks and keep an error log (roadmap stage 1).
Build and test a library of five review prompts (roadmap stage 2).
Draw your data flow map and mark every manual step (roadmap stage 1).
Meet your data protection focal point and draft AI data rules (roadmap stage 3).
Add AI lines to your next budget revision with finance (roadmap stage 3).
Write a one-page change plan. Change management is the profile's first priority (roadmap stage 3).
Set two KPIs for one workflow and review them after one cycle (roadmap stage 4).
Mission goal: commit to a realistic 12-week plan and track it week by week.
Step 1 of 5Weeks 1–2: Understand AI
About 3–4 hours a week (a planning suggestion). Tick steps as you do them: +5 XP each. The mission is complete when every step is done, so it takes the full 12 weeks.
Stage 1 · Weeks 1–20 / 5 done
Understand AI
Build AI literacy and see where AI touches your M&E system.
Loot: an error log, a data flow map and a running time log.
Step 2 of 5Weeks 3–4: Apply AI
Stage 2 · Weeks 3–40 / 4 done
Apply AI
Practise on real management tasks with past or anonymised material.
Loot: a tested prompt library and a comparison note.
Step 3 of 5Weeks 5–8: Integrate AI
Stage 3 · Weeks 5–80 / 5 done
Integrate AI
Design the controls, rules, budget and people plan around one workflow.
Loot: a decision matrix, QA plan, data rules, budget lines and change plan.
Step 4 of 5Weeks 9–12: Demonstrate competence
Stage 4 · Weeks 9–120 / 5 done
Demonstrate competence
Show you can run AI-supported M&E responsibly.
Loot: a portfolio project, a responsible pilot and a development plan.
Step 5 of 5Check
Mission check+20 XP
In which stage do you design QA checkpoints, data rules and AI budget lines?
Show answer
Stage 3 (weeks 5–8).
End of Mission 07
MISSION08
Your practical learning project
Boss level · Badge: System Architect
1 challenge + 8 deliverables
Mission goal: produce a portfolio piece that proves you can run AI-supported reporting responsibly.
Step 1 of 6Your boss-level project
Boss level project
Build an AI-ready reporting control plan for your M&E portfolio
Objective: show you can decide what to automate, set the controls around it, fund it and bring your team along, while staying accountable for what is reported.
Case study · Integrated Food Security Programme Illustrative
No programme of your own to use? Build the plan for this one, using the examples from mission 4.
Context: use your own programme and one recurring report, such as the quarterly pack, or an anonymised copy.
Resources: your IPTT, last two quarterly reports and the M&E budget; Excel; an approved AI assistant; Power Automate or Power BI if available; your data protection policy.
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 plan?
Think about it for a moment, then reveal.
Every figure in the plan, against source.
The data rules, reviewed by your data protection focal point.
The change plan, discussed with at least two team members.
Any AI-generated text, reviewed and marked as AI-assisted.
No personal or confidential data left in the portfolio sample.
Step 5 of 6How it's judged
Criterion
What good looks like
Evidence-based choices
Automation follows the time log and error consequences, not tool features.
Defensible QA
Each automated step has a checker, a sample rule, a source and a stop rule.
Traceability
Every reported figure can be traced to source in one step.
Data protection
Rules name the data type, the tool and the condition.
Realistic resources
Budget and staff time are costed with stated assumptions.
People
The change plan names roles, training and how concerns are heard.
Learning
The memo sets measures and a decision point.
Step 6 of 6Check
Mission check+20 XP
Which part of the project must be reviewed by your data protection focal point?
Show answer
The AI data handling rules.
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 responsibilities stay with you, and the controls for each AI risk.
Step 1 of 6Would you approve this report?
Click every sentence you would not sign. There are five problems: an invented figure, an unsupported causal claim, an incorrect percentage, a missing indicator and a misleading interpretation.
Case study · Integrated Food Security Programme Illustrative
Copilot drafted part of the annual report to the donor. You are asked to sign it today.
Indicator (annual IPTT)
Result
Households reached
21,400 against a target of 25,000
3.1 Children enrolled
4,000 at baseline, 4,720 now (+18%)
2.1 Improved food consumption
+40% overall; +4% at the 18 sites reporting both years; 7 new sites
4.2 Women in savings groups
3,150 against a target of 3,000
Women's income
No indicator
Problems found: 0 of 5+40 XP
Show answer
Five problems: the 28% (it is 18%), the invented income claim, the unsupported causal claim about training, the misleading "works everywhere" reading of the 40%, and the claim that every indicator is reported (4.2 is missing).
Step 2 of 6Your decision
Your decision+20 XP
You found the problems. Can you approve this report?
Show answer
Return it for correction. Fix each problem, recheck every figure against the IPTT, record the check, then approve.
Use the QA checklist on every AI-supported report:
The report is corrected. Before you sign it, answer the five questions. Tick each one you can honestly answer yes to.
Step 4 of 6Mini-game: delegate or never?
AI can draft, sort and summarise. It cannot be accountable. For each task: can AI help (with checks), or must you never delegate it?
Delegate?+10 XP
Listing every quantitative claim in a donor report
Show answer
AI can help. AI builds the checklist; you check each claim against the IPTT.
Delegate?+10 XP
Signing off figures sent to the donor
Show answer
Never delegate. You sign and you accept accountability.
Delegate?+10 XP
Classifying 3,500 feedback messages, with hand-checked samples
Show answer
AI can help. With a sample check and a stop rule, this is a good use of AI.
Delegate?+10 XP
Deciding whether a 40% jump is real
Show answer
Never delegate. AI can suggest explanations; the data and the field team decide.
Delegate?+10 XP
Choosing what personal data may go into which tool
Show answer
Never delegate. Agreed with your data protection focal point, never left to a tool.
Delegate?+10 XP
Drafting the first version of a management brief
Show answer
AI can help. Fine as a first draft that you edit for accuracy, tone and overclaiming.
Delegate?+10 XP
Deciding which staff roles change
Show answer
Never delegate. A management decision, made with the people affected.
Delegate?+10 XP
Stopping a workflow that falls below the quality threshold
Show answer
Never delegate. Your rule to enforce, even when a deadline slips.
Step 5 of 6Six risks, six controls
What can go wrong when AI helps with your reporting? Name three risks before you look.
Think about it for a moment, then reveal.
Risk
What it looks like
Your control
Hallucination
A figure or claim with no source
Claim tracing; sample checks
Overclaiming
A change called a success without evidence
Review tone and causal claims
Bias
AI misreads feedback from some groups or languages
Hand-check samples across groups and languages
Privacy
Personal data pasted into a public tool
Agreed data rules; approved tools only
Silent failure
A flow runs on missing or changed data
Failure alerts; completeness checks
Transparency
Readers don't know AI was used
An AI-use statement in reports
Step 6 of 6Tell readers how AI was used
Transparency is a control too. Add a short statement to reports that used AI, and adapt it to what you actually did.
AI tools were used to consolidate monitoring data and to draft parts of the narrative. All figures were checked against the indicator tracking table, a sample of [X]% was verified against source data, and the final text was reviewed and approved by the M&E Manager.
Agree which data may go into which tool before you start:
The AI-Enabled M&E Specialist is covered in Tutorial 02, the AI-Enabled Senior M&E Specialist in Tutorial 03, the AI-Enabled M&E Coordinator in Tutorial 05, the Director of AI-Enabled M&E in Tutorial 06, the AI-Powered Monitoring Specialist in Tutorial 07, the AI-Enabled Evaluation Manager in Tutorial 09, the AI-Driven Programme Performance Specialist in Tutorial 10, the AI-Enabled Results-Based Management Specialist in Tutorial 11, the AI-Powered Performance Measurement Specialist in Tutorial 12 and the AI-Enabled MEAL Manager in Tutorial 15.
Step 2 of 5What you need to show
Strengths you build on
Owning M&E systems and reporting calendars
DQA experience, the basis of AI validation
Budget, procurement and team management
Judgement about credible results
What you need to show
A documented automation decision
A QA process you designed and enforced
AI data rules you agreed
AI budget lines you planned
A team you brought through change
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
Led the redesign of [quarterly reporting for 6 projects], automating [data consolidation and formatting] while keeping indicator interpretation with staff.
Introduced QA checkpoints for automated reports, including [10%] source checks and claim tracing before donor sign-off.
Agreed AI data handling rules with the data protection focal point and budgeted for AI licences, API usage and training.
Managed the team through role changes linked to automation, including [training plan / reskilling].
Replace the brackets with what you have actually done.
Qualifications in the profile
Master's degree in development studies, public policy, statistics, economics or a related field.
7–12 years in M&E, including at least three years managing an M&E team and budget.
Useful certifications: MEAL DPro (PM4NGOs); PMD Pro (PM4NGOs).
Step 5 of 5Check
Mission check+20 XP
What's the best way to use "AI-Enabled M&E Manager" 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 AI-ready reporting control plan from mission 8, focused on your quarterly performance pack.
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-Enabled M&E Manager do?
Manages the M&E function for a country programme or a large multi-project portfolio, including the team, the M&E budget and the systems that produce performance data. Oversees AI-powered monitoring and automated reporting, supervises the quality of AI-supported analysis and makes sure programme performance is reported accurately and on time.
How is AI changing the M&E Manager role?
The manager now runs a system in which much of the routine data processing and report drafting is automated, so the main job becomes setting controls, staffing and budgets around those workflows. The manager is accountable for what the automated system reports and must know where it fails.
Which judgements should an AI-Enabled M&E Manager never hand over to AI?
The career profile lists these as human judgement:
Signing off figures and narratives sent to donors and accepting accountability for them.
Deciding how to allocate staff time and budget between automation and fieldwork.
Managing people through changing roles and new tools.
Handling difficult performance conversations with programme teams and partners.
Which skills should I learn first?
The learning priorities in the career profile, in order:
Change management. AI changes staff roles, and adoption fails without clear communication and training.
Workflow automation. The manager needs to judge what automation can do well enough to approve and budget for it.
Validating AI outputs. Sign-off on automated reports requires a defensible QA process.
Power Automate. Many organizations on Microsoft 365 build reporting automation in this tool.
Budget and resource management. AI licences and API costs are now regular M&E budget lines.
The profile lists 13 skills in all: Team leadership, M&E planning, Budget and resource management, Performance and results-based management, Data quality assessment, Validating AI outputs, Change management, Workflow automation, Results frameworks and theory of change, Stakeholder engagement, Data protection and privacy, Project management and Analytical report writing.
Which tools does this role use?
The career profile names: Power BI, Power Automate, Microsoft 365 Copilot, SharePoint, Excel, SurveyCTO, KoboToolbox, Claude, LLM APIs (OpenAI, Anthropic, Google) and ActivityInfo. The tutorial shows how to check every AI output, whichever tool you use.
What qualifications and experience does the profile list?
Master's degree in development studies, public policy, statistics, economics or a related field.
7–12 years in M&E, including at least three years managing an M&E team and budget.
Useful certification: MEAL DPro (PM4NGOs) and PMD Pro (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-Enabled M&E Manager" 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.
