Week 9 of 12 · 12 hours · Diploma

Steering with MEAL: indicators, monitoring, dashboards and AI

Objectives of the week
  • Build a MEAL plan linked to the logframe: SMART indicators, baselines, targets, disaggregation, sources and frequencies.
  • Track progress with an indicator tracking table (ITT) and a dashboard, and turn them into decisions.
  • Use AI to analyse quantitative and qualitative data while respecting humanitarian data responsibility.

Keep the course at home: free PDF handout with exercises, key takeaways and quiz answers.

Download this lesson as PDF Complete manual (PDF)

1The MEAL plan, backbone of project steering

MEAL stands for Monitoring, Evaluation, Accountability and Learning. The MEAL plan turns the logframe into a concrete system. For each indicator: precise definition, calculation method, baseline, target (total and per period), disaggregation (sex, age, disability, location), source and collection method, frequency, owner, intended use.

A SMART indicator is specific, measurable, achievable, relevant and time-bound. "Improve access to water" is not an indicator; "% of target households with at least 15 litres of drinking water per person per day (Sphere standard), by December 2027" is.

The MEAL plan also covers evaluations (baseline, mid-term, final), the feedback and complaints mechanism and learning moments (quarterly reviews, lessons-learned workshops). The PM4NGOs MEAL DPro guide is a free reference.

2The indicator tracking table (ITT)

The ITT (Indicator Tracking Table) shows, for each indicator, the baseline, target, achievements by period, cumulative total and achievement rate. Donors often require it with each report.

Reading an ITT means comparing the achievement rate with elapsed time. In quarter 3 of 4 (75% of the time), an indicator at 45% calls for analysis: a late start that can be caught up, an access problem, an unrealistic target, or a poorly defined indicator?

Watch out for frequent mistakes: adding up people counted several times (one family receives three distributions: that is 3 distributions, but a single family); mixing people and households; presenting a percentage without the denominator; forgetting disaggregation by sex, age and disability.

3From monitoring to decisions

Monitoring is useless if it does not change decisions. An effective one-hour monthly performance review examines, in order:

  1. outcome indicators and community feedback: is change happening?
  2. outputs and the schedule: are we on time?
  3. the budget and earned value;
  4. risks and issues;
  5. decisions: who does what, by when.

The one-page dashboard combines these elements with colour codes (green, amber, red) defined in advance. For example: green if achievement is at least 90% of the time-based pro rata, amber from 70 to 89%, red below. Written thresholds avoid subjective debates.

4Analysing data with AI

Quantitative data. AI can write spreadsheet formulas, clean a database (duplicates, outliers), calculate rates and disaggregations, propose charts and a first commentary. It can also run small calculations in versions with an analysis tool. Always check totals with an independent calculation.

Qualitative data. For hundreds of open answers, focus group notes or community feedback, AI can propose themes, count occurrences and extract representative quotes. Good practice: define the coding frame yourself or validate it; manually check a sample (for example 10% of answers); keep minority voices, which automated summaries tend to erase.

What AI must not do: process sensitive complaints (sexual exploitation and abuse, gender-based violence, fraud) outside the planned confidential channel; decide alone on a complaint's category; produce figures without a verifiable database.

5Humanitarian data responsibility

The IASC Operational Guidance on Data Responsibility (2023) and the ICRC Handbook on Data Protection set the framework. Before entering data into an AI tool:

  • Minimise: only the columns needed;
  • Truly anonymise: remove names, phone numbers, registration numbers, precise GPS coordinates, and check that combinations (village + age + disability + household composition) do not make anyone identifiable;
  • Think about demographically sensitive data: even without names, the location of an ethnic group or of displaced people can expose them;
  • Use the approved offer of your organisation, whose terms exclude the use of your data for training;
  • Respect the consent given at collection.

Simple rule: if a leak of this file could harm a person or a group, it does not go into a non-approved AI tool, even "just for a summary".

6Learning and adapting

The "L" in MEAL is often forgotten. For a project to learn:

  • keep an adaptive decision log: date, finding, data, decision, approval;
  • hold short after-action reviews after each major stage (what was planned? what happened? why? what to repeat or change?);
  • share lessons with the cluster and partners;
  • check the assumptions of the theory of change at least once every six months.

AI can help synthesise reviews and find lessons from past projects in your archived reports, provided it cites the source documents.

Exercise: reading an ITT and deciding

Situation. Child protection project of the fictitious NGO Amal, 12 months. Indicator: "number of children receiving individual case management". Annual target: 2,400 children. Achievements: Q1 450; Q2 620; Q3 530. Cumulative disaggregation: 58% boys, 42% girls, whereas the target population is 51% girls. Dashboard thresholds: green ≥ 90% of the time-based pro rata, amber 70 to 89%, red < 70%.

Task. 1) Cumulative total and achievement rate at the end of Q3? 2) Dashboard colour? 3) How many children in Q4 to reach the target? 4) What question does the disaggregation raise?

Answer. 1) 450 + 620 + 530 = 1,600 children; 1,600 ÷ 2,400 ≈ 66.7%. 2) Time-based pro rata at the end of Q3 = 75%; 66.7 ÷ 75 ≈ 89%: amber (just below green). 3) 2,400 − 1,600 = 800 children in Q4, well above the best quarter (620): actual capacity must be analysed before promising the target. 4) Girls are under-represented (42% against 51%): access barriers specific to girls? Unsuitable reporting channels? This must be discussed with communities and the team, without artificially pushing the figures, because case management must respond to a real need.

Key points

  • The MEAL plan details each indicator: definition, baseline, target, disaggregation, source, frequency, owner.
  • Reading an ITT: compare the achievement rate with elapsed time; avoid double counting.
  • A monthly review starts with outcomes and ends with written decisions.
  • AI analyses anonymised data and feedback; check a sample and keep minority voices.
  • Minimise, anonymise, approved tool, consent: no potentially harmful data in a non-approved tool.
A question about this module?

Ask the other participants: they are taking the same course as you.

Ask my question See this module's discussions

💼 Programme management jobs

Put this course into practice: the latest jobs in this field.

See all 1171 Programme management jobs →

Prepare for the written test and interviewA model written test in this field, interview questions with sample answers and the STAR method.See the model test →
Further reading: course references
  1. Project Management Institute (PMI) (2025). A Guide to the Project Management Body of Knowledge (PMBOK Guide), 8th edition. www.pmi.org
  2. Project Management Institute (PMI) (2026). PMP Examination Content Outline. www.pmi.org/certifications/project-management-pmp
  3. PeopleCert (AXELOS) (2023). Managing Successful Projects with PRINCE2, 7th edition. www.peoplecert.org
  4. Ken Schwaber and Jeff Sutherland (2020). The Scrum Guide. scrumguides.org
  5. Group of 17 signatories (2001). Manifesto for Agile Software Development. agilemanifesto.org
  6. PM4NGOs (2020). Project DPro: project management guide for development professionals (PMD Pro 2nd edition). pm4ngos.org
  7. PM4NGOs (2019). MEAL DPro: guide to monitoring, evaluation, accountability and learning. pm4ngos.org
  8. European Commission (2004). Project Cycle Management Guidelines. international-partnerships.ec.europa.eu
  9. United Nations Sustainable Development Group (UNSDG) (2011). Results-Based Management Handbook. unsdg.un.org
  10. CHS Alliance, Groupe URD and Sphere (2024). Core Humanitarian Standard on Quality and Accountability. www.corehumanitarianstandard.org
  11. Sphere Association (2018). The Sphere Handbook. spherestandards.org
  12. Inter-Agency Standing Committee (IASC) (2023). Operational Guidance on Data Responsibility in Humanitarian Action. interagencystandingcommittee.org
  13. ICRC and Brussels Privacy Hub (2020). Handbook on Data Protection in Humanitarian Action (2nd edition). www.icrc.org
  14. National Institute of Standards and Technology (NIST) (2023). AI Risk Management Framework (AI RMF 1.0). www.nist.gov
  15. United Nations (Chief Executives Board) (2022). Principles for the Ethical Use of Artificial Intelligence in the United Nations System. unsceb.org

Weekly quiz

5 demanding questions to check your learning. Answer them all to complete the week.

1Target 1,000, 600 achieved at mid-project. Compared with the time-based pro rata (50%), the project is:
2One family receives three distributions. It counts as:
3Before analysing community feedback with AI, you must first:
4A complaint of sexual exploitation and abuse must be handled:
5The "L" in MEAL stands for:

← Week 8Week 10 →