Why This Matters for M&E

NotebookLM is particularly useful when your work involves analysing a defined collection of documents and evidence.

For M&E professionals, that might include:

• evaluation reports;
• research papers;
• baseline and endline studies;
• MEL plans;
• Theories of Change;
• programme proposals;
• donor reports;
• monitoring reports;
• interview transcripts;
• survey documentation;
• policy documents;
• learning reports.

The temptation is to upload the documents and immediately start asking questions.

What You Will Learn

• Why isolated prompts become limiting for recurring M&E work.
• What an analytical protocol is.
• How to structure NotebookLM work into four phases.
• How to analyse evaluation reports systematically.
• How to compare multiple sources.
• How to identify unsupported claims and evidence gaps.
• How to challenge assumptions and methodology.
• How to build evidence-synthesis workflows.
• How to create reusable protocols for your own M&E work.

1. Prompt vs. Protocol

A prompt is a request.

A protocol is a sequence of analytical steps.

2. The Four-Phase M&E Protocol

A practical starting point is:

Discovery

Understand what the sources contain before making judgments.

Comparison

Identify agreement, disagreement, convergence, and divergence.

Challenge

Test claims, evidence, assumptions, methodology, and alternative explanations.

Synthesis

Form conclusions from the evidence that remains after examination.

3. Phase 1 — Discovery

Discovery is the orientation phase. The objective is to create a structured map of the source.

For an evaluation report, identify:

• evaluation purpose;
• evaluation questions;
• programme context;
• methodology;
• data sources;
• major findings;
• conclusions;
• recommendations;
• limitations;
• important assumptions.

Discovery Prompt

Analyse this evaluation report.
Identify:
1. Purpose and scope
2. Evaluation questions
3. Methodology
4. Data sources
5. Major findings
6. Conclusions
7. Recommendations
8. Limitations
9. Important assumptions
For each major finding, identify the evidence
used to support it and cite the relevant source.

4. Discovery for Different M&E Documents

Evaluation Report

Map purpose, questions, methodology, findings, conclusions, recommendations, evidence, and limitations.

MEL Plan

Map objectives, results, indicators, definitions, targets, data sources, collection methods, responsibilities, frequency, assumptions, and risks.

Theory of Change

Map inputs, activities, outputs, outcomes, impact, causal pathways, assumptions, contextual factors, and supporting evidence.

Donor Report

Map achievements, targets, evidence, deviations, explanations, lessons, recommendations, and unresolved issues.

Research Paper

Map the research question, methodology, sample, analysis, findings, limitations, and conclusions.

5. Phase 2 — Comparison

Once you understand the sources, compare them deliberately.

Useful comparison questions include:

• Where do the sources agree?
• Where do they disagree?
• Which findings appear across multiple sources?
• Which claims depend on a single source?
• What information appears in one source but not another?
• Are different sources using the same evidence differently?
• What contradictions remain unresolved?

Comparison Prompt

Compare the major findings across the selected sources.
For each important finding:
- identify areas of agreement;
- identify areas of disagreement;
- identify which sources support it;
- identify important differences in interpretation;
- identify unresolved contradictions.
Cite the relevant evidence.

6. Practical Example — Programme Report vs. Evaluation

Imagine you have:

Do not immediately ask:

Instead:

1. Identify the achievements reported in Source A.
2. Identify the major findings in Source B.
3. Compare the reported achievements with
   the evaluation findings.
4. Identify areas of agreement.
5. Identify where the evaluation qualifies,
   challenges, or contradicts the programme report.
6. Identify unresolved differences.
7. Identify what additional evidence could
   help resolve those differences.

7. Phase 3 — Challenge

Understanding a source does not mean accepting its conclusions.

Ask NotebookLM to examine:

• unsupported claims;
• weak evidence;
• hidden assumptions;
• contradictions;
• methodological limitations;
• alternative explanations;
• missing perspectives;
• evidence gaps;
• overstated conclusions.

8. Challenge the Evidence

For each major finding:
1. What evidence supports the finding?
2. Is the evidence sufficient for the claim?
3. What limitations affect the evidence?
4. What alternative explanations are possible?
5. What additional evidence would strengthen it?
Cite the relevant source for each assessment.

9. Challenge the Assumptions

M&E documents often contain assumptions that are not immediately visible.

For example:

That relationship may depend on several assumptions.

Identify the causal assumptions underlying
the conclusions in this source.
For each assumption:
- state the assumption;
- identify supporting evidence;
- identify evidence challenging it;
- identify whether it remains untested.

This is especially useful for:

• Theories of Change;
• programme logic;
• causal claims;
• contribution claims;
• outcome pathways.

10. Challenge the Methodology

Identify the main methodological choices
made in this evaluation.
For each important choice:
- explain its purpose;
- identify strengths;
- identify limitations;
- explain how it may affect interpretation
  of the findings.
Use evidence from the source.

The objective is not to ask NotebookLM to declare an evaluation “good” or “bad”.

The objective is to make the reasoning behind the assessment visible.

11. Find the Missing Evidence

One of the most useful challenge questions is:

For example, a programme may claim that training improved employment outcomes while providing evidence only of:

• participation;
• training completion;
• satisfaction.

The missing evidence may be the actual employment outcome.

12. Phase 4 — Synthesis

Only after discovery, comparison, and challenge should you move to synthesis.

1. Identify the strongest findings supported
   across the sources.
2. Identify findings where evidence is mixed
   or contradictory.
3. Identify conclusions that depend on weak
   or single-source evidence.
4. Identify the most important evidence gaps.
5. Identify conclusions that can be made
   with relatively high confidence.
6. Identify conclusions that should be
   treated cautiously.
7. Identify implications for programme
   decision-making.
Cite the evidence supporting each conclusion.

13. Build a Complete Evaluation Review Protocol

14. Protocols for Other M&E Tasks

Evidence Review

Research question → Relevance → Claims → Evidence → Methodology → Agreement → Contradictions → Gaps → Synthesis

Theory of Change Review

Pathways → Causal assumptions → Evidence → Context → Risks → Alternatives → Gaps

MEL Plan Review

Objectives → Results → Indicators → Definitions → Targets → Data → Responsibilities → Frequency → Risks → Gaps

Donor Report Review

Achievement → Target → Evidence → Variance → Explanation → Documentation → Missing evidence → Credibility

Research Paper Review

Question → Method → Sample → Analysis → Findings → Limitations → Conclusions → Applicability

15. Use Source Selection Strategically

Do not automatically ask every question against every source.

For example:

16. Treat Citations as a Verification Step

When NotebookLM provides a citation for an important claim, do not simply accept it because a citation is present.

Use the citation as a verification mechanism:

This is particularly important for:

• evaluation findings;
• causal claims;
• recommendations;
• evidence syntheses;
• statements about programme effectiveness.

17. Build an Evidence Matrix

For more systematic evidence work, ask NotebookLM to organize your findings into an evidence matrix.

Create an evidence matrix with these columns:
1. Claim
2. Source
3. Supporting evidence
4. Evidence limitations
5. Contradictory evidence
6. Important assumptions
7. Evidence gaps
8. Implication for the conclusion
Use citations for source-based entries.

This can be useful for:

• evaluation synthesis;
• literature reviews;
• theory testing;
• programme learning;
• donor evidence reviews;
• policy analysis.

18. Add a Confidence Check

After synthesis, explicitly examine how strongly each conclusion is supported.

For each major conclusion:
- What evidence supports it?
- How many independent sources support it?
- Are there contradictory findings?
- What limitations remain?
- What assumptions does it depend on?
- What information could change the conclusion?
Classify it as:
High confidence
Moderate confidence
Low confidence
Explain the reasoning and cite the sources.

19. Turn the Protocol Into a Reusable Asset

Once you have a protocol that works, save it and improve it.

A reusable protocol should contain:

• purpose;
• required sources;
• analytical phases;
• prompts for each phase;
• expected outputs;
• quality checks;
• human review points.

20. Improve the Protocol Over Time

After using your protocol on several projects, ask:

• Which questions consistently produced useful findings?
• Which questions produced little value?
• Which evidence gaps did we repeatedly miss?
• Which assumptions should be added?
• Which stakeholder perspectives are missing?
• Where does AI frequently overinterpret?
• Where is human review most important?

21. Practical Exercise — Review an Evaluation

22. Build Your Own M&E Protocol

Choose one recurring task:

• evaluation report review;
• evidence synthesis;
• research paper analysis;
• Theory of Change review;
• MEL plan review;
• donor report review;
• indicator review;
• programme document analysis.

Then define:

PROTOCOL NAME:
[Name of workflow]
PURPOSE:
[What should this protocol accomplish?]
SOURCES:
[What documents or evidence are required?]
DISCOVERY:
[What must be understood first?]
COMPARISON:
[What should be compared?]
CHALLENGE:
[What should be tested?]
SYNTHESIS:
[What conclusions should be produced?]
QUALITY CHECK:
[What must be verified?]
HUMAN REVIEW:
[What requires professional judgment?]
FINAL OUTPUT:
[What should the completed analysis contain?]

23. NotebookLM Is Not the Evaluator

AI can help you locate evidence, compare sources, identify patterns, surface contradictions, and organize information.

M&E professionals still need to determine:

• whether evidence is credible;
• whether a causal inference is justified;
• whether findings are contextually appropriate;
• whether evidence is sufficient for a decision;
• whether a conclusion is defensible.

The objective is to strengthen professional judgment, not replace it.

24. Privacy and Responsible Use

Before uploading programme documents, check whether they contain information you are authorized to share with the service.

Take particular care with:

• personally identifiable information;
• beneficiary records;
• interview transcripts;
• case-management information;
• confidential donor documents;
• sensitive assessment material;
• internal programme records.

25. The Bigger Shift

At first, NotebookLM is something you interact with.

You upload a source and ask questions.

With more experience, you can begin designing a method of inquiry around those questions.

You can improve that method each time you use it.

You can identify questions that consistently produce useful findings, add missing challenge steps, improve evidence checks, and create specialized protocols for different M&E tasks.