AI and Local Data
Using AI to Analyse Local Data for Climate-Resilient Water Decisions
A practical M&E learning scenario based on the Fredericton, Canada – Port St. Johns, South Africa case documented by the OECD.
01 The case
The OECD describes a partnership between the city of Fredericton, Canada, and the city of Port St. Johns, South Africa. Fredericton dispatched specialists from its municipal water utility to support Port St. Johns in assessing and managing climate risks to water-system assets.
The specialists used local data and helped map and visualise it using GIS technology. This supported communication with local authorities and politicians and helped make the case for investment in the city’s water infrastructure. The partnership also helped reduce silos between municipal divisions.
After the intervention, specialists in Port St. Johns had the capacity to independently manage and visualise their data and maintained improved working relationships with other parts of the administration to support climate resilience of the local water system.
02 Where does this case fit in M&E?
The case is particularly relevant when an evaluator needs to examine whether technical support and knowledge exchange have contributed to capacity, data use, planning and longer-term results.
Technical support and peer learning
Track data use, capacity and collaboration
Assess effectiveness and sustainability
Use evidence for planning and investment discussions
The OECD also reports that M&E systems among local and regional governments often use qualitative evidence such as activity reports and lessons learned, while surveys, interviews and feedback sessions are also widely used. This makes the case relevant to evaluators working across both qualitative and quantitative evidence.
03 The M&E question
To what extent did the partnership strengthen the capacity of Port St. Johns to use local data to understand and manage climate risks to its water-system assets?
This question is an EvalCommunity learning formulation. It is not presented by the OECD as the original evaluation question.
04 What evidence would an evaluator examine?
Start with the evidence that is actually relevant to the case:
- Local data used to understand the water-system assets and climate risks.
- Maps and GIS visualisations developed during the work.
- Evidence of technical support provided by Fredericton specialists.
- Evidence of whether Port St. Johns specialists can independently manage and visualise their data.
- Evidence of working relationships between the water utility and other municipal divisions.
- Evidence of how climate-risk information was communicated to local authorities and politicians.
- Evidence relevant to decisions or proposals concerning investment in water infrastructure.
05 A simple results pathway
Fredericton water specialists
Support to Port St. Johns
Mapping and visualisation
Independent data use
Climate-resilient water planning
Evaluation caution: This pathway is a learning framework constructed from the case evidence. It should not be treated as proof that each step caused the next.
06 Where AI can support the evaluator
The following are proposed AI applications for an evaluator, not activities reported in the OECD case.
| M&E task | Possible use of AI | Human responsibility |
|---|---|---|
| Document review | Organise large volumes of project documents and identify passages related to capacity, data use, collaboration and decisions. | Check the original documents and confirm the evidence. |
| Qualitative analysis | Assist with an initial coding structure for interview notes or other qualitative evidence. | Review coding, context and alternative interpretations. |
| Local data analysis | Help explore structured datasets, identify missing values or unusual patterns, and prepare questions for further investigation. | Validate data, methods and interpretation. |
| Evidence synthesis | Compare findings from documents, interviews, data and observations against evaluation questions. | Triangulate evidence and judge its strength. |
| Reporting | Help organise evidence into a clear findings structure and draft summaries for review. | Own the findings, conclusions and recommendations. |
07 Example: analysing local data with AI
Imagine that the evaluator receives a dataset used to understand the water-system assets and climate risks. The evaluator could ask an AI system to help:
- Describe the variables and data structure.
- Identify missing or inconsistent records.
- Group the data into useful categories for review.
- Suggest patterns that deserve investigation.
- Generate a list of questions that should be checked against GIS information and other evidence.
08 Triangulating the evidence
A useful evaluation would not rely on AI analysis alone. For this case, the evaluator could compare several forms of evidence:
Partnership records, technical materials and planning evidence.
Interviews with municipal and water-utility staff.
Local data, maps and visualisations.
Can staff independently manage and visualise the data?
The OECD case specifically reports improved capacity for specialists in Port St. Johns to independently manage and visualise their data, as well as improved working relationships across the administration. An evaluator should therefore test these claims against the available evidence rather than assuming that the presence of technical assistance automatically produced lasting capacity.
09 A practical evaluator workflow
Clarify what change you are assessing.
Identify data, documents, people and practices.
Check quality, structure and permissions before AI use.
Use AI for organisation, exploration and first-pass analysis.
Check AI outputs against source evidence.
Compare data with documents, interviews and practice.
Explain what the evidence supports—and what it does not.
Present evidence clearly to decision-makers.
10 Learner challenge
You are the evaluator supporting a review of the Fredericton–Port St. Johns partnership. You have access to project documentation, local data and GIS outputs, interviews with relevant staff, and evidence about how the information was used.
Design a short evaluation approach that answers:
- What evidence would show strengthened local capacity?
- How would you assess whether staff can independently manage and visualise data?
- How would you examine changes in collaboration between municipal divisions?
- How would you assess whether climate-risk information supported communication and investment discussions?
- Where could AI reduce analytical workload without replacing evaluator judgement?
- What evidence would you require before making a conclusion?
11 The key M&E lesson
The interesting evaluation question is not simply whether technical support was delivered. It is whether the partnership strengthened the ability of the local system to use knowledge and data after the intervention.
The OECD case is useful because the reported results include independent management and visualisation of data and improved relationships across municipal administration. For an evaluator, this shifts attention from the activity itself toward capacity and how evidence is used.
Further reading and practical resources
- OECD — The Impact of Decentralised Development Co-operation: Driving Local Governance and the SDGs — main source for this case.
- EvalCommunity Academy — Principles for AI Use in Monitoring & Evaluation — practical guidance on purpose, human judgement and responsible AI use.
- EvalCommunity Academy — The 7 Layers of AI for Monitoring & Evaluation Professionals — helps match AI capabilities to concrete M&E tasks.
- EvalCommunity Academy — AI in Monitoring & Evaluation Course — practical, no-code learning for applying AI across M&E workflows.
Ready to apply AI to your M&E work?
Go beyond individual AI tools. Learn how to use AI for evaluation design, evidence analysis, qualitative and quantitative work, reporting, data quality and responsible M&E practice.
Enroll in AI in Monitoring & Evaluation
Explore EvalCommunity Academy courses for M&E, AI and international development professionals.
Primary source: OECD (2026), The Impact of Decentralised Development Co-operation: Driving Local Governance and the SDGs, especially Box 4.4, “Impact of DDC activities in partner LRGs across selected case studies,” and the report’s discussion of DDC M&E systems. The Fredericton–Port St. Johns case is reproduced here only in summary form for educational purposes. AI uses described in this learning case are EvalCommunity-proposed applications and are not presented as technologies used in the original OECD partnership. Where the OECD reports quantitative relationships elsewhere in the report, it cautions that these should be interpreted as associations rather than causal effects.
