A water treatment plant can be finished on the day the contractor hands over the keys. A reader of a development project report still has several questions. Is the plant working at the expected level? Who receives water through the network? Can the operator maintain the equipment in the years after handover? Those questions concern different kinds of evidence. The fact that the building exists is an important starting point.
Japan International Cooperation Agency, or JICA, publishes evaluations that give readers a way to follow these questions after a project has ended. This guide explains how to read one without losing track of dates, indicators and the scope of a conclusion. It uses a historical evaluation of the Niroth Water Supply Project in Cambodia and a separate fictional example. Neither example describes a current water service advisory.
Identify the project and the stage of the report
Begin with the cover. Record the project name, country, type of assistance and the evaluation year. Then look for the period in which the work took place and the dates of the observations. A fiscal year in an evaluation title can be different from the year the project was approved, completed or actually examined. Putting those dates in one short timeline makes it harder to read an old observation as a current condition.
JICA’s description of post-implementation evaluation explains that its ex-post work examines completed projects. It distinguishes evaluation from later monitoring, which can revisit certain results and recommendations. The page also describes external and internal approaches. Check the report itself for who conducted the evaluation and which approach it used. Knowing how the document was produced gives you a starting point for reading its evidence.
An appraisal describes what was expected before the work. An ex-post evaluation can compare those expectations with evidence collected after completion. A later monitoring exercise can ask whether effects continued or recommendations were followed. Keep the three tasks separate in your notes. If you only have an appraisal, you can describe the plan. To describe what happened, locate a report that actually provides observations.
When there are several phases, write down which one the report covers. An evaluation of one treatment plant may mention an earlier distribution project or another phase of investment. Those references can explain the setting. They do not necessarily put every facility in the scope of the evaluation. Look for the project boundary before combining numbers from adjacent reports.
Use the rating to find the questions behind it
JICA’s 2024 annual evaluation report overview describes six criteria adapted from the OECD Development Assistance Committee framework. It groups relevance and coherence together, effectiveness and impact together, and considers efficiency and sustainability as separate rating items. Historical evaluations may use the framework in effect at the time. Check the method named in the report before comparing ratings across years.
For a reader, relevance can prompt a question about the problem the project was meant to address. Coherence can prompt a question about how the intervention fit with other policies or work. Effectiveness asks how far the intended results were realised. Impact directs attention to broader consequences. Efficiency brings time and resources into the discussion. Sustainability asks whether the benefits can continue. These prompts help you find the relevant evidence even when an evaluation groups the criteria together.
An overall rating is a short summary. Read the separate judgements and their explanations before trying to use it. A project can show substantial progress against service goals while taking longer than expected. A report can also identify an operational concern even where the installed equipment is in use. Keep the assessment of each question visible rather than letting a single overall word stand in for the whole report.
You can make a four-column reading note. Put the criterion in the first column, the report's conclusion in the second, the strongest piece of evidence in the third and an open question in the fourth. Use a page number or table label in the evidence column. The open question can be as simple as whether the latest reading covers the same area as the original target. This note is short enough to review before you cite the report.
Trace one indicator from the plan to the observation
Choose an indicator and read across its row. Note its unit, the baseline if one is given, the target and the reported result. Then identify the time and location the result covers. A daily volume at a treatment plant and a count of connected households are both useful, but they describe different stages between production and service.
Imagine a fictional project expected a plant to produce 8,000 cubic metres a day. The evaluation reports an average of 7,600 for a particular period. Dividing the reported figure by the target yields 95 percent. That comparison is meaningful if the target and observed average use the same scope and time basis. It tells you how the stated production indicator compares with its target. It does not tell you that 95 percent of households were connected.
Suppose the same fictional evaluation also reports 3,000 new connections. Before describing the change, read the definition. Does it mean installed connections, active customer accounts or households reached? If the number includes businesses, say so if the report gives that detail. If the definition is unclear, preserve the report’s wording and leave the more specific claim open. Translating a figure into plain English should preserve what was counted.
When the baseline is missing, you can still report a comparison with a target if the two measures match. A claim about improvement from the original situation requires information about that original situation. A blank in a table is a gap in evidence, and the surrounding notes may explain why it exists.
Read the Niroth example by separating the findings
The FY 2016 external evaluation of the Niroth Water Supply Project examines a Japanese official development assistance loan project for water supply in Phnom Penh and surrounding areas. Its summary says the cost slightly exceeded the plan and the project period exceeded the plan, leading the evaluators to judge efficiency fair. Separately, it describes broad achievement of the planned operation and effect indicators, with an exception concerning the water colour indicator.
That exception is worth reading for the reasoning, not just the mark in a table. The evaluators discuss the basis for the original colour target and explain why they did not treat the difference as decisive for effectiveness. A careful account can state both that the target was missed and that the report explains how it interpreted the miss. A claim that all targets were met would erase a documented exception. An assertion about current water quality would ask the historical report to answer a different question.
Notice how the findings can coexist. The project period can exceed the original plan while the operation indicators later show substantial achievement. These are observations from different parts of the evaluation. A reader interested in delivery should follow the cost and schedule comparison. A reader interested in service should look at the actual indicators and how they were measured. The summary is a map to those passages.
For comparison, Humanitarian Aid Funding Figures discusses the evidence behind another type of international development figure. An appeal and a completed project evaluation answer different questions. The shared reading skill is to check what the document is trying to measure before carrying its headline number into a wider argument.
Look for the evidence that benefits can continue
Sustainability can be difficult to describe with a single observation. Find out who is expected to operate the project after completion. Look for the report's discussion of staffing, maintenance arrangements and resources where those matters are relevant. A recent observation that a facility is in use is valuable. The ability to repair it when a component fails is a different piece of evidence.
Imagine the fictional water plant is producing close to its target but a replacement part takes months to obtain. The output figure may still be accurate for the measured period. A reader interested in future operation would want to see what the report says about the replacement process and whether there is an alternative in place. A maintenance plan and evidence that it has been carried out answer different parts of that question.
When a report recommends an action, record who it addresses and what would show progress. “Improve maintenance” could refer to a schedule, a training need or a purchase arrangement. If the evaluation specifies the problem, keep it in the note. A later review can then look for the actual change. The recommendation itself is evidence of the evaluator’s judgement, and follow-up records can help establish what happened next.
Check how much of the result the project can explain
Return to the fictional plant and imagine that the surrounding city has also replaced several kilometres of distribution pipes. If interruptions fall during the year after the plant opens, both changes might matter. The plant could increase the volume available for distribution. Work on the pipes could change what reaches people at the other end. Read what the report says about the relationship before attributing the entire change to one investment.
The dates can help you sort possible explanations. If a service improvement appeared before one of the works was complete, ask whether the report explains that sequence. If the only observations come after both changes, the available information may support a combined account without isolating either contribution. A useful summary can identify a plausible contribution and state what the evidence shows without giving it a numerical share the evaluation did not establish.
Look for how the evaluators collected information. An operating record, a visit and an interview each tell you something different. A record might show what the operator measured every day. A visit might show how an installation was being used at a specific time. An interview might help explain why a procedure changed. If a report uses all three, check whether the sources address the same question or provide different parts of an answer.
Also check who appears in the evidence. A result for the whole service area can hide variation between neighbourhoods or groups. If the report gives a breakdown, retain it when it matters to the question you are answering. If there is only a combined figure, report it at that level. In the fictional example, the 3,000 connections could be evidence of expansion if the definition supports it. Without a breakdown, the number alone cannot establish where the expansion was concentrated or which households were served.
Write a summary another reader can verify
Start with a sentence that identifies the document and its date. Then choose one finding and give the comparison that supports it. In the Niroth case, the report’s summary describes cost and schedule overruns and calls efficiency fair. If you also mention the operation indicators, include the exception and the evaluators’ separate interpretation. This lets a second reader open the document and see what you used.
For the fictional example, a defensible sentence would say that the reported daily average was 7,600 against a target of 8,000 during the period being compared. It would leave household coverage out unless the report provided evidence for it. Adding one sentence about what remains uncertain can be enough to keep the conclusion within its scope.
Finally, check whether the person reading your summary could distinguish the old observation from current conditions. Put the evaluation year beside the finding and use past tense for the observation. If you want to discuss what has happened since, look for a later source that actually measures it. That discipline preserves the value of an ex-post report. You can use it to understand what evaluators found after completion while keeping newer questions open.
Sources consulted 7 November 2026. JICA's evaluation method guidance, 2024 evaluation overview and the FY 2016 Niroth external evaluation provide the factual basis. The 8,000 and 7,600 volume figures, the connection count and the maintenance example are original hypothetical examples.