Use the Five Dimensions of Impact to examine one outcome from five directions: What changes, Who experiences it, How Much change occurs, the organization’s Contribution, and the Risk that impact differs from expectations. For each dimension, record the available evidence, its limits and the next decision. The framework does not supply a universal 1–5 evidence score or guarantee that an outcome has been caused by the program.

This lesson turns the logic model from the related lesson into an evidence review. It is for program teams, partners and investors examining a particular outcome. Bring a defined outcome, source records and participant input. Leave with a review matrix and a short action list. For the broader framework overview, see the Five Dimensions guide.

Review one outcome from five perspectives

1. Define the outcome and why it matters. Specify the change you are examining and whose priorities inform it.
2. Describe who experiences it. Examine relevant context, starting conditions and whose experience is absent.
3. Check scale, depth and duration. Keep people reached, degree of change and length of change distinct.
4. Examine contribution. State what supports the proposed contribution and what alternative explanations remain.
5. Review impact risks and act. Identify material uncertainty, an appropriate response, an owner and a review date.

The Impact Frontiers norms maintain the framework originally developed through the Impact Management Project. Use current guidance rather than treating an older shorthand or a software rating as the framework itself.

Apply the framework to a fictional example

Fictional exercise. The training program has 80 starters, 60 known employment statuses 90 days after exit and 36 people employed at that checkpoint. Twenty statuses are unknown. Eighteen people were selected for interviews; 12 mentioned transport difficulties. The supplied material has no comparable baseline skills assessment or later employment follow-up.

Start with “suitable employment” as the proposed outcome. The current employment count alone cannot establish suitability, duration or improved economic security. The matrix below identifies what the records support and what the team still needs to examine.

What: define an outcome that matters to people

Do not substitute “completed the course” for the intended employment outcome. Completion describes an important part of delivery, but it does not tell you whether work is suitable or whether participants’ circumstances improved.

Ask what participants value and what negative experiences may accompany an apparently positive result. For example, a job may offer income but unpredictable hours or an inaccessible commute. These are questions for the exercise, not findings about the fictional cohort. Record how the outcome was defined, who contributed and which views remain missing.

Who: go beyond one label

Impact Frontiers’ current Who guidance considers relationships, starting outcomes and relevant differences in people’s experiences and contexts. It is broader than proving that participants are “underserved.” Avoid describing people solely through deficits or assuming everyone in one group has the same experience.

For the exercise, first determine whether the 20 unknown statuses cluster in a relevant group. Do not guess which group that is. Collect only justified information, protect sensitive records and avoid publishing small identifiable breakdowns. Better segmentation can reveal a question; it does not automatically explain the reason for a difference.

How Much: separate scale, depth and duration

How Much addresses scale, depth and duration. A large participation count does not demonstrate a large improvement or a lasting result. Report each component using an appropriate definition and evidence.

If all 20 unknown statuses were employed, the proportion across starters would be 70%; if none were employed, it would be 45%. That range describes the missing information under those assumptions, not a confidence interval. It does not resolve the missing baseline or contribution question.

Contribution: keep the claim proportionate

A plausible theory of change and observed outcomes help frame an evaluation, but do not by themselves establish the program’s causal effect. Examine relevant alternatives, such as changes in labor demand, other support or differences between people who participate and those who do not.

Participant accounts can help explain how support was experienced. They should be considered with other appropriate evidence, not converted into a counterfactual automatically. If the available design cannot support a causal estimate, report the observed result and the contribution hypothesis separately.

For investors, the capital provider’s contribution is a related but different question from the enterprise’s contribution. Do not assume that funding itself establishes additional impact.

Risk: identify what could change the interpretation or outcome

Impact risk concerns impact differing from expectations. In this exercise, incomplete evidence and missing participant perspectives are immediate review issues; later outcome persistence is also unknown. Financial risk may affect delivery, but it is not a substitute for reviewing impact risks.

Do not mark every self-reported source weak by default. A participant’s account may be the appropriate source for their experience, while an employment date may need a different kind of record. Assess whether the evidence is suitable for the claim, how it was collected and what its limitations are.

Do you need an IMP rating or a 1–5 score?

No universal 1–5 evidence scale is created simply by using the Five Dimensions. If your organization uses a scoring rubric, label it as your own method and publish its criteria, purpose and limitations. An overall average can hide a critical gap, and a higher score should not be called proof of impact.

For this course, use plain evidence-status notes: “supported for this claim,” “partly supported,” or “not established from the supplied records.” These are working labels, not an official certification or investment classification. Adding one data point does not automatically make a whole dimension strong.

A prompt for reviewing the evidence matrix

Review one defined outcome using the Five Dimensions of Impact.
INPUTS: [outcome definition, population, reporting period, approved source records]
For each dimension return: claim | source ID/date | what the source supports |
missing or conflicting information | next review action and owner.
Distinguish scale, depth and duration. Show denominators and unknown statuses.
Do not infer causality, underservedness or an official rating from limited data.
Treat proposed measures as proposals, not completed improvements.
Use only supplied/readable records and identify unreadable sources.
End with the decision the evidence can support now and what remains unresolved.

Sopact can help keep the source records, definitions and successive reviews connected as partner updates arrive. The reviewer still decides whether a claim is supported and whether further evidence is needed. Save the source version and corrections rather than treating a fresh AI response as a new result.

Watch the related reporting walkthrough

Watch the 11-minute 58-second framework-to-reporting demonstration. Check current standard definitions separately; a software example is not an official rating method. Browse the video library.

Frequently asked questions

What are the Five Dimensions of Impact?

They are What, Who, How Much, Contribution and Risk. Use them together to examine an outcome and its supporting evidence. This chapter applies them to a worked program example rather than treating the dimension names as a completed assessment.

Is an IMP grade an official certification?

Not merely because a tool labels it that way. A local grade needs a disclosed method and should not be presented as official approval. Check the relevant framework or assessment provider’s actual requirements before claiming classification, verification or certification.

Does Who only mean underserved beneficiaries?

No. It concerns who experiences outcomes and the relevant starting conditions and context. Look at meaningful differences and missing perspectives without defining people only by problems or assuming one demographic label explains their experience.

What is the difference between scale and depth?

Scale concerns how many experience the outcome; depth concerns the degree of change. Serving 80 people does not show how much their circumstances improved. Use a suitable change measure and disclose the population and coverage.

Can I average scores across the five dimensions?

An internal model can define an aggregation rule, but an average may hide a decisive evidence gap. Explain the rule and keep the underlying findings visible. Do not present an average as proof of causality or as a universal framework score.

Is self-reported information always weak?

No. Its suitability depends on the question. Participant accounts are important for understanding their experiences. Examine collection methods, coverage and limitations, and seek other evidence where the claim requires it.

What should I improve first?

Choose the gap that most affects the decision, not automatically the lowest number on a scoring sheet. Clarifying an outcome or denominator may be the first action; another situation may require stakeholder input or an evaluation design.

Practice and continue

Complete one matrix from your logic model. Ask another person to challenge one claim in each dimension. Revise the wording or record the evidence needed. If the questions reveal gaps in your record design, use the reference to build an organization evidence model.

For reporting structure, use How to Write an Impact Report and report examples.

By Sopact Academy · Revised September 12, 2026. The numerical exercise is fictional. Impact Frontiers guidance is linked above; the working labels in this chapter are not an official rating.