Understand the framework, map the components and build a theory you can test with evidence.
A theory of change explains how and why an activity is expected to lead to a particular change. It identifies the people affected, the outcomes you want, the steps that could lead to those outcomes, and the assumptions behind those connections. It also sets out what evidence would help you judge whether the explanation holds.
For a training program, delivering workshops is an activity. People using a new skill at work is an outcome. The theory explains the steps between them: relevant teaching, practice, feedback and an opportunity to use the skill. It asks what could prevent those steps from working.
A useful theory of change helps a team make choices: what to deliver, what to measure, which partners to involve and when to change course. It can support a community program, a public service or an employer’s development initiative. It is a planning and learning tool; the diagram alone does not prove that the program caused a result.
The Center for Theory of Change describes this as backward mapping: identify a long-term goal, then work through the earlier changes it depends on.
A theory of change is a reasoned explanation, not a promise that the intended result will happen. You may have strong evidence for one connection and very little for another. Making that distinction visible gives the team a useful starting point.
Suppose a training team expects a qualification to improve employment prospects. Its records show strong attendance and completion, but follow-up interviews reveal that employers want work experience as well as a qualification. The team now has a specific assumption to revisit. Adding placement support may make more sense than adding another workshop.
Keep earlier versions of the theory. Record what changed, what evidence prompted the change and who reviewed it. This prevents a team from quietly rewriting its original expectations to fit whatever results arrive.
Sopact describes this continuing practice as a living hypothesis. The underlying discipline is not new: a sound theory should draw on research, experience and evidence, and remain open to revision. BetterEvaluation’s guidance explains how to bring those different sources into the process.
Teams can develop a theory of change in a facilitated workshop, on a shared canvas or in a document. Each format can work. What matters is whether people can challenge the reasoning and whether the team can connect its claims to evidence.
Digital tools make collaboration and revision easier. AI can help organize a program description, suggest questions and identify missing information. Neither replaces the people who understand the setting or the evidence needed to support a claim.
The practical improvement is to connect planning with ongoing learning. Instead of rebuilding the story for each report, maintain a record of the intended outcomes, agreed measures, evidence and unresolved questions. Review it while there is still time to improve delivery.
A six-part pathway is a useful starting structure: problem, inputs, activities, outputs, outcomes and impact. It is not a universal template. Some theories focus on a network of outcomes, distinguish several timescales or show several pathways. In every version, explain the assumptions and context connecting the parts.
Scroll the table sideways to read all columns →
| Component | Question it answers | Illustrative example |
|---|---|---|
| Problem | What needs to change, and for whom? | Adults seeking work lack opportunities to develop and demonstrate relevant skills. |
| Inputs | What resources and relationships are available? | Instructors, funding, equipment and employers willing to offer practice. |
| Activities | What will the program do? | Deliver practical training, coaching and supported work placements. |
| Outputs | What will be delivered or completed? | Training sessions, completed assessments and placement opportunities. |
| Outcomes | What changes for people? | Participants demonstrate skills, secure suitable work and remain employed. |
| Impact | What broader, longer-term change is intended? | More sustained economic security for participants and their households. |
Assumptions run across the pathway. In this example, participants must be able to attend, employers must value the skills and suitable jobs must be available. Those are different questions, requiring different evidence. Attendance records cannot answer all three.
Also name factors the program does not control: transport, caring responsibilities, employer demand and changes in the local economy. For other settings, see the theory of change examples guide.
A theory of change diagram makes the proposed connections visible. Read each arrow as a question: why should this step lead to the next? Put a short explanation beside the connection, then note the evidence needed to check it. A long list of boxes without those explanations can conceal weak reasoning.
Evidence to follow: attendance and assessment → participant and employer feedback → placement and follow-up records. A missing link remains a question to investigate.
This simplified diagram shows one pathway. A real program may need parallel paths, feedback loops or partners’ contributions. Keep the overview readable and place detailed assumptions and measures in an accompanying table. A short narrative should explain the parts that a diagram cannot.
People often use these terms interchangeably. There is no single agreed hierarchy in which a “framework” must be more advanced than a “model.” Check how a funder or partner uses the terms before preparing a document for them.
The deliverable should make the same things clear whatever its name: the intended change, the proposed pathway, why the pathway is plausible, what could interrupt it and how the team will learn whether it is working.
Begin with one defined initiative and the people it is meant to benefit. A small, familiar program may reach a draft quickly. A complex partnership may need several rounds of research and consultation. Agree on a workable first version without treating speed as a substitute for participation.
Describe the current situation, who experiences it and what existing evidence tells you. Include participants, frontline staff and relevant partners. Check whether the proposed problem matches their experience.
Be specific about whose circumstances should improve and in what way. “Improve lives” is too broad to guide a decision. “Sustained access to suitable employment” gives the team something to examine.
Ask what needs to happen first. For each outcome, identify the earlier changes it depends on. Then decide where the program can contribute and where another organization or external condition matters.
Connect activities to the outcomes they are intended to support. Write down why you expect each connection to hold. Distinguish a finding supported by research from an assumption that still needs testing.
Choose indicators, sources, timing and responsibilities. Define terms before collection begins. For employment, for example, specify whether the measure includes temporary work, the follow-up period and whose responses are missing.
Ask people who know the setting to challenge the draft. Agree when to review it and what evidence would trigger a change. Keep a dated record of decisions so later reports can explain what the team learned.
If adults receive practical training and supported work experience, then they are more likely to secure suitable employment, because they can demonstrate skills employers need—provided suitable vacancies and access to work are available.
This is an illustrative statement, not evidence that the approach succeeds in every setting.
For a structured starting document, use the theory of change template. For a guided exercise, continue to the Academy walkthrough.
A logic model usually provides a concise view of resources, activities and intended results. A theory of change puts more emphasis on why the connections should work, the conditions they depend on and alternative explanations. The distinction is useful, but the formats overlap: a well-developed logic model can also include assumptions and context.
Scroll the table sideways to read all columns →
| Question | Logic model typically emphasizes | Theory of change typically emphasizes |
|---|---|---|
| What does it show? | Resources, delivery and intended results. | The explanation linking actions and outcomes. |
| What is it useful for? | A concise overview and delivery plan. | Examining assumptions, context and plausible pathways. |
| How does evidence help? | Track delivery and results against the plan. | Investigate whether the proposed explanation is supported. |
You can use both. A logic model can summarize the program while an accompanying theory explains the reasoning. See the logic model guide and the existing detailed comparison.
A theory of change helps explain what the organization is trying to achieve. A reporting framework determines what a particular audience needs to see. Related frameworks serve different purposes; they are not interchangeable labels for the same report.
For impact-investment reporting, IRIS+ provides metrics and guidance; the five dimensions of impact help examine what changes, for whom, how much, contribution and risk. A metric label alone does not make unlike programs comparable. Check definitions, populations, periods and collection methods before combining results. Consult the GIIN’s IRIS+ resources and Impact Frontiers’ five dimensions when choosing the approach.
Maintain a data dictionary: the agreed meaning of each measure, its source, reporting period, calculation and owner. Reuse compatible evidence across reports, but make any differences explicit. A new funder requirement may need additional data; software cannot supply evidence that was never collected.
Start by separating three questions: did the activity happen, did the intended outcome occur, and how much did the program contribute? Delivery records answer the first. Follow-up evidence helps with the second. The third needs a stronger evaluation design and consideration of other explanations.
Combine numbers with accounts of people’s experience. A completion rate can show where participation drops. Interviews may help explain whether scheduling, support or course relevance played a part. An uploaded report is useful only when its period, source and meaning are clear enough to interpret.
Where consent and access arrangements allow, keep repeat observations linked to the same person, group or organization. Retain dates, source documents and definition changes. Record missing responses as well as positive results, so the people who are easiest to reach do not become the whole story.
AI can help classify feedback against a reviewed rubric, summarize records and flag inconsistencies. Review the source evidence before acting. A numerical score or a confident summary does not establish causation. Contribution analysis offers one approach to examining a program’s role alongside other influences.
Sopact’s Loop methodology connects collection, analysis and improvement. Applied here, its purpose is practical: keep the theory and the evidence close enough that the team can act on what it learns.
The theory can provide continuity from an application through delivery and review. The work changes at each stage, so the questions should change too.
Explain the need, proposed change and reasoning. Distinguish previous evidence from outcomes the new grant is intended to achieve.
Agree on definitions, reporting periods, responsibilities and evidence requirements with the partner. Leave room to explain differences between programs.
Read reported numbers alongside documents and feedback. Investigate missing data, changed definitions and findings that challenge the original assumptions.
Show what happened, what remains uncertain and what the team changed. Link claims to their sources rather than presenting activity totals as proof of impact.
For teams collecting reports from several partners, the portfolio management solution explains the related reporting workflow. The method above remains useful whichever system you use.
Choose the next lesson according to the work in front of you. These are related starting points; the course page provides the full learning path.
It means a reasoned explanation of how a particular change is expected to happen. It connects activities and outcomes and makes the assumptions behind those connections explicit. It remains open to testing and revision.
A common starting structure is problem, inputs, activities, outputs, outcomes and impact, with assumptions and context across the connections. Other formats organize the theory around an outcomes pathway. There is no mandatory number of boxes.
Name the people, proposed activity, intended outcome and reason the connection should work. An “if, then, because” sentence can help. Add the important conditions and check the statement against the people’s experience and available evidence.
It depends on the initiative and available evidence. A team familiar with a small program may draft an initial model in a workshop. A complex partnership may need repeated consultation and research. Agree on a draft and a review process rather than assuming the first version is final.
It helps identify what to monitor and which evaluation questions matter. It connects outcomes to indicators and evidence sources, highlights gaps and gives the team an explanation to test. Monitoring change alone does not prove the program caused it.
AI can help organize source material, draft a pathway and suggest questions. People still need to check the sources, involve those affected, challenge assumptions and decide what is plausible. An AI-generated diagram is a draft, not evidence of effectiveness.
Use the worked examples to compare settings and the template guide to organize your own draft. Adapt the structure to the initiative instead of copying another program’s assumptions.
Customer practice / Open Play Foundation
Open Play’s published story describes bringing facility activity, maintenance, water use and program evidence into one connected record. The team used that information for operational decisions and funder reporting.
“I’m digitizing our entire business through Sopact.”
Marco Botha
CEO and co-founder, Open Play Foundation
The lesson for a theory of change: keep the evidence needed to examine your assumptions available during delivery. This customer story documents the data practice; it does not establish causal impact from the theory illustrated above.
Read Open Play’s story →
