What is a theory of change example?
A theory of change example is a completed model for one program: the activities, what they deliver, the changes expected for people, the long-term aim, and the assumption that links each step to the next. Seeing one filled in for a setting close to yours is the fastest way to spot what your own model is missing.
The four examples below cover workforce training, reading support, a health service and an environmental program. All are illustrative, not documented customer results; the first is worked in full, with indicators.
THE SHORT VERSION
- In every example, find the line between outputs (what the program delivers and counts) and outcomes (what changes for people); most weak models blur it.
- Each example depends on at least one condition the program does not control, such as employer demand or a working collection service; that is the assumption to test first.
- Borrow an example’s structure, not its assumptions: rewrite the people, outcomes and conditions for your own setting.
For the definition and build method, read the complete theory of change guide. The seven-minute introduction below uses a skills-development example; watch where it separates what a program delivers from what changes.
Example 1: What does a theory of change look like for workforce training?
A workforce theory of change runs from training and mentoring, through certification, to placement in a job, retention a year later and, in the long run, graduates in living-wage jobs. This example is fully worked: it is the running example from Sopact’s measurement and reporting course, a fictional regional workforce fund and one of its four job-training partners.
On the onboarding call, the partner said it wants graduates in jobs that pay a living wage, and still there a year later; the two sides turned that into this model and five agreed metrics.
| Step | In this program | Output or outcome? | Agreed indicator |
|---|---|---|---|
| Activities | Training and mentoring | What the program does | Not reported |
| Outputs | People enrolled, completed training, certified | Output: counted by the program | Enrolled (unique people, not visits); completed training; both quarterly |
| Outcome | Placed in a job | Outcome: a change for each person | Within 90 days of exit; quarterly |
| Outcome | Retained | Outcome: a change that lasts | Same job at 12 months; annual |
| Long-term aim | Graduates in living-wage jobs | Impact | Starting wage, hourly, by track; annual |
The outputs row is where many workforce reports stop, and it proves the least. Enrolled and completed training show the program ran. Placement and retention show whether anything changed for the people in it, so they are counted per person, with those the program could not reach named rather than dropped.
Assumptions to test: employers recognize the certification; the job fits the person, so they stay; the job, or the next one, pays a living wage. The one that would most change the plan is that suitable jobs are within reach, because training cannot create vacancies.
Review question: if completion is high but placement within 90 days is low, is the missing link employer demand, the skills people can show, or access to interviews?
The walkthrough below takes a job-training and placement program like this one from its theory of change to the IRIS+ metrics and report built on it. Watch how each outcome box becomes a metric with a definition.
Example 2: How does a theory of change work for reading support in schools?
In a reading-support model, trained tutors deliver structured practice, students take part, and the intended change is stronger reading, not more sessions attended. A school offers small-group support to students who need extra instruction, so the model must link instruction to learning.
| Step | In this program | Output or outcome? |
|---|---|---|
| Activities | Trained tutors deliver structured practice in small groups | What the program does |
| Outputs | Sessions delivered; students attending and practicing between sessions | Output |
| Outcome | Reading accuracy and comprehension improve on a comparable assessment | Outcome |
| Longer-term aim | Students engage more confidently in classroom learning | Impact |
The main assumptions are that instruction fits the students’ needs and that attendance is enough for learning. A teacher’s observation can help explain an assessment change, but it does not replace a comparable assessment, and every later school result should not be credited to the reading support.
Example 3: What is a theory of change example for a health service?
This model aims to keep people connected to a service after a first appointment, so its outcome is continued contact, not a claim about any treatment’s clinical effect. The team agrees follow-up arrangements, sends reminders and helps people past access barriers.
| Step | In this program | Output or outcome? |
|---|---|---|
| Activities | Agree follow-up arrangements, send reminders, help with access barriers | What the program does |
| Outputs | Appointments arranged, reminders sent, support contacts completed | Output |
| Outcome | More people attend the agreed follow-up or arrange a suitable alternative | Outcome |
| Longer-term aim | People stay connected to the care they need | Impact |
The assumptions are that contact details stay usable, the appointment is affordable and reachable, and the service has capacity. Record why someone did not return without reading it as low motivation: transport, timing, changed needs and service availability are different explanations.
Example 4: How do you write a theory of change for an environmental program?
For a household food-waste program, the chain runs from containers and guidance, to correct sorting, to sustained use of collection, to an environmental benefit that also depends on what happens after collection. Providing containers is an output; continued use is a behavior change.
| Connection | Assumption | Evidence to review |
|---|---|---|
| Containers and guidance (output) → correct sorting (outcome) | Households have space and understand what goes in each container | Participation feedback, sorting observations, contamination records |
| Correct sorting → sustained collection (outcome) | Collection is reliable all year | Missed collections, repeated participation measures |
| Collected material → environmental benefit (impact) | The processing route delivers the claimed benefit | Verified processing records and a suitable environmental assessment |
A short participation survey cannot establish the final environmental benefit. Keep the claim in proportion to the evidence and say which parts of the chain are still untested.
What do the four theory of change examples have in common?
All four use the same levels, all four draw a line between what is delivered and what changes, and each one depends on a condition the program does not control. The table puts them side by side so you can see that line and the assumption most likely to break.
Scroll the table sideways to read all columns →
| Example | Output (counted) | Outcome (what changed) | Assumption most likely to break |
|---|---|---|---|
| Workforce training | People enrolled and completed training | Placed within 90 days; retained at 12 months | Suitable jobs are within reach |
| Reading support | Sessions delivered and attended | Reading improves on a comparable assessment | Instruction fits each student’s needs |
| Health service | Reminders sent, appointments arranged | People attend follow-up | The service has capacity and is reachable |
| Environmental practice | Containers and guidance provided | Households sort correctly and keep using collection | Collection stays reliable all year |
They also share one shape. The same five levels appear under different labels in a logic model, a logframe and a results framework, so any example here can be re-laid without new evidence.
For the blank structure, use the theory of change template. The chapter One change model, four formats lays Example 1 out as a logframe and a results framework, row by row.
How do you adapt a theory of change example to your own program?
Keep the structure and rewrite everything else: the people, the outcomes and their time windows, and above all the assumptions, which are the part least likely to carry over. An employer that values a certification in one county may ignore it in the next.
Start with a short conversation with participants, delivery staff and key partners: what is different in your location, and which barriers can your activities not remove? Then write your long-term change for a named group and work backward.
Give each outcome a time window and a definition you could count, as Example 1 does with “within 90 days of exit”. When conditions change, update the reasoning as well as the boxes.
PROMPT · PASTE INTO CLAUDE, CHATGPT OR YOUR AI TOOL
Below is an example theory of change and a description of our program. Adapt the example's structure to our program. Rules: 1. Keep the levels: activities, outputs, outcomes, long-term aim. 2. Take our activities, people and outcomes only from our description. Do not copy the example's content. 3. Label each step "output" (what we deliver and count) or "outcome" (what changes for people). 4. For each arrow, list any assumption from the example that may not hold for us, and say why. Do not state it as true. 5. If our description does not give an outcome's time window or definition, write "not in our data". 6. Do not invent numbers, targets or results. Example: [PASTE ONE EXAMPLE] Our program: [PASTE]
What can a theory of change example not tell you?
An example shows how someone expected change to happen in their setting; it is not evidence that the same chain works in yours, and none of the four here reports real results. Your own records have to test each arrow.
Even with your own data, a rise in placements or reading scores is not proof the program caused it without a comparison. Follow-up relies on people answering, so self-report and non-response shape every outcome, and the people you could not reach should be counted, not dropped. BetterEvaluation’s guidance covers checking the model with stakeholders and existing evidence.
An AI tool can adapt an example quickly, which makes it tempting to inherit assumptions nobody checked. People who know the setting decide what is plausible. In Sopact Sense, each participant keeps one ID from the first form, so the outcome rows are answered from the same person’s records and every line of an AI Assistant answer links to a record you can open.
Start with the example closest to your program
Pick the one example nearest your setting and turn it into your own first model this month.
- Choose the example and copy its table layout.
- Replace the activities and outputs with what you deliver and count.
- Rewrite each outcome for your people, with a time window and a definition.
- Strike every assumption you cannot vouch for, then write your own under each arrow.
- Mark the one assumption most likely to break and the records that would show it.
- Ask a participant and a frontline staff member whether the chain matches their experience.
After the first cycle you have your own model with outputs and outcomes separated, one assumption to test, and an indicator ready for each outcome.
Frequently asked questions
Can I use these examples for my own program?
Yes, as starting points. Keep the structure, then rewrite the people, context, outcomes and assumptions for your setting and check the evidence for each important link. The examples are illustrative, not models validated for your program, and the workforce example is fictional. Borrowed assumptions are where adapted models most often go wrong, so test those first.
What is an example of an outcome in a theory of change?
An outcome is a change for people, with a time window. In the workforce example, “placed in a job within 90 days of exit” and “retained in the same job at 12 months” are outcomes. “Participants enrolled” and “completed training” are outputs: they show the program delivered, not that anything changed. Outcomes are measured per person, and the people you could not reach are reported.
Does every theory of change need six boxes?
No. Problem, inputs, activities, outputs, outcomes and impact are a useful teaching structure, and the examples here use a shorter version. Complex programs may need several pathways, feedback between stages and a narrative. What every version needs is a clear line between outputs and outcomes and an assumption stated for each important link.
What makes a theory of change example testable?
It names the intended changes with time windows, explains why each step should lead to the next, states the assumptions and names the evidence that could support or challenge each link. Example 1 is testable because each outcome has an agreed indicator and a source, such as the placement survey and the 12-month follow-up.
How often should we revise the model?
Review it on a planned cycle, such as once a year, and whenever evidence or conditions change an important assumption. Record the date, what changed, why and who approved it, and keep the earlier model with the reports that used it, so last year’s report can still be read against the pathway it assumed.
