Picture your last board or funder report — people served, sessions run, programs delivered. All real. Now the question that report doesn't answer: is any of it actually creating the change you exist to create? Not "did we stay busy" — did we move the outcome? For a lot of good organizations, the honest answer is: we're not sure.
That's the gap between your activity and your impact — and the Impact Model closes it. It's a single framework that connects your mission to measurable results and keeps the work on point. Two connected parts: why it works — the reasoning for how your work creates change — and how you measure it — the grid that shows whether it actually is.
Done right, it isn't a document you write once for a grant application. It's the reference you run the work against: state what you expect, measure against it, and adjust. Where the work is landing, you reinforce it. Where it isn't, you change course — with evidence instead of guesswork.
The reasoning that connects your mission to the change — from the result you want back to what you actually do, and the assumptions it rests on, named and made testable.
The grid — inputs, activities, outputs, and outcomes — laid out so you can track whether the work is creating results, and catch it when it isn't.
The grid at a glance — a simple example from a job-training program:
| InputsWhat you invest | ActivitiesWhat you do | OutputsWhat you produce | OutcomesHow things change |
|---|---|---|---|
| Program funding | Recruit & enroll participants | # of participants enrolled | Participants gain job-ready skills |
| Trained instructors | Deliver training sessions | # of sessions held | Participants secure employment |
| Curriculum | Provide job-placement support | # completing the program | Employment sustained at 6 months |
| Employer partnerships | Connect to employers | # of job placements | Household income rises |
The discipline: outputs are what you produce (countable). Outcomes are how things change. Keep them separate and the model stays honest.
Reading the concept is one thing. Building an Impact Model against your own programs and data — the argument, the testable assumptions, and the measurable grid — is the work. That's where we come in.
Start at the result you want and work back to what you do. At each step ask "what has to be true for this to happen?" until you reach things your activities can actually create.
Then make each assumption testable: ask how you'd know if it were false, and what you'd observe. An assumption you can't test is a hope. One you can test becomes a question your measurement answers — and the basis for adjusting when the answer comes back. Where evidence says an assumption doesn't hold, the work gets refined, not defended.
Before you trust the grid, check every item in the Outcomes column. Is it a change, not an activity? ("People served" is activity — move it left.) Is it measurable — could you say what you'd observe if it happened? Is it caused by your work — does something in Activities credibly produce it? If nothing produces an outcome, it's aspiration, not outcome.
AI is a strong drafting partner and a weak decision-maker. Use it for a first pass on either layer, then bring your judgment:
Forcing it to ask questions first keeps it from producing a generic model that looks plausible and means nothing.
AI can draft the structure in seconds. Knowing whether it's honest — an output from an outcome, which assumptions are dangerous, what the work truly requires — is the judgment that comes from doing the work, not from a tool.
If a funder or board has asked you for a "theory of change" or a "logic model," the Impact Model is that — the two unified into one framework and built to be measured.