Nonprofit Logic Model Template

A theory of change is the argument underneath the boxes: why your activities should lead to the changes you want. The Center for Theory of Change, which sells training and software for the method, says it requires you to articulate underlying assumptions which can be tested and measured¹. Write it first, so the model follows the argument.

Base it on the program as it runs, not on your last proposal. Ask what happens, to whom and in what order, and where it goes wrong: who drops out, which referrals go nowhere. Then write the claim as a chain of links, rate how well each is supported, and name the weakest.

CDC's 2024 evaluation framework defines a logic model as a graphic depiction of the relation between a program's activities and its intended outcomes². Each arrow should read as an if-then. Draft it as a list first: one element per row, with its column and the element it leads to.

Test the chain both ways. The University of Kansas Community Tool Box suggests asking "But why?" forward from each activity and "But how?" back from each outcome³. An outcome that nothing upstream reaches is a wish. In the example, nothing reaches the two-year outcome, because contact stops at six months.

An indicator is the count or rate that shows whether an element is happening. The Urban Institute's outcome management guide says outcome indicators should almost always begin with words such as "The number of" or "The percent of"⁴. "Improved knowledge" is a direction. "The percent of participants whose score rises on the course's knowledge check" is an indicator.

Give each element one indicator, because an element with three is usually three elements, or one vague one, and a funder can ask for all three. The Center for Theory of Change builds each indicator from four parts: population, target, threshold and timeline⁵. In plain words: who changes, how many, how much counts as success, and by when. For the clinic: 70% of completers check their glucose at least weekly by session six.

Add a target only where evidence supports one. The University of Wisconsin-Extension's logic model guide notes that numeric targets may not be appropriate when the evidence base is weak⁶. "No target, aspiration only" is an honest entry; a blank cell is not.

When you choose indicators, the Urban Institute's guide says the key criteria are the feasibility and cost of measurement⁴. So hold each one against the fields you record today, and give it one of four statuses.

Expect the share of live indicators to fall as you move right. In the worked example, 8 of 14 elements from inputs through outputs are live (57%), but only 1 of 9 outcomes is (11%). That drop is structural, not sloppy: you run the systems that record activities, but outcomes happen to people after they leave.

The CDC framework notes that outcomes can be more challenging to measure than inputs or outputs², and long-term outcomes are often hard to attribute to one program. Showing that your program caused a population-wide change takes a comparison group. That is a study, not a field on an intake form, so plan it in a program evaluation plan.

Next, list what your current grants require. In the US, federal grant rules say the award should state any expected outcomes, indicators, targets, baseline data, or data collections⁷ you must measure and report. Check foundation grant letters and reporting forms too.

Score each required indicator the same way. In the worked example, six funding agreements commit the clinic to 11 distinct indicators, and five aren't live. Two have nothing behind them: a six-month clinical measure and a two-year self-management figure. Close each gap or renegotiate the indicator now, before it shows up as a blank in a grant report.

Then price every gap in staff hours and cash, in separate columns. In the example, six gaps covering 10 elements close for 94 staff hours and no cash, lifting the live share from 39% to 83%. The other four need $129,800, a decision for the board.

The model usually goes into a grant proposal, so use the funder's headings and terms. Labels vary: the Community Tool Box notes that some people call the earliest effects "outcomes" and the later ones "impacts," and others the reverse³. If the template has fewer columns than yours, merge columns rather than drop an outcome.

For any indicator below live, add one plain sentence on what you can measure today and what the rest would take. Label any proxy and say what it misses. Never include an outcome figure you can't trace to participant records.

Scored indicators also show which figures your annual impact report can honestly include. This page's template has a sample clinic's model built and scored, and Edit with AI opens it in River to build yours.