An evaluation plan is the section of a grant proposal that tells a funder how you will know whether the program worked. It answers four questions for each thing you promise: what change you expect (the outcome), how you will measure it (the indicator), how you will collect the data (the method), and how much change by when (the target). The cleanest way to present it is an evaluation matrix — one row per outcome, tied directly to your logic model — plus a short paragraph on who collects data, how often, and how you will report. Funders fund plans that are specific, measurable, and honest about limits, not plans that promise to count everything.
Why funders ask for an evaluation plan
When a funder asks for an evaluation plan, they are really asking one question: if we give you this money, how will we (and you) know it made a difference? A weak proposal promises to "track participation" and "survey clients." A strong proposal shows exactly which changes it expects, how each will be measured, and what success looks like in numbers. The plan is also a risk signal — a clear evaluation design tells the funder you have thought past the launch to the results.
Three respected, free frameworks shape almost every funder's expectations, and reading even one will sharpen your plan:
- The CDC Framework for Program Evaluation lays out a six-step cycle — engage stakeholders, describe the program, focus the design, gather credible evidence, justify conclusions, and use what you learn — and four standards (utility, feasibility, propriety, accuracy). It is the most widely cited public-health evaluation model in the U.S.
- The W.K. Kellogg Foundation Evaluation Handbook is a plain-language guide that walks small organizations through planning and using evaluation, and it pairs naturally with a logic model.
- BetterEvaluation is a global library of methods and options for every step, useful when you need to pick a specific data-collection approach.
You do not need to master all three. You need one matrix that maps cleanly to what you promised in your logic model, written so a reviewer can follow it in two minutes.
The plan must match the rest of the proposal
Reviewers read the need statement, the activities, and the evaluation plan together. If your goals say you will improve reading levels but your evaluation only counts tutoring hours, the proposal contradicts itself. Every outcome you promise should appear as a row in your evaluation matrix — and nothing should appear there that you did not promise.
Outputs vs. outcomes (and why it matters)
The single most common evaluation mistake is reporting outputs as if they were outcomes. Getting this distinction right is what separates a fundable plan from a list of activities.
- Outputs — what you did, counted: 24 workshops delivered, 200 people served, 5,000 meals distributed. Outputs prove effort and dosage. They are easy to count and necessary, but they do not prove anything changed.
- Outcomes — the changes your activities produce in people: knowledge gained, behavior shifted, income raised, a condition improved. Outcomes are usually grouped as short-term (learning), intermediate (action/behavior), and long-term (condition). They prove the work mattered.
- Impact — the long-term, population-level change your outcomes accumulate into over years, usually shared with other organizations. Most single grants are evaluated at the outcome level, not the impact level — be careful not to over-claim.
A useful test: an output you control directly (you decide how many workshops to run); an outcome you can only influence (whether participants actually change). Funders pay for influence on outcomes. Your plan should report a few key outputs to show the program ran as designed, then focus its weight on outcomes. For a deeper treatment of choosing what to measure, see how to measure outcomes.
| Output (effort) | Outcome (change) |
|---|---|
| Number of people who attended job training | % of participants employed 90 days after completing training |
| Hours of tutoring delivered | % of students gaining ≥1 reading level on a pre/post assessment |
| Meals distributed | % of households reporting reduced food insecurity at 6 months |
Indicators, methods, and targets
For every outcome you promise, the evaluation plan needs three things. Get these right for each row and the matrix almost writes itself.
1. The indicator. An indicator is the specific, observable thing you will measure to show the outcome happened. A strong indicator is tied to the change you promised, can be measured consistently the same way each time, and would move only if the program is working. Phrase indicators as a number you can put a value on — a percentage, a count, an average score, or a rate — not as a vague statement like "clients feel better."
2. The method. The method is how you will collect the data for that indicator. Common, affordable methods include pre/post surveys, intake and exit forms, administrative records you already keep, a validated assessment or screening tool, attendance logs, and short follow-up calls. The method should fit your capacity — choose tools you can actually run with the staff and budget you have. BetterEvaluation is a good place to compare options for a specific measure.
3. The target. A target states how much change you expect and by when, against a starting point. Targets need a baseline — where participants start, measured before the program — or the number means nothing. Set targets that are ambitious but defensible; reviewers are skeptical of "100% of participants" and of targets with no basis. If you have run the program before, base the target on past results; if it is new, say so and set a conservative, explainable number.
Write a clear evaluation question first
Before the matrix, state the one or two questions the evaluation will answer — for example, "Do participants who complete the program find and keep employment?" The CDC framework calls this focusing the evaluation design. A sharp question keeps you from measuring everything and reporting nothing.
Decide early whether your evaluation is formative (running during the program to improve it) or summative (judging results at the end). Most grant plans include both: process measures to manage the work, and outcome measures to prove it. Tie each measure back to a box in your logic model so the funder can see the throughline from activity to result.
Worked example: an evaluation matrix
Here is a filled-in evaluation matrix for a fictional Pathways to Work job-readiness program serving 120 adults a year. Each row maps to one outcome from the program's logic model. This is the format most funders are happiest to receive — and the one you can adapt directly.
Program goal: participants gain employment skills, secure jobs, and retain them. Baselines shown are from the prior program year; figures are illustrative.
| Outcome | Indicator | Data method | Baseline → Target |
|---|---|---|---|
| Short-term Participants gain job-search skills | % scoring "proficient" on a resume/interview skills rubric at exit | Pre/post skills rubric scored by staff | 0% → 80% of completers |
| Intermediate Participants find employment | % of completers employed within 90 days of finishing | Follow-up call + employer verification | 41% → 60% of completers |
| Intermediate Participants earn a living wage | Average starting hourly wage of those employed | Wage reported at 90-day follow-up | $15.10 → $17.50 avg. |
| Long-term Participants retain employment | % still employed at 6 months | 6-month follow-up call | Not tracked → 75% of those placed |
| Process Program runs as designed (output) | Number completing ≥80% of sessions | Attendance log | — → 96 of 120 enrolled |
Notice what makes this work: every indicator is a number with a defined method; every outcome target has a baseline beside it (or honestly says "not tracked"); and the last row carries a key output to show the program actually ran. A reviewer can scan it and immediately judge whether the plan is realistic.
Below the matrix, add a short narrative — three or four sentences is enough: "The program coordinator will collect all data using the tools above. Skills rubrics are scored at intake and exit; employment and wage data are gathered by phone at 90 days and 6 months and verified with employers where possible. Data is entered into our case-management system monthly and reviewed by the program director quarterly to adjust delivery. We will report progress in interim reports and full results in the final report." That paragraph answers who, how often, and so what — the parts a table cannot show.
Data collection without a research team
Small organizations do not need a research department to evaluate well — they need a few consistent measures collected the same way every time. The discipline matters more than the sophistication. Keep these principles in mind as you design collection:
- Reuse what you already collect. Intake forms, attendance, case notes, and service records are evaluation data. Add two or three outcome questions to existing forms before you build anything new.
- Collect a baseline at intake. The moment to capture a starting point is when someone enrolls — after the program begins, it is gone. Build the baseline question into the first form a participant sees.
- Pick one strong indicator per outcome. One measure collected well beats five collected sloppily. You can always add measures in year two.
- Use validated tools when they exist. For common outcomes (food security, depression screening, housing stability) free validated instruments already exist; using one strengthens your credibility and saves design time.
- Mind consent and privacy. Tell participants what you collect and why, and store identifiable data securely. As of 2026 — verify — federal grants can carry specific data-protection and human-subjects requirements, so check the notice of funding opportunity.
For larger awards there is a point where outside help pays off. Many federal grants and large foundation grants either require or strongly prefer an independent (third-party) evaluator — someone outside your organization who designs the evaluation, analyzes the data, and reports objectively. Budget for it as a line item; a common range is a single-digit to low-double-digit percentage of the total grant, but confirm against the funder's rules and your scope (as of 2026 — verify). For smaller grants, a well-run internal plan is usually exactly what the funder expects, and over-engineering it can hurt more than help.
When an external evaluator is worth it
Hire one when the funder requires independent evaluation, when the grant is large enough that a credible outside assessment de-risks renewal, when you are testing a model you hope to scale or publish, or when the design needs methods (comparison groups, statistical analysis) beyond your team's capacity. For a small program grant, an internal matrix run consistently is the right call.
Reporting back to the funder
The evaluation plan you write in the proposal becomes the script for the reports you file after the award. The closer your reporting mirrors your plan, the easier your grant is to manage — and the more likely it is to renew. Tie your reporting rhythm to the data-collection rhythm in your matrix.
- Interim reports update the funder mid-grant: progress against each outcome, what data you have so far, what is on or off track, and what you are adjusting. Honesty about a lagging number, paired with a plan to address it, builds far more trust than silence.
- The final report compares results to the targets in your matrix, row by row. Report what you hit, what you missed, and what you learned. A missed target with a thoughtful explanation reads better than a suspiciously perfect scorecard.
- Pair outcomes with a budget-to-actual statement. Funders read your results next to your spending; make sure the program your money describes is the program your data describes.
This is also where evaluation pays off beyond compliance: the same data that satisfies a funder is what you use to improve the program and to win the next grant. Strong, honestly reported outcomes are the backbone of grant management and reporting and of building durable funder relationships. The CDC framework's last two steps — justify conclusions and ensure use — are a reminder that an evaluation no one acts on is wasted effort. Build the plan so its results actually change what you do.
One caution worth stating plainly: do not let a heavy evaluation requirement push you to over-promise. Set targets you can defend, methods you can sustain, and a scope that matches your capacity. A modest plan delivered fully beats an ambitious plan abandoned by month three.
Evaluation plan checklist
Before you submit, run your evaluation plan through this list. If every box is checked, you have a plan a reviewer can fund and a plan you can actually deliver.
Your evaluation plan is ready when
- Every outcome in your logic model appears as a row in the evaluation matrix — and nothing extra does
- Each row has an outcome, a measurable indicator, a data-collection method, and a target
- Every target has a baseline beside it (or honestly states it is not yet tracked)
- Outputs and outcomes are clearly separated, with weight on outcomes
- At least one clear evaluation question frames the whole plan
- The methods fit the staff, budget, and timeline you actually have
- A short narrative names who collects data, how often, and how you will report
- An external evaluator is budgeted if the funder requires or expects one
- Consent, privacy, and any funder data rules are addressed
- The reporting schedule in your plan matches the funder's required reports
Keep the finished matrix as a working document, not just a proposal exhibit. The teams that evaluate best treat the matrix as a dashboard they check quarterly — which is exactly the behavior funders reward with renewal.
Fund the program your evaluation measures — automatically
A strong evaluation plan proves your program works; it still needs unrestricted dollars to run. With Good Circles, supporters pick your cause once, then a share of their everyday local spending funds you automatically — about $72 per active supporter per year, or roughly $36,000 a year from 500 supporters. It is recurring, unrestricted, and free to join. (An estimate, not a guarantee.)
Claim a Founding Nonprofit spot →Sources & tools
Free first
- CDC Framework for Program Evaluation — The most widely cited U.S. evaluation model — a six-step cycle (engage stakeholders through ensure use) and four standards (utility, feasibility, propriety, accuracy) that funders' expectations are built on.
- W.K. Kellogg Foundation Evaluation Handbook — A plain-language guide that walks small nonprofits through planning, conducting, and using an evaluation, designed to pair with a logic model.
- BetterEvaluation — A free global library of evaluation methods and options for every step, useful when you need to choose a specific data-collection approach for an indicator.
Paid — optional labor-savers
- Independent (third-party) evaluator — An outside expert who designs the evaluation, analyzes the data, and reports objectively — separate from the program team delivering the work. Worth it when The funder requires independent evaluation, the grant is large or federal, or you are testing a model to scale or publish. For small program grants, an internal matrix run consistently is usually what funders expect.
Last verified 2026-06-16. Figures and rules change — verify at the source before you act.
FAQ
What is the difference between an output and an outcome in an evaluation plan?
An output is what you did, counted — 24 workshops delivered, 200 people served. An outcome is the change those activities produced in people — knowledge gained, behavior shifted, employment secured. Outputs prove effort; outcomes prove the work mattered. A fundable evaluation plan reports a few key outputs to show the program ran as designed, then focuses its targets on outcomes, which is what funders actually pay for.
What is an evaluation matrix?
An evaluation matrix is a table with one row per outcome and columns for the indicator (the specific measure), the data-collection method, and the target with its baseline. It maps directly to your logic model and lets a reviewer see in two minutes how you will know whether the program worked. It is the clearest, most fundable way to present an evaluation plan, paired with a short narrative on who collects data and how often.
Do I need an external evaluator?
Not for most small grants — a well-run internal plan with consistent measures is usually exactly what the funder expects. An independent (third-party) evaluator becomes worth it when the funder requires independent evaluation, when the grant is large enough that an objective outside assessment de-risks renewal, or when you are testing a model to scale or publish. Many federal and large foundation grants either require or prefer one; budget it as a line item and confirm the funder's rules (as of 2026 — verify).
How do I set realistic targets in a grant evaluation plan?
Start with a baseline — where participants stand before the program — because a target means nothing without a starting point. If you have run the program before, base the target on past results; if it is new, set a conservative number and say so. Make targets ambitious but defensible; reviewers distrust round claims like 100% of participants. State how much change you expect and by when, and tie each target to an outcome in your logic model.