Grant Evaluation Methods: How to Choose the Right One for Your Grant
A chooser for grant evaluation methods: what each design proves and costs, which federal evidence level it can support under 34 CFR 77.1, four questions that settle the choice, and a worked example.

In this article (14)
- Quick Answer: What Are Grant Evaluation Methods?
- What Grant Evaluation Means
- Types of Grant Evaluation
- Grant Evaluation Methods Compared
- Which Evidence Tier Each Design Can Support
- How to Choose a Grant Evaluation Method for Your Grant Type
- Quantitative and Qualitative Grant Evaluation Methods
- Cost and Rigor Tradeoffs
- Worked Example: Choosing a Method for a Two-Year Workforce Grant
- Reporting Requirements That Shape the Method
- Common Grant Evaluation Method Mistakes
- Tracking Evaluation Across the Grant Lifecycle
- FAQ
- Pick the Method Before You Write the Section
Grant evaluation methods decide what your final report can claim. Pick a design the award cannot pay for and it never runs; pick one too weak for the notice and reviewers score the section down. Fit is the first question, and fit means the design works inside the award, the sample and the staff hours you actually have.
This page is a chooser. It compares the methods on what each proves and what each costs, shows which federal evidence tier each design can support, and walks through one choice from start to finish.
Quick Answer: What Are Grant Evaluation Methods?
Grant evaluation methods are the data collection and analysis designs used to measure whether a funded project did what it promised. The common ones are pre-post measurement, comparison group designs, randomized trials, surveys, interviews, focus groups, administrative records, observation, and case studies. Most grants combine two or three, chosen to match the award and the staff you have.
Free download: Evaluation methods comparison workbook as Excel .xlsx. No email address required.
The workbook sits in a library of free grant templates that also holds a project evaluation matrix and an evaluation plan template for writing the method up.
Writing your own? Start it as a draft in GrantCue, with AI help for each section. Start the plan free
What Grant Evaluation Means
Grant evaluation is the systematic collection of evidence about what a funded project delivered and what changed as a result. It is separate from monitoring, which tracks whether activities happened on schedule and whether money was spent as budgeted.
The distinction decides which method you need. Monitoring needs records and counts. Evaluation needs a design that can show change, and sometimes a design that can show the program caused the change. Once the method is chosen, the grant proposal evaluation plan template covers how to write it into the proposal.
Types of Grant Evaluation
Grant evaluations are described along two axes: when the evaluation runs and what it judges. Formative and summative describe timing and purpose. Process and outcome describe focus. Most grants need one from each axis, which is why a process evaluation is usually formative and an outcome evaluation is usually summative.
Formative evaluation
Formative evaluation runs during the project and feeds findings back while there is still time to change the program. Funders ask for it when the work is new or untested, usually in year one of a multi-year award or in a pilot. Its deliverable is a memo of recommended changes.
Summative evaluation
Summative evaluation runs at or near the end of a project and judges whether it achieved its stated outcomes. It is only as good as the baseline captured at the start. A design decided in the final quarter has nothing to compare against, which is how final reports end up describing activities.
Process evaluation
Process evaluation measures whether the program was delivered as designed, to the intended people, at the intended dose. When outcomes come in flat, process data separates a program that does not work from one that was never fully delivered. Without it, both look identical in a report.
Outcome evaluation
Outcome evaluation measures the change in participants the project intended to produce, such as skills gained, health indicators, or employment. Outcomes need a defined time horizon. A workforce program measuring job placement at exit reports something different from one measuring retention six months later, so state the horizon.

Grant Evaluation Methods Compared
The methods differ mainly on one axis: how confidently they let you attribute a change to your program rather than to everything else in a participant's life. Cost and staff time rise along that same axis.
Pre-post measurement
Pre-post measurement compares the same participants before and after the program using the same instrument. It needs no comparison group, which is the usual reason a small nonprofit picks it.
Its weakness is attribution. A gain shows participants changed, not that the program caused the change. That is acceptable for a short, modest award, and it fails the moment a notice asks for evidence of impact. Build baseline collection into intake, or the pre-test never happens.
Comparison group designs
A comparison group design measures a second group that did not receive the program and sets its results against participants' results. Most comparison groups fall apart on one thing: the group has to resemble participants on whatever actually drives the outcome. People who never sought your service usually differ in motivation from people who did, and that difference shows up in the results looking exactly like a program effect.
A waitlist is usually the most defensible source, because everyone in that pool already chose to enroll. A cutoff is another: if the program serves everyone below a set score, the people just above the line are a natural comparison, which is the logic of a regression discontinuity design.
Randomized controlled trials
A randomized controlled trial assigns eligible participants to the program or to a control condition by chance, so the groups are comparable in expectation on every characteristic, measured or not. Small samples can still come out unbalanced, which is why trials publish a baseline comparison alongside their results. Trials produce the strongest causal evidence available and cost the most to run. They suit research awards and funders whose notices ask for causal evidence in those words. They rarely suit a one-year service grant, where the sample is too small to detect anything and randomization means withholding a service from people who need it.
Surveys
A survey collects self-reported data from participants using an identical set of questions. Once it is written, each additional respondent costs almost nothing, which is what makes surveys the default for reach. Response rate then decides whether the results mean anything. A survey answered mostly by people who finished the program describes finishers, not the program. An exit survey handed out before participants leave the room reaches people an emailed one never will. Validated instruments already exist for many common outcomes, and using one saves the cost of writing and testing your own.

Interviews and focus groups
Interviews and focus groups collect open-ended accounts from the people who experienced the program. They explain why a number moved, or why it did not, in a way a fixed-response survey cannot.
Take a composite example: an after-school program at a small youth nonprofit sees flat test scores, and interviews reveal that attendance collapsed after a bus route changed. That turns a failed outcome into a fixable implementation problem. How interviewees were selected belongs in the write-up, since staff-chosen samples skew toward people who liked the program.
Administrative and secondary data
Administrative data evaluation uses records that already exist: attendance logs, service counts, enrollment files, or case notes. Collection already happened as part of the work, so the marginal cost is close to zero.
Two conditions make those records usable: consistent definitions across the grant period, and permission to extract. Both need settling before submission, and a partner's data, such as a school district's, is where that usually stalls.
Observation and fidelity checks
Observation measures whether the program was delivered as designed, scored against a structured checklist. If the proposal names a specific curriculum or model, the funder is buying that model, and fidelity data is the proof they got it. Two observers scoring the same session, at least for a sample of sessions, is what makes the checklist credible.
Case studies
A case study is a detailed account of one participant, site, or cohort followed through the program. A case study illustrates. It cannot generalize, and it should never be the only method in a plan. Consent has to be collected at the same time as the story, because going back for it later rarely works.
Which Evidence Tier Each Design Can Support
Some federal programs rank designs by name. The Department of Education publishes its ranking in regulation. Its regulations at 34 CFR 77.1 define four levels of evidence for its programs, and each level names the study designs that can reach it (eCFR, Title 34 current as of October 21, 2026; read October 2026).
The same section defines a quasi-experimental design study as one that "attempts to approximate an experimental study by identifying a comparison group that is similar to the treatment group in important respects," and says such a study can meet the clearinghouse standards with reservations but not without them. In practice a well-built comparison group study can count as moderate evidence at most, and only an experimental study, such as a randomized trial or a regression discontinuity study, can count as strong. The definitions also accept existing research reviewed by the clearinghouse, so a project can meet a level by citing earlier studies of the same component instead of running its own.
These definitions bind Department of Education competitions, not every funder. They still spell out, more exactly than most notices, what each design can prove, and other notices that ask for "evidence-based" or "causal" findings are asking the same question in looser words. For a program grant that asks for none of this, the bottom row is the realistic bar: a logic model whose main components cite research, plus a pre-post measure of the outcome.
How to Choose a Grant Evaluation Method for Your Grant Type
Choose the method by working backwards from three constraints: what the funder's notice requires, how long the project runs, and who will do the measurement work. Anything that survives all three is a candidate.
Then run four questions in order. Stop at the first one that settles the design.
- Does the notice name a method, a measure or an evidence level? If it does, the funder has decided for you. Use its words in the evaluation section.
- Does the notice ask you to show the program caused the change? Words like impact, causal or effectiveness point to a comparison group or a trial. If the answer is no, a pre-post design with records and one qualitative method is enough.
- Is there a group you can compare against without withholding service? A waitlist, a cutoff score, or sites that start later all qualify. If none exists, a comparison group design is off the table, whatever the notice prefers, and the section should say why.
- Who collects the data, and how many hours do they have? Count the hours per collection point and multiply by the number of points. If the total does not fit inside a funded role, drop to the next simpler design.
Write the chosen method into the logic model so the measures line up with the outcomes you claimed instead of being bolted on afterward.
Quantitative and Qualitative Grant Evaluation Methods
Quantitative methods count things and produce numbers a funder can compare across grantees. Qualitative methods gather descriptions and produce explanations. A mixed methods design uses both, and it is what most program funders ask for when their guidelines specify a design at all.
A workable default for a program grant: one quantitative outcome measure, one administrative measure of delivery, and one qualitative method. That combination fits inside a normal staff workload and covers both halves of what a funder asks, the number and the reason behind it.
Cost and Rigor Tradeoffs
Evidence strength and cost move together, so the decision is how much rigor the award justifies. Spending a large share of a small grant on evaluation is a legitimate reason for a reviewer to question the budget.

Score a candidate method on five things before committing to it: the strength of the evidence it yields, what it costs to set up, how much ongoing staff time it consumes, how large a sample it needs, and whether it requires an external evaluator. A method that scores well on the first and badly on the other four is the wrong method for a small award, however impressive it would look in the narrative.
Most budgets miss two costs. Instrument development is the first, and it applies whenever you cannot reuse an existing tool. Data cleaning is the second, and it routinely takes longer than collection. Both belong in the budget narrative as named lines.
Worked Example: Choosing a Method for a Two-Year Workforce Grant
The applicant is a composite: a workforce nonprofit asking a community foundation for $90,000 over two years to train 80 adults for warehouse and logistics jobs in two cohorts a year. The foundation's guidelines ask for "measurable outcomes" and do not mention causal evidence.
Illustrative figures, not data from a real organization.
The four questions settle most of it. The notice names no method (question 1). It does not ask for causal evidence (question 2). There is a waitlist, because the program turns away about as many applicants as it enrolls (question 3). The program coordinator can give about 4 hours a week to measurement (question 4), roughly 400 hours over two years.
The randomized design fails question 4 and would withhold training from people the foundation wants served. The waitlist comparison fits the hours only just, and only if the waitlist answers follow-up calls. The first design fits with room to spare and answers what the guidelines ask. The proposal names it, budgets the skills test and the coordinator's hours, and adds one sentence: placement for waitlisted applicants will be tracked where they respond, as context, not as a comparison. That sentence costs little and leaves the door open to a stronger design in a renewal.
Reporting Requirements That Shape the Method
A method has to produce data on the funder's reporting schedule, so the schedule is a constraint on the choice. A design whose main result arrives after the final report is due will not help you.
Federal awards set that schedule in regulation. Under 2 CFR 200.329, part of the Uniform Guidance as revised by OMB in April 2024 and effective for awards made on or after October 1, 2024, recipients submit performance reports at intervals set by the award, no less frequent than annually and no more frequent than quarterly unless the agency has applied a specific condition. That section also fixes the due dates: annual reports within 90 calendar days of the reporting period, quarterly or semiannual reports within 30 calendar days, and the final performance report within 120 calendar days of the end of the period of performance. Agencies must collect the information using OMB-approved common information collections, such as the Research Performance Progress Report. Your specific measures sit in the notice of funding opportunity and the award notice, so confirm them per award.
A six-month follow-up on the last cohort is the usual casualty. If the period of performance ends before that follow-up can happen, either measure the last cohort at exit or end enrollment earlier.
Common Grant Evaluation Method Mistakes
The mistakes below are choices of design, not errors in statistics.
- A design nobody is funded to run. A comparison group study appears in the narrative with no evaluator and no data staff in the budget.
- A comparison group chosen for convenience. People who declined the program are compared with people who joined it, and motivation does the work the program gets credit for.
- A trial on a sample too small to show anything. Randomizing a few dozen people produces a result no one can read.
- A survey that only finishers answer. The results describe the people who stayed, not the program.
- No pre-test at intake, so the pre-post design the proposal promised cannot be run.
- Outcome horizons left vague. The proposal says participants will find employment without saying when that gets measured.
Decide the measures while writing the statement of need, because the problem you document there is what your outcomes have to move. Our complete proposal-writing guide shows where the evaluation section sits among the others.
Tracking Evaluation Across the Grant Lifecycle
Evaluation gets designed before the award and executed after it, and the person who wrote the method is rarely the person collecting data eighteen months later. That handoff is where plans tend to fail.
Put the collection schedule somewhere shared the day the award is signed: every measure with an owner, a collection point and a date. In GrantCue each one can be a task with an owner and a due date, next to the award's deadlines and its pipeline status, and opportunities found in grant search across federal and state portals carry into the same pipeline, so the grant you applied for is the same record you report on.
For client-level measures, how to track client outcomes for a grant sets out the data sources, collection points and aggregate reporting that keep an evaluation plan on schedule.
FAQ
These are the questions funders and program staff ask most often when choosing a grant evaluation method.
What is grant evaluation?
Grant evaluation is the systematic collection of evidence showing what a funded project delivered and what changed as a result. It differs from monitoring, which tracks whether activities happened on schedule and whether spending matched budget. A complete evaluation names the outcome, the method, the data source, the owner, and the timing.
What are the different types of grant evaluation?
Evaluations are described along two axes. Timing and purpose give you formative evaluation, which runs during the project and feeds findings back while changes are still possible, and summative evaluation, which judges results at the end. Focus gives you process evaluation, which measures whether the program was delivered as designed, and outcome evaluation, which measures the change in participants. Most grants need one from each axis.
What evaluation method should a small nonprofit use?
For a short grant with limited staff, pre-post measurement of one clear outcome, combined with records you already collect and a few participant interviews, is usually enough. It fits a normal workload and it produces both a number and an explanation. Avoid proposing a comparison group design without funded evaluation capacity.
What study design counts as strong evidence for a federal grant?
At the Department of Education, a single study counts as strong evidence when it is an experimental study, such as a randomized trial, that meets What Works Clearinghouse standards without reservations, shows a significant favorable effect, and includes at least 350 people across more than one site (34 CFR 77.1, read October 2026). A quasi-experimental study can reach moderate evidence at most. Other agencies set their own bars in each notice.
How much of a grant budget should go to evaluation?
There is no universal figure. Some funders cap evaluation spending in the notice of funding opportunity, some require a specific evaluation line, and others expect it absorbed into program costs. Read the budget instructions first, then budget instrument development and data cleaning explicitly.
Do you need a control group for a grant evaluation?
Usually not. Funders expect control groups in research awards and large federal demonstrations where the notice asks for causal evidence. Program grants more often accept a pre-post design supported by administrative and qualitative data. Check the notice: words like impact and causal signal that a comparison or control condition is expected.
Pick the Method Before You Write the Section
The evaluation section gets easier once the method is settled, because everything else follows from that decision. Run the four questions, score the two or three designs that survive on evidence strength, setup cost, staff time, sample size and evaluator requirement, then take the winner into your draft. Use the evaluation plan template and examples to write it up. If you are formalizing this across a portfolio, the grants lifecycle management guide covers where evaluation sits in the cycle.