How to Write a Grant Evaluation Plan (With Examples)
Updated 22 official sources
Quick Answer: A grant evaluation plan is the section that says how you'll measure whether your program did what it promised: process measures (did you deliver it as designed?), outcome measures (did participants change?), baseline data, named instruments, a timeline, and a budget sized to the design.
Full answer and how we checked it
For nonprofit applicants, the weight is set by each notice. Grantsights analysis of 37 nonprofit-eligible federal NOFOs open on Grants.gov on October 20, 2026 found only 3 scored a separate evaluation plan (HUD gave it 10 of 100 points; ACF gave 20 of 125), none required a comparison-group study to apply, and all 37 asked for performance data.
Data note: Amounts may shift with annual appropriations.
Source review by Grantsights, October 20, 2026: Grantsights checked this guide against 37 NOFOs open on Grants.gov that day (21 from DOJ's Office of Justice Programs, 11 from IMLS, 2 from HUD, 2 from AmeriCorps, and 1 from HHS's Administration for Children and Families), ED's FY2025 Education Innovation and Research notice, 2 CFR 200.301 and 200.455, 34 CFR 75.590 and 77.1, 45 CFR 2522.700-740, AmeriCorps' January 2026 evaluation FAQ, 20 CFR 677.155, and SAMHSA's SPARS guidance. Where official sources stop: each agency publishes its own measures and scoring, but none shows how often notices actually score an evaluation plan or require a comparison group. The analysis below counts that. A common mistake follows from it: promising an impact evaluation without a comparison group, baseline data, a data-sharing agreement, or an evaluation budget. For example, a workforce proposal should name its 2nd-quarter employment measure and wage-record source instead of saying outcomes will be tracked after completion.
In this guide
- What 37 Open Federal NOFOs Actually Score
- Why Federal Agencies Care About Evaluation
- Three Types of Evaluation
- Performance Measures vs. Evaluation Questions
- SMART Outcomes: What the Framework Actually Requires
- Baseline Data: The Most Overlooked Requirement
- Agency-Specific Evaluation Requirements
- Internal vs. External Evaluators
- How Much Budget to Allocate to Evaluation
- IRB Requirements
- Common Reviewer Critiques (and How to Fix Them)
- Related Guides
- What Federal Program Officer Feedback Doesn't Tell You
- Sources
- Related Reading
If you're a nonprofit program director or grants lead writing the evaluation section of a federal or foundation proposal, this guide shows you what reviewers score and how to write each part.
Federal funders don't pay for good intentions alone. They want to know you can show whether the program worked. That's what a grant evaluation plan is for: it's your documented case that you have the systems, measures, and methods to tell whether your program produced the outcomes you claimed it would.
Learning how to write a grant evaluation plan that scores well means knowing what the notice in front of you actually asks for, how to structure process and outcome evaluation, how to write SMART outcomes, and where applicants lose the points this section is worth. For the full proposal structure and how the evaluation plan fits alongside the needs statement, project narrative, and budget, see our guide to writing a grant proposal.
What 37 Open Federal NOFOs Actually Score
Only 3 of 37 open notices gave the evaluation plan its own scored criterion. Grantsights analysis of 37 nonprofit-eligible federal NOFOs found that the other 34 fold evaluation into performance reporting, evidence, or project design instead of scoring a stand-alone evaluation section.
Method: on October 20, 2026, Grantsights pulled all 869 posted opportunities from the public Grants.gov search API (opens in a new tab), kept every competitive notice from DOJ's Office of Justice Programs, ACF, HUD, IMLS, and AmeriCorps that lists nonprofits as eligible and has a downloadable NOFO (37 notices), and read each one's review criteria and evaluation language. Formula, invitation-only, and government-only notices were left out.
| Agency (notices) | Separate scored evaluation criterion | What the notice asks for | Third-party or comparison-group evaluation |
|---|---|---|---|
| DOJ Office of Justice Programs (21) | 0 of 21 | Performance reports and data after award; no evaluation criterion | Priority consideration in 2 (both Second Chance Act); required in 0 |
| IMLS FY2027 museum and library programs (11) | 0 of 11 (no point values published) | A required Performance Measurement Plan, 2 pages recommended | 0 |
| HUD Fair Housing Initiatives Program (2) | 2 of 2: 10 of 100 base points | Measurement (5 points) and an evaluation plan (5 points) | 0 |
| ACF Office of Refugee Resettlement (1) | 1 of 1: 20 of 125 points | A project performance evaluation plan, plus a 5-point logic model | 0 |
| AmeriCorps State and National and RSVP (2) | 0 of 2 (evaluation plan summary scored at 0 percent) | Evidence (20 points) and data collection (8 points) in State and National; performance measures (30 percent) in RSVP | Independent impact evaluation after award for State and National grantees averaging $500,000+ |
| Total (37) | 3 of 37 (8%) | All 37 ask for performance data | 0 of 37 required to apply |
Three things stand out. First, performance measurement is universal and formal evaluation is not. Second, where a notice does score evaluation, it's worth real points: HUD's 10 points equal its whole Rating Factor 4, and ACF's 20 points are 16% of the total. Third, none of the 37 made a comparison group or outside evaluator a condition of applying. That's why proposing one you can't deliver is a red flag rather than a bonus: the rigor wasn't required, and a reviewer can see when the data access, sample, and budget behind it don't exist.
Here's what the scored criteria look like in real notices:
- HUD Fair Housing Initiatives Program, Education and Outreach Initiative (opens in a new tab) (OFH-2600-DC-021A): Rating Factor 4, "Achieving Results and Program Evaluation," is worth 10 points. It asks for strategies to gather valid, quantifiable output and outcome data, how you'll validate that data, and a plan to evaluate progress toward goals and long-term outcomes in your service area. HUD's Private Enforcement Initiative notice uses the same 5-plus-5 split.
- ACF Home Study and Post-Release Services (opens in a new tab) (HHS-2026-ACF-ORR-ZU-0019): the 20-point Impact criterion is four 5-point items: a performance evaluation framework, continuous quality improvement, data collection and management, and obstacles plus a review timeline. A logic model earns up to 5 more points under Purpose and need.
- DOJ Second Chance Act Community-based Reentry Program (opens in a new tab) (O-BJA-2026-172717): no evaluation criterion, but the notice gives priority consideration to applicants that provide for an independent evaluation using random assignment "to the maximum extent feasible," and asks them to name an evaluation partner not otherwise tied to the project. It also asks every applicant to identify the baseline recidivism figure for its target population. OJJDP's Second Chance Act Youth Reentry (opens in a new tab) notice carries the same statutory priority.
- IMLS Museums for America (opens in a new tab) (FY2027): the Performance Measurement Plan covers 4 measures (effectiveness, efficiency, quality, and timeliness). For each one you name the data, source, method, and schedule. Reviewers judge it under the Project Work Plan criterion, which asks whether your tracking methods will give "reliable and measurable information" about results.
The pattern outside this sample can look very different. ED's FY2025 Education Innovation and Research mid-phase notice (opens in a new tab) gave up to 30 of 100 points to the quality of the project evaluation, 15 of them for a design that would meet What Works Clearinghouse standards without reservations, and required every grantee to run an independent evaluation and submit the final report to ERIC. Read the review criteria before you decide how ambitious your design should be. If you're still choosing which program to apply to, search open federal grants and open the full NOFO before you draft.
Looking for your next grant?
About once a week, we email a short list of open grant calls with real deadlines from different funders. It lists open calls only.
No spam. Unsubscribe anytime.
Why Federal Agencies Care About Evaluation
Two laws drive most of what agencies ask for: the GPRA Modernization Act of 2010 and the Foundations for Evidence-Based Policymaking Act of 2018.
Under the GPRA Modernization Act (GPRAMA), agencies set performance goals and report progress to Congress. The data your project reports feeds those reports, which is why so many notices hand you a fixed list of measures instead of letting you invent your own. The Evidence Act added a second layer: under 5 U.S.C. 312 (opens in a new tab), each agency must publish a plan of the policy questions it wants answered and an evaluation plan issued with its annual performance plan.
The Uniform Guidance turns this into grant terms. According to 2 CFR 200.301 (opens in a new tab), the agency must measure each recipient's performance and should spell out in the award any expected outcomes, indicators, targets, baseline data, or data collections you're responsible for, plus any requirement to take part in federally funded evaluations.
The practical upshot: federal funders value outcomes measured the same way across all their grantees, not just a good story for your project. When a notice lists specific measures, use them exactly. It's how the agency adds your results to everyone else's.
Three Types of Evaluation
There are 3 types, and you don't need all of them for every grant. You do need to know which ones your notice asks for and describe them specifically.
Process Evaluation
Process evaluation answers: Did you do what you said you'd do?
It tracks program implementation: whether you delivered the activities you described, served the number of participants you projected, and operated the program as designed. Process data includes:
- Number of participants enrolled
- Number of sessions or activities delivered
- Attendance rates
- Staff qualifications and training completion
- Adherence to the program model or curriculum
Process evaluation is how you tell whether weak outcomes came from the program model or from poor implementation. If you didn't reach your target enrollment, you can't draw conclusions from your outcome data. Documenting fidelity is what lets you make that distinction.
Most community-based federal grants collect process measures through quarterly or semiannual performance reports. All 21 DOJ notices in the sample say OJP will judge success by reviewing your performance reports and data.
Outcome Evaluation
Outcome evaluation answers: Did your program produce the intended changes?
It measures whether participants changed in knowledge, skills, attitudes, behaviors, or circumstances. Examples:
- Pre-test and post-test knowledge scores
- Employment in the 2nd quarter after program exit
- Symptom reduction on validated clinical scales (PHQ-9, AUDIT-C)
- Housing stability at 6-month follow-up
- Re-arrest rates at 12 months post-release
Outcome evaluation requires measuring participants before the intervention (baseline) and at defined points after it (follow-up). Without a baseline, you can't show change.
Impact Evaluation
Impact evaluation answers: Did your program cause those changes?
It requires a comparison or control group, people similar to your participants who didn't get the program, so you can separate your program's effect from everything else going on. According to AmeriCorps' rule at 45 CFR 2522.700 (opens in a new tab), that comparison is exactly what separates evaluation from performance measurement: a performance measure shows something good is happening, but it "does not indicate that the change can be wholly attributed" to the program. Randomized controlled trials are the strongest version.
A real-world example: MDRC's 2015 evaluation of Reading Partners (opens in a new tab), a nonprofit volunteer tutoring program, randomly assigned more than 1,100 students in 19 schools across 3 states to tutoring or a control group. It found statistically significant gains on 3 reading measures, equal to about 1.5 to 2 months of extra growth. That's what an impact study looks like, and it needed school cooperation, a large sample, and a professional evaluator.
When is impact evaluation required? Only when the notice or program rule says so. Examples include ED competitions like Education Innovation and Research, which require an independent evaluation designed to meet What Works Clearinghouse standards, and AmeriCorps State and National grants averaging $500,000 or more a year. In the 37-notice sample, no notice required one to apply, and the 2 DOJ reentry notices that mention random assignment only give priority consideration for it. If you can't execute a comparison design, don't promise one. A rigorous design that collapses after award does more damage than an honest pre-post design.
| Evaluation Component | What Reviewers Look For | Common Mistake |
|---|---|---|
| Logic Model | Clear causal chain, measurable outcomes | Too many boxes, vague connections |
| Process Evaluation | Implementation fidelity measures | Ignoring process metrics |
| Outcome Evaluation | Specific, time-bound targets | Promising outcomes beyond grant period |
| Data Collection | Named instruments, collection timeline | Saying "TBD" for instruments |
| Analysis Plan | Statistical methods appropriate to data | Overclaiming causal inference |
| Evaluator Qualifications | Independence, topic expertise | Using internal staff only when the notice asks for independence |
Performance Measures vs. Evaluation Questions
Plan on 3-5 performance measures and 2-3 evaluation questions. These are two different tools, and many federal applications want both.
Performance measures are specific, quantifiable indicators you'll track throughout the project. They answer "how much" and "how many" at defined intervals. Examples:
- Number of individuals enrolled (tracked monthly)
- Percentage of participants completing the full program (tracked at program exit)
- Percentage achieving the primary outcome (tracked at 90-day follow-up)
Evaluation questions are broader questions the evaluation is designed to answer by the end of the project period. Examples:
- To what extent did program participation improve employment outcomes among the target population?
- What factors were associated with higher rates of treatment retention?
- Were outcomes consistent across subgroups of participants?
Performance measures are operational: you track them on a dashboard. Evaluation questions require synthesis and analysis at the end of the project period or at a defined midpoint.
When a notice gives you required measures, they come first. IMLS, for instance, asks applicants in 10 of its 11 open FY2027 programs to plan around 4 fixed measures (Native Hawaiian Library Services uses 3), and AmeriCorps RSVP puts 30 percent of the score on work plans built from its national performance measures. Add your own measures only where the required ones don't capture your outcomes.
SMART Outcomes: What the Framework Actually Requires
SMART has 5 elements, and each one does a specific job in a federal outcome statement.
Specific: Name exactly what you're measuring and for whom. "Improve health outcomes" isn't specific. "Reduce HbA1c levels among program participants with Type 2 diabetes" is specific.
Measurable: You must have a data collection method that produces a number. If you can't measure it with an instrument, assessment, administrative record, or survey, it's not a measurable outcome. Name the measurement tool in the outcome statement or right below it.
Achievable: Ground your target in results from comparable programs. Don't claim 95% employment if similar programs reach 55%. Reviewers know the research in their field, and unrealistic targets signal that you either don't know it or are overpromising to score points.
Relevant: Each outcome must connect directly to the need you documented. If your needs statement establishes that 35% of youth in your county read below grade level, your outcomes must include reading proficiency. A gap between documented need and proposed outcomes is an easy critique for a reviewer to write.
Time-bound: State exactly when the outcome will be measured. "By end of program year 1," "within 90 days of program exit," "at 6-month follow-up." Without a timeframe, the outcome can't be evaluated.
Example of a weak outcome statement: "Participants will have improved employment outcomes."
Example of a SMART outcome statement: "60% of participants who complete the 12-week workforce training program will gain employment within 90 days of program exit, as documented by employer verification or pay stub."
The SMART version names the proportion (60%), the population (participants completing the 12-week program), the outcome (employment), the timeframe (90 days), and the data source (employer verification or pay stub). Each outcome should also appear in your logic model as an outcome, not an output.
Baseline Data: The Most Overlooked Requirement
A baseline is the 1 number that makes every later number mean something: where participants stand before your program starts.
Without it, you can't show change. You might have post-program data showing 70% of participants employed, but if you don't know their employment rate before the program, you can't claim credit for that result. Some notices make this explicit. DOJ's FY2026 Second Chance Act Community-based Reentry notice asks every applicant to identify the baseline recidivism figure for its target population inside the 35% Project Design and Implementation criterion.
How to collect baseline data:
Pre-test / post-test: For knowledge, attitude, or skill outcomes, give an assessment at program entry and again at completion or follow-up.
Administrative records at intake: For employment, housing, criminal justice, or health status outcomes, document baseline status at enrollment (employment verification, housing documentation, medical records).
Validated clinical instruments at intake: For mental health outcomes, give a validated scale (PHQ-9 for depression, GAD-7 for anxiety, AUDIT-C for alcohol use) at intake and at follow-up intervals.
A missing baseline plan is one of the easiest gaps for a reviewer to spot. Even if you don't have existing baseline data for your population, your plan must say exactly how you'll collect it within the first 30-90 days of each participant's enrollment.
State it plainly: "Baseline data will be collected at intake using the PHQ-9 depression scale and a structured employment status assessment. Follow-up data will be collected at 90 days and 6 months post-enrollment."
Agency-Specific Evaluation Requirements
At least 5 agencies publish evaluation or performance rules you must build around. Here's what each one requires as of October 2026.
SAMHSA Client-Level Measures and SPARS
SAMHSA service grantees report client-level data through SPARS (opens in a new tab), SAMHSA's Performance Accountability and Reporting System. For years that meant the CSAT GPRA tool or the CMHS National Outcome Measures (NOMs) tool, which ask about areas such as substance use, employment or education, housing, criminal justice involvement, and social connectedness. That's changing: according to the SPARS home page, GPRA and NOMs data entry closes on November 5, 2026 for grants that have moved to the new SUPRT-A (administrative) and SUPRT-C (client) tools. Check your notice for which tool applies.
Follow-up is where these plans break. According to SAMHSA's SPARS CSAT GPRA follow-up fact sheet, the minimum follow-up rate is 80 percent, one 6-month follow-up interview is required for every client, and the window for most CSAT programs runs from 1 month before to 2 months after the 6-month anniversary of intake. Clients can receive up to $30 for completing a follow-up interview. Your evaluation plan should say who collects intake data, how you'll locate clients at 6 months, and how you'll hit 80%.
ED Evidence Levels and What Works Clearinghouse Standards
ED's evidence tiers come from the Every Student Succeeds Act and are defined in 34 CFR 77.1 (opens in a new tab). The What Works Clearinghouse doesn't create the tiers; its standards (now in Handbook version 5.0) are what ED uses to judge whether a study counts. A notice only asks you to claim a tier when the competition uses an evidence priority or requirement.
- Strong evidence: at least 1 experimental study that meets WWC standards without reservations, shows a statistically significant positive effect, and includes at least 350 students or other individuals across more than one site
- Moderate evidence: at least 1 experimental or quasi-experimental study that meets WWC standards with or without reservations, with the same 350-person, multi-site sample rule
- Promising evidence: at least 1 experimental, quasi-experimental, or well-designed correlational study with statistical controls for selection bias, showing a significant positive effect
- Demonstrates a rationale: a key project component in your logic model is supported by citations of high-quality research suggesting it's likely to improve relevant outcomes
These tiers describe the evidence behind the intervention you propose, which is a separate question from the evaluation you'll run. ED scores the evaluation itself under the "quality of the project evaluation or other evidence-building" criterion in 34 CFR 75.210(h) (opens in a new tab) when a competition uses it. Under 34 CFR 75.590 (opens in a new tab), if the notice required you to describe an evaluation, you must later show ED that the evaluation met the standards in your approved application, and a notice may also require an independent evaluation and a public final report.
DOL Employment and Training Administration Performance Measures
Under WIOA, 20 CFR 677.155 (opens in a new tab) sets six primary indicators of performance for the adult, dislocated worker, adult education, and vocational rehabilitation programs (youth programs use education-or-employment versions of the first two). DOL-funded workforce grants generally track these same measures:
- Employment rate, 2nd quarter after exit: Percentage of participants in unsubsidized employment during the second quarter after exit
- Employment rate, 4th quarter after exit: Percentage of participants in unsubsidized employment during the fourth quarter after exit (measured across all exiters, not just those employed earlier)
- Median earnings: Median earnings in the 2nd quarter after exit
- Credential attainment rate: Percentage of participants who obtain a recognized postsecondary credential or secondary diploma during participation or within 1 year after exit
- Measurable skill gains: Percentage of participants in education or training who achieve documented progress toward a credential or employment during a program year
- Effectiveness in serving employers: A program-level measure of how well the program serves employers
The older "entered employment" and "employment retention" rates were pre-WIOA Common Measures, so don't use those labels in a current proposal.
Build these measures into your design. Describe how you'll collect post-exit employment data, typically through state unemployment insurance wage records, participant self-report, or employer verification. Wage records are the most credible source and are what DOL expects.
AmeriCorps Evaluation Requirements by Grant Size
AmeriCorps is the clearest case of evaluation rules set by grant size. According to 45 CFR 2522.710 (opens in a new tab), a State competitive or national direct grantee whose average annual AmeriCorps grant is $500,000 or more must arrange an independent evaluation, and a grantee under $500,000 (or an Education Award Program grantee) must conduct an internal one. The evaluation must cover at least 1 year (45 CFR 2522.720). These sections haven't changed in substance since 2005; the 2024 amendments only replaced "the Corporation" with "AmeriCorps."
AmeriCorps' January 2026 evaluation FAQ adds the details. Large grantees must run an independent impact evaluation using an experimental or quasi-experimental design. Small grantees may choose a process, outcome, or impact design and use an internal evaluator such as staff, board members, or partners. Grantees that already completed an impact study, or that replicate a model already tested elsewhere, can request an Alternative Evaluation Approach.
The FY2027 State and National notice (opens in a new tab) (applications due January 13, 2027) shows how this plays out in scoring:
| Requirement in the FY2027 notice | Who it applies to | Weight |
|---|---|---|
| Evidence Tier | All applicants | 12 points |
| Evidence Quality | All applicants | 8 points |
| Data Collection (budget and process for performance data) | All applicants | 8 points |
| Evaluation Plan Summary Form | Applicants with 3+ years of competitive funding for the same project | 0 percent (not scored) |
| Completed evaluation report | Applicants with 6+ years of competitive funding for the same project | Not scored; a report missing its title-page details may not get credit for meeting evaluation requirements |
The notice also reports that among 2026 applicants recommended for funding, 51% were in the strong evidence tier, 10% moderate, 16% preliminary, and 23% pre-preliminary. For a first-time applicant, the evaluation plan isn't scored at all; the 28 points tied to evidence and data are. For more on the program itself, see our AmeriCorps State and National guide.
Internal vs. External Evaluators
Of the 37 open notices Grantsights read, 0 required an outside evaluator to apply. You have two options for who runs your evaluation: internal staff or an external independent evaluator.
Internal evaluation is done by your own staff. It's less expensive and benefits from your team's program knowledge. The limit is objectivity: reviewers know organizations have reasons to report good news.
External evaluation is done by an individual or firm with no stake in the outcome. AmeriCorps' FAQ defines independent as an evaluator with "no formal or personal relationship with, or stake in" the grantee's administration, management, or finances. It's more credible and more expensive.
When an external or independent evaluator is required:
- When the notice says so (read the program requirements section, not just the review criteria)
- For AmeriCorps State and National grantees averaging $500,000 or more a year
- For ED competitions that require an independent evaluation under 34 CFR 75.590, such as Education Innovation and Research
- When you're seeking a priority that depends on one, such as the Second Chance Act independent-evaluation priority in DOJ's reentry notices
Even when it isn't required, an external evaluator can add credibility on a large award. But a named evaluator who hasn't agreed to the work is worse than none.
If you're committing to one, either name the firm or person (with qualifications) or describe your selection process: "We will issue an RFP within 60 days of award and select an external evaluator through a competitive process based on relevant experience, methodological expertise, and cost." DOJ's reentry notice asks applicants seeking the priority to describe an evaluation partner "not otherwise associated with the project," such as a university researcher or private research firm.
How Much Budget to Allocate to Evaluation
A common rule of thumb is 10-20% of total project budget, but no federal rule sets a percentage. Size the budget to the design the notice asks for: tracking 4 IMLS performance measures with existing staff costs far less than the independent impact study an AmeriCorps grantee averaging $500,000 a year must fund.
The cost rules are clearer than they used to be. According to 2 CFR 200.455(c) (opens in a new tab), added in the 2024 Uniform Guidance revision, "costs related to data and evaluation are allowable." Evaluation costs include evidence reviews, evaluation planning and feasibility assessment, conducting evaluations, and sharing results. Data costs include data systems, personnel, dashboards, and cybersecurity.
For internal evaluation: Staff time for data collection, entry, and analysis is your main cost. Calculate it honestly. If case managers collect data at intake and follow-up, that time has a cost, so build it into personnel at the right effort percentage.
For external evaluators: Get a written quote before you finalize the budget. Costs depend on design, sample size, data access, and reporting, and an experimental study costs far more than a performance-measure report. AmeriCorps' FY2027 notice scores the budget for data collection and evaluation directly, under its 8-point Data Collection item.
Line items to include in the evaluation budget:
- Evaluator personnel time (internal) or contract cost (external)
- Licensing fees for validated instruments (some require them)
- Data management software or database development
- Incentives for participant follow-up (if allowed; SAMHSA's CSAT guidance permits up to $30 per follow-up interview)
- Travel for external evaluators to site visits
For how to justify each line, see our grant budget justification guide.
IRB Requirements
Under 45 CFR 46, activities designed to produce generalizable knowledge count as research, and research with human subjects usually needs Institutional Review Board (IRB) approval.
Purely programmatic data collection for internal improvement, such as tracking your own clients' outcomes to improve services, usually doesn't. Data collection that's designed to produce generalizable knowledge does. A randomized study meant for publication will almost always need IRB review.
If you're partnering with a university evaluator, they almost certainly have IRB capacity. If you're a standalone nonprofit, independent IRBs serve organizations without their own.
Address this in your evaluation section if it's relevant: "This project's evaluation constitutes research involving human subjects. We will seek IRB approval from [institution] within 60 days of award. The principal investigator will maintain IRB certification and ensure all data collection protocols comply with 45 CFR 46."
If your evaluation doesn't meet the federal definition of research, say so briefly: "This project's evaluation involves programmatic data collection for quality improvement purposes and does not meet the federal definition of research requiring IRB oversight."
Common Reviewer Critiques (and How to Fix Them)
These 6 critiques map directly onto the scored items in the notices above, and all of them are fixable before submission.
"Outcomes are not measurable." Every outcome statement needs a specific number and a data collection method. Rewrite any outcome that doesn't state a percentage, threshold, or measurable change, and pair it with its instrument or data source. HUD's FHIP notices score exactly this under their 5-point Measurement item.
"No baseline data plan." Describe how you'll collect pre-program data within the first 30-90 days of enrollment. Name the instrument. State who administers it and when.
"Evaluation is underfunded." If the evaluation line can't pay for the design you describe, either add money or scale back the design. Reviewers compare the two.
"No description of data analysis methods." State how you'll analyze the data: pre-post comparison with paired t-tests, descriptive statistics by subgroup, logistic regression for factors linked to outcomes. You don't need a statistics textbook, just enough to show you've thought about it.
"External evaluator not identified." Either name a firm or describe a competitive selection process with a timeline. "We will identify an external evaluator" without a plan doesn't satisfy reviewers.
"Outcomes don't connect to the documented need." Go back to your needs statement. If you claimed a specific problem exists, your outcomes must measure whether your program reduced it.
Related Guides
The Evaluation Budget Line Reviewers Check Early
A skeptical reviewer will often flip to the budget and find the evaluation line first, because an underfunded evaluation is a quick signal of how seriously the applicant took this section.
The 2024 revision to 2 CFR Part 200, effective October 1, 2024, removed one excuse for a thin line. Section 200.455(c) now says plainly that data and evaluation costs are allowable, including evidence reviews, evaluation planning, feasibility assessments, conducting evaluations, and sharing results. Before October 2024, § 200.455 covered only organization costs such as incorporation fees. You can now show the evaluation line openly and explain it. Separately, the same revision raised the Single Audit threshold (opens in a new tab) to $1,000,000 in federal expenditures a year, which matters for small grantees adding a new federal award but doesn't change how evaluation costs are charged.
What many applicants still miss is building the external evaluator's start-up time into the grant period. Some proposals promise a university partner will run the evaluation but don't mention that the partner hasn't agreed yet and won't be selected until Month 3. Data collection can't start until the evaluator is named, the data sharing agreement is signed, and any IRB protocol is approved. If your evaluation touches protected health information or identifiable client data, plan for weeks to months of IRB lead time. For a grant that starts October 1, that can push the first follow-up data into the new calendar year, so write the evaluator selection timeline into the evaluation plan itself, not just the project timeline. ACF's ORR notice scores this directly: 5 of its 20 Impact points go to naming obstacles and a review timeline.
SAMHSA's SPARS system is where client data submission happens, and first-time grantees often underestimate how strict it is. A 6-month follow-up has to match a complete intake record. Organizations that don't train intake staff on the exact fields, or don't build their case management system around them, end up with follow-up data they can't use and fall short of the 80% follow-up minimum. The problem usually isn't a weak evaluation on paper. It's a data system that wasn't designed around SPARS, or now SUPRT, from day one.
Don't promise a pre-post design without describing attrition. If you're enrolling 120 participants and your outcome needs a 90-day follow-up, reviewers will ask what share you expect to lose. Programs serving people with unstable housing or active substance use often lose a meaningful share between intake and follow-up. If you assume, say, 30% attrition, you'll have complete data on about 84 people. If that isn't enough for your stated analysis method, your plan has a validity problem a reviewer will catch even if you didn't.
What Federal Program Officer Feedback Doesn't Tell You
Program officers will tell you what to report, but 3 things rarely make it into their feedback or the notice:
Most notices don't reward rigor you can't deliver. In the 37 open notices Grantsights read, 0 required a comparison group to apply, and only 2 gave priority consideration for one. Write your plan around the evaluation capacity you can actually secure, not the ideal design. A well-run simple evaluation beats a poorly run mixed-methods study.
Process metrics matter more in Year 1 than outcome metrics. Program officers reviewing your first annual report want to know you implemented the program as proposed. Outcome data comes later; fidelity data comes first.
Don't promise a comparison group unless you've already secured one. A "quasi-experimental design with matched comparison" fails because the comparison site never agreed to share data. If you can't guarantee access, propose a pre-post design with statistical controls instead, and say why.
Sources
- Grants.gov search API (opens in a new tab) (37 NOFOs open October 20, 2026)
- HUD FHIP Education and Outreach Initiative NOFO, OFH-2600-DC-021A (opens in a new tab)
- ACF Home Study and Post-Release Services NOFO, HHS-2026-ACF-ORR-ZU-0019 (opens in a new tab)
- DOJ BJA FY2026 Second Chance Act Community-based Reentry Program NOFO (opens in a new tab)
- ED FY2025 Education Innovation and Research Mid-Phase notice (opens in a new tab)
- AmeriCorps FY2027 State and National Grants notice and evaluation resources (opens in a new tab)
- 45 CFR 2522.710, AmeriCorps evaluation requirements (opens in a new tab)
- 34 CFR 77.1, ED evidence definitions (opens in a new tab)
- 2 CFR 200.455, data and evaluation costs (opens in a new tab)
- What Works Clearinghouse Handbooks (opens in a new tab)
- SAMHSA SPARS (opens in a new tab)
- MDRC, Mobilizing Volunteer Tutors to Improve Student Literacy (2015) (opens in a new tab)
- CDC Framework for Program Evaluation in Public Health (opens in a new tab)
- NIH Grants Application Guide (opens in a new tab)
- NSF Proposal and Award Policies and Procedures Guide (PAPPG) (opens in a new tab)
Related Reading
For the full proposal structure that surrounds your evaluation plan, including agency-specific scoring criteria and SF-424 requirements, see how to write a federal grant proposal.
The evaluation plan can't be built in isolation from your logic model: every outcome you commit to measuring must appear in your logic model first. See what is a logic model for grants for the 5-component framework and the difference between outputs and outcomes.
For the reporting that follows the award, see grant reporting requirements by agency. For the federal cost rules behind your evaluation budget, see the 2 CFR 200 Uniform Guidance guide.
The CDC's Framework for Program Evaluation in Public Health is publicly available at cdc.gov/evaluation (opens in a new tab) (CDC published an updated version in 2024) and is widely cited for health-related federal grants. It's worth referencing in your evaluation narrative if your grant is health-related.
Federal evaluation requirements are demanding but learnable. Document where participants start, describe how you'll measure change, fund it honestly, match the rigor to what the notice actually scores, and make sure every outcome connects back to the need you documented at the front of the proposal.
Frequently Asked Questions
Do I need an external evaluator for my federal grant?
Only when the notice or the program rule says so. AmeriCorps State and National grantees averaging $500,000 or more a year must arrange an independent evaluation under 45 CFR 2522.710, and AmeriCorps' January 2026 evaluation FAQ says it must be an experimental or quasi-experimental impact study. ED competitions such as the Education Innovation and Research program require an independent evaluation under 34 CFR 75.590. In the 37 nonprofit-eligible NOFOs Grantsights read on October 20, 2026, none made a third-party evaluator a condition of applying, though both DOJ Second Chance Act reentry notices give priority consideration for one. If the notice is silent, an external evaluator is optional.
What are SMART outcomes in a grant evaluation plan?
SMART stands for Specific, Measurable, Achievable, Relevant, and Time-bound. A SMART outcome names what will change, by how much, for whom, measured by what instrument, and by when. For example: 70% of program participants will be employed at 90-day follow-up, as documented by employer verification. Outcomes missing any of the five elements are easy for reviewers to mark down.
How much of my budget should go to evaluation?
No federal rule sets a percentage. Many applicants plan on roughly 10-20% of project costs as a rule of thumb, which would be $50,000 to $100,000 on a $500,000 award, but size it to the design the notice asks for. A performance-measure plan with existing staff costs far less than an independent impact study. Since the 2024 Uniform Guidance revision, 2 CFR 200.455(c) states that data and evaluation costs are allowable, so price them openly in the budget.
What is the difference between process evaluation and outcome evaluation?
Process evaluation measures whether you delivered the program as planned: sessions held, participants enrolled, fidelity to the model. Outcome evaluation measures whether participants changed: knowledge gained, employment reached, symptoms reduced. Process answers did you do what you said, and outcome answers did it work. Most federal notices ask for both, because process data explains weak outcome data.
What agencies have specific evaluation requirements I need to follow?
DOL workforce programs report the six WIOA primary indicators in 20 CFR 677.155. SAMHSA grantees report client-level data through SPARS, where many grants have moved to the SUPRT-A and SUPRT-C tools. ED scores evaluation under 34 CFR 75.210(h) when a competition uses that criterion and defines evidence levels in 34 CFR 77.1. AmeriCorps sets evaluation rules by grant size in 45 CFR 2522.700-740. HUD's Fair Housing Initiatives Program gives 10 of 100 points to measurement and an evaluation plan.
How many points is the evaluation section worth?
It depends on the notice. In the 37 open NOFOs Grantsights read on October 20, 2026, HUD's two Fair Housing Initiatives Program notices gave 10 of 100 base points to Achieving Results and Program Evaluation, and an ACF Office of Refugee Resettlement notice gave 20 of 125 points to its Impact criterion. The 21 DOJ Office of Justice Programs notices had no separate evaluation criterion at all. ED's FY2025 Education Innovation and Research mid-phase notice gave up to 30 of 100 points to evaluation.
Last updated October 20, 2026. Corrected October 20, 2026: what changed. Published by Grantsights from public sources. See our editorial policy, how we build and check guides, corrections and how we make money.
Related Guides
Complete Guide
Federal Grant Proposal Guide: Sections & Scoring
Federal grant proposals average 50-100 pages. This guide covers every section and scoring criteria for writing a proposal that reviewers score highly.
How to Write a Data Management Plan for Grants
NIH, NSF, DOE and NASA each want a data management and sharing plan. Compare 5 agencies' rules, page limits and review, plus 2 templates.
How to Write a Grant Dissemination Plan (With Examples)
Write a dissemination plan that scores well with federal reviewers. Covers 6 required components, audience targeting, and budget for 5+ methods.
How to Write a Grant Management Plan (With Examples)
Write a grant management plan reviewers can score: staffing, 2 CFR 200 financial controls, data plans, and the 25 points SAMHSA gives to staffing.
How to Write a Grant Staffing Plan (With Examples)
Build a grant staffing plan that matches your work plan and budget: FTE math, fringe, consultants, and 2 CFR 200.430 effort records reviewers check.
How to Write a Grant Work Plan (With Template)
Federal grant work plans need objectives, activities, timelines, and responsible staff. Get the format reviewers expect with 3 examples.