Outputs, Outcomes, and Impact

An output is what a project produces and controls — sessions delivered, people enrolled. An outcome is a change in the people or systems the project touched, which the project influences but does not control. Impact is the broader, longer-term change that outcomes contribute to and no single project can claim alone.

Key takeaways

• Outputs are within your control; outcomes are only influenced by you.
• The controllability test separates the two in one question.
• Outcomes run in tiers: learning, then action, then condition.
• Claiming impact without a counterfactual costs credibility.
• The outcomes you promise become the outcomes you report.

What is the difference between an output and an outcome?

An output is a direct product of program activity. An outcome is a change in a participant or a system. The W.K. Kellogg Foundation defines outputs as “the direct products of program activities” that “may include types, levels and targets of services to be delivered by the program,” and outcomes as “the specific changes in program participants’ behavior, knowledge, skills, status and level of functioning” (W.K. Kellogg Foundation, 2004).

The distinction is one of ownership. Delivering forty workshops is entirely inside an organization’s control; whether anyone learned anything is not. The Urban Institute states the same boundary from the measurement side: end outcomes are “the consequences/results of what the program did, not what the program itself did” (Lampkin and Hatry, Urban Institute). The table below compares the two across the five dimensions that determine which column a result belongs in.

Dimension	Output	Outcome
What it describes	Work delivered	Change experienced
Degree of control	Full staff control	Influence only
When it is countable	During delivery	After exposure, often with a lag
Value as evidence	Proves effort, not benefit	Proves benefit, if measured well
Typical wording	Delivered, enrolled, distributed	Improved, increased, reduced, sustained

Grant reviewers read the wording as a competence signal. A results table full of verbs like “delivered” and “held,” presented under a heading that says outcomes, says that the applicant has not yet separated activity from benefit. Everything else in the evidence, evaluation, and data section rests on getting that separation right.

What is impact, and when can a project claim it?

Impact is the longer-horizon, broader change that outcomes contribute to. Two definitions of the word circulate, and confusing them is expensive. Kellogg uses impact to mean “fundamental intended or unintended change occurring in organizations, communities or systems as a result of program activities within 7 to 10 years.”

Federal evaluation practice uses the word differently, to mean a causally attributed effect. The CDC defines impact evaluation as an approach that “compares the outcomes of a program, policy, or organization to estimates of what the outcomes would have been without it,” and states that outcome evaluation “cannot determine what caused specific outcomes (causality), only whether they have been achieved” (CDC Approach to Program Evaluation; CDC Program Evaluation Framework, 2024). The Urban Institute uses the federal sense: “Impact measures the achievement of outcomes relative to a comparison, which helps estimate causality” (Urban Institute, Performance-Based Strategies).

In a federal proposal, use the causal meaning. Claiming impact means claiming a counterfactual, and a counterfactual requires a comparison condition — a control group, a matched comparison, a discontinuity, or a credible modeled estimate. A single-group before-and-after comparison is not one, no matter how large the change. That is the difference between attribution and contribution: attribution says the program caused the change, contribution says the program was one of several forces moving in the same direction. Over-claiming attribution is the fastest way to lose a reviewer who has run an evaluation, because the design on the page will not support the verb in the sentence.

How do you tell an output from an outcome?

Apply the controllability test: if program staff can guarantee the result by doing their jobs well, the result is an output. If achieving it also depends on how a participant, an employer, or a system responds, the result is an outcome.

Two worked cases show the line. “Fifty people attended a workshop” is an output — the organization scheduled the room, recruited, and counted heads. “Thirty-five participants demonstrated improved knowledge on a validated instrument ninety days after the workshop” is an outcome — it names a change, in a defined subset, measured with a named instrument, at a stated interval, and it can fail even if attendance was perfect.

The word “difference” is the shortcut. Outputs relate to what a project does; outcomes refer to what difference it makes (Pennsylvania Coalition Against Rape, 2018). Candidate outcomes and indicators for common program types are catalogued in the Urban Institute and Center for What Works Outcome Indicators Project, which is the fastest way to check whether a proposed outcome is measurable at all. Two habits keep the categories clean: write outputs as counts and outcomes as changes with a direction and a baseline, and attach a measurement point to every outcome, because an outcome with no measurement date is a wish.

Funders differ in vocabulary. Some agencies say “performance measures” for the whole set, some say “results,” some use “deliverables” where others say outputs. Federal education regulations define a performance measure as “any quantitative indicator, statistic, or metric used to gauge program or project performance,” which covers both categories (34 CFR 77.1). The same regulations reward applications that propose “specific, measurable targets, connected to strategies, activities, resources, outputs, and outcomes” (34 CFR 75.210). Translate into the funder’s terms, but keep the separation intact in your own planning documents.

What are short, intermediate, and long-term outcomes?

Outcomes arrive in three tiers, and each answers a different question. Short-term outcomes are learning changes — awareness, knowledge, attitudes, skills, intentions. Intermediate outcomes are action changes — behavior, practice, decision-making, policy adoption. Long-term outcomes are condition changes — social, economic, civic, or environmental status (Pennsylvania Coalition Against Rape, 2018).

Kellogg attaches rough horizons: short-term outcomes at one to three years, long-term outcomes at four to six, impact at seven to ten. A three-year award whose only stated outcome sits at the long-term tier has committed to something it cannot observe before the final report is due.

The design rule that follows is to pair tiers. Name the long-term outcome the funder cares about, then name at least one earlier-tier outcome expected to move inside the period of performance, and state why the earlier change should produce the later one. Tiering also determines where the ceiling of accountability sits: outcomes below the ceiling get indicators and targets, and the condition change above it gets described as a contribution rather than promised as a deliverable.

What do paired outputs and outcomes look like?

Paired examples are the fastest way to calibrate. The table below shows the same project activity expressed first as an output and then as the outcome it is supposed to produce, across five program types. The figures are illustrative.

Program type	Output	Corresponding outcome
Job training	180 adults complete a 120-hour certificate course	Share of completers employed in the target occupation within 90 days of exit, against a documented baseline
Food security	42,000 meal-equivalents distributed through a mobile pantry	Reduction in household food insecurity among enrolled families on a validated screening tool at six months
Youth mentoring	1,400 mentor-mentee contact hours logged	Share of matched youth on track for grade promotion at the end of the academic year
Clinical outreach	900 residents screened at community events	Share of residents screening positive who complete a first follow-up appointment within 30 days
Small business support	60 firms receive advisory hours and a financing readiness review	Share of assisted firms securing external financing, and change in employment at the firm 12 months later

Read the right column carefully. Every entry names a population, a direction of change, and a measurement window, and every entry can fail even when the left column is fully delivered. That is what makes it an outcome.

How does the distinction change what you measure and budget?

The output-outcome distinction determines the cost of measurement, which makes it a budget question and not only a vocabulary question. Outputs are captured by systems an organization already runs — attendance sheets, enrollment records, service logs. Outcomes usually require reaching people after service ends, which means tracking contact information, administering an instrument, following up with non-responders, and often paying participant incentives.

Three cost drivers follow from committing to an outcome. Follow-up windows extend staff time past the delivery period. Instruments carry licensing fees or development and validation work. Administrative data — wage records, school records, claims — requires a data-sharing agreement that takes months to execute and sometimes a fee. The National Institute of Justice notes that a process evaluation “may require only a few months,” while “a large-scale outcome evaluation may require years and a substantial financial outlay” (NIJ, 2015).

The cheapest defensible move is to prefer outcomes measurable from data that already exists, and to budget the measurement when you write the outcome. An outcome promised in the narrative with no corresponding cost in the evaluation plan is a commitment the project cannot keep.

What goes wrong when outputs are labeled as outcomes?

The most common failure in first-time applications is a results table where every row is an output sitting under an outcomes heading. Reviewers scoring an outcomes criterion in front of a list of service counts mark the section incomplete, because nothing in it describes a benefit.

Five failures recur, and each has a specific fix. There are five:

• Outputs presented as outcomes. Rewrite each row as a change with a direction, a population, and a measurement point; move service counts to their own table.
• Causal verbs on non-causal designs. “Increased” and “caused” require a comparison condition. Report single-group change as a performance measure.
• Outcomes with no baseline. Federal regulation defines a baseline as “the starting point from which performance is measured and targets are set,” and judges targets against it (34 CFR 77.1).
• Impact claimed on a short award. Condition change on a seven-to-ten-year horizon belongs above the accountability ceiling, as contribution.
• Outcomes that no data source can produce. Every outcome needs a named system, instrument, or record set before it goes in the logic model.

The post-award consequence is the reason to get it right at submission. Federal performance reporting requires “a comparison of accomplishments to the objectives of the Federal award established for the reporting period” and an explanation of why goals were not met (2 CFR 200.329). The outcomes written into the application become the rows of a report filed against them for the life of the award — which is why the objectives you write should be ones you would be content to be graded on.

Frequently asked questions

Are outputs worth reporting at all?

Yes. Outputs establish that the program was delivered at the intended dose, which is the only thing that lets anyone interpret a disappointing outcome. Without output data, a null result is unreadable: the theory may have been wrong, or the service may never have reached anyone.

Can an output ever be an outcome?

For a capacity-building or systems-change grant, a product can be the intended change — an adopted policy, a functioning referral network, a data system in production. The test still applies: adoption by an entity you do not control is an outcome; producing the document is an output.

How many outcomes should a proposal commit to?

Enough to cover the objectives, and few enough to measure well. Most competitive proposals carry two to four outcome objectives, each with a baseline, a target, a data source, and a measurement schedule. Additional outcomes add reporting obligations for the life of the award.

What if the funder uses “impact” to mean long-term outcomes?

Follow the funder’s vocabulary in the application and keep the distinction in your own planning documents. When a solicitation uses “impact” loosely, avoid causal verbs unless your design supports them, and describe the long-horizon change as a contribution.

Should outcome targets be set before the baseline is known?

No. If no baseline exists, say so plainly and budget a baseline collection period in the first year rather than inventing a starting value. A target floating free of a starting point cannot be judged ambitious or unambitious, which makes the whole target set unscoreable.

Related topics

• Evidence, Evaluation, and Data — the hub for proving a project will work, and then proving it did
• Performance Measurement and Indicators
• Levels of Evidence in Grant Funding
• Public Data Sources for Needs Statements
• Grant Reporting Requirements

Sources

1. W.K. Kellogg Foundation. Logic Model Development Guide. 2004. https://www.naccho.org/uploads/downloadable-resources/Programs/Public-Health-Infrastructure/KelloggLogicModelGuide_161122_162808.pdf (accessed 2026-08-11)
2. Lampkin, L. M., and Hatry, H. P. Key Steps in Outcome Management. Urban Institute. https://www.urban.org/sites/default/files/publication/42736/310776-Key-Steps-in-Outcome-Management.PDF (accessed 2026-08-11)
3. Marazzi, M. Performance-Based Strategies: Defining Terms and Comparing Common Strategies. Urban Institute, July 2019. https://www.urban.org/sites/default/files/publication/100452/performance-based_strategies_defining_terms_and_comparing_common_strategies_2.pdf (accessed 2026-08-11)
4. Centers for Disease Control and Prevention. CDC Approach to Program Evaluation. https://www.cdc.gov/evaluation/php/about/index.html (accessed 2026-08-11)
5. Kidder, D. P., Fierro, L. A., Luna, E., et al. CDC Program Evaluation Framework, 2024. MMWR Recommendations and Reports 73(RR-6):1–37. https://www.cdc.gov/mmwr/volumes/73/rr/rr7306a1.htm (accessed 2026-08-11)
6. Pennsylvania Coalition Against Rape. Theory of Change and Logic Models. 2018. https://pcar.org/sites/default/files/resource-pdfs/tab_2018_logic_models_508.pdf (accessed 2026-08-11)
7. U.S. Department of Education. 34 CFR 77.1 — Definitions that apply to all Department programs. Legal Information Institute, Cornell Law School. https://www.law.cornell.edu/cfr/text/34/77.1 (accessed 2026-08-11)
8. U.S. Department of Education. 34 CFR 75.210 — General selection criteria. Legal Information Institute, Cornell Law School. https://www.law.cornell.edu/cfr/text/34/75.210 (accessed 2026-08-11)
9. Office of Management and Budget. 2 CFR 200.329 — Monitoring and reporting program performance. Legal Information Institute, Cornell Law School. https://www.law.cornell.edu/cfr/text/2/200.329 (accessed 2026-08-11)
10. National Institute of Justice. Plan for Program Evaluation from the Start. March 1, 2015. https://nij.ojp.gov/topics/articles/plan-program-evaluation-start (accessed 2026-08-11)
11. Urban Institute and The Center for What Works. Outcome Indicators Project. https://www.urban.org/policy-centers/cross-center-initiatives/performance-management-measurement/projects/nonprofit-organizations/projects-focused-nonprofit-organizations/outcome-indicators-project (accessed 2026-08-11)