A practical approach to measuring nonprofit work without confusing a busy room with a lasting result
By Arbab Naseebullah Kasi, Chief Executive Officer - CEO, Feel Worldwide Foundation Inc.
Looking carefully at programme evidence can reveal more than attendance totals alone.
An attendance sheet can tell a reassuring story. The room was used. People made time to come. The team delivered the sessions it promised. Those are real accomplishments, especially when organising even a small programme involves transport arrangements, competing schedules and limited resources.
But the attendance sheet cannot tell us everything we want to know.
Did participants learn something they can use? Did an obstacle become easier to manage? Did a new skill survive beyond the afternoon when it was taught? A programme can be well attended and still leave those questions unanswered.
For a small nonprofit, the answer is not necessarily a complicated evaluation system. It may be a more careful question, a modest piece of evidence and a willingness to report what remains unknown. Measurement becomes useful when it helps a team decide what to keep, what to change and what it can honestly claim.
Give attendance its proper place
Counting participation is not a mistake. A programme that reaches nobody cannot do much for its intended community. Attendance can reveal practical problems, including a session time that excludes working parents or a venue that people struggle to reach.
The mistake is asking that count to stand in for a different kind of evidence.
The CDC's August 2024 programme-description guidance distinguishes outputs, the products of activities, from outcomes, the changes expected to result. That distinction gives a simple starting point: sessions delivered and materials distributed describe the work performed; changes in knowledge, behaviour or conditions describe what the work hopes to achieve. CDC guidance on describing a programme.
A useful report can contain both. It can say how many people participated and what evidence was collected about their progress. Neither number needs to pretend to be the other.
This matters particularly when an impressive total is easy to produce. Counting every visit may be appropriate for planning staffing, but it is not the same as counting distinct people. If one person attends six sessions, the report should make clear whether it records six attendances or one participant. A small label can prevent a large misunderstanding.
Choose a change that the programme can reasonably influence
Consider an illustrative programme that helps adults practise explaining a project to an audience. Its broad ambition might be greater confidence and opportunity. Those hopes are worthwhile, but they are too expansive to measure with a sign-in sheet or one final questionnaire.
A more manageable question would be whether participants can give a short explanation that identifies the project's purpose, describes the main idea and makes the next step clear.
The team could ask participants to attempt a brief explanation near the beginning and another near the end. It could use the same simple criteria on both occasions, with different but comparable prompts. Participants should know that the exercise is intended to improve the programme and support learning, not embarrass them or decide who deserves help.
This is an example of a possible approach, not a tested claim about that imaginary programme. Its value is that the proposed evidence is close to the activity being delivered.
A later promotion, new job or successful business may involve the skill, but it also involves circumstances beyond the programme's control. Starting with a nearer, observable change makes the evaluation more feasible and the claim more defensible.
Make the starting point visible
A strong final performance does not automatically show improvement. Some participants may have arrived with that ability already. Others may make substantial progress while still needing support.
For the illustrative presentation programme, a final score alone could hide both situations. Comparing each person's starting and later performance would provide more useful information, provided the tasks and scoring are sufficiently comparable.
Even then, the measurement is imperfect. A participant might be unfamiliar with the topic, anxious about speaking or more comfortable using another language. A written response would assess something different from an oral explanation. The team should decide what it actually wants to measure and avoid allowing an unrelated difficulty to dominate the result.
That does not require pretending every small organisation can conduct specialist assessment. It requires describing the measure plainly. A report might explain that facilitators assessed short practice presentations against three shared criteria. That is more informative than announcing an undefined improvement in leadership.
Where possible, have a second person review a few examples. If two reviewers interpret the same criterion very differently, the team has learned something about the measure before using it to make a confident claim.
Keep the denominator attached
Percentages become much less impressive, and much more useful, when readers can see who they describe.
Imagine, purely for illustration, that 40 people join a programme. A later questionnaire receives 24 responses, and 18 respondents say they have used the skill outside the sessions. That is 75 percent of respondents. It is not evidence that 75 percent of all participants used the skill.
The response rate is 60 percent. The remaining 16 participants have unknown outcomes in this example. Their silence is not proof of failure, and it is not permission to assume success.
An honest summary would state the counts: 18 of the 24 respondents reported using the skill; 40 people originally participated. It would also make clear that this is self-reported use, not independently observed performance.
With small groups, counts can be easier to understand than percentages alone. Moving from three people to four can create an eye-catching percentage increase while still representing one person. That individual matters. The report simply needs to keep the scale visible.
The same care applies when people leave a programme early. Reporting only those who finish may be useful for a particular question, but it should not quietly erase those who could not continue.
Separate improvement from proof of cause
If participants improve between the first and last session, that is worth knowing. It does not, by itself, establish that the programme caused all the improvement. People may practise elsewhere, receive support from colleagues or encounter new opportunities during the same period.
The CDC's August 2024 evaluation-design guidance distinguishes outcome evaluation from impact evaluation. Outcome evaluation examines whether intended outcomes were achieved; impact evaluation compares results with estimates of what would have happened without the programme. The latter addresses a harder causal question. CDC guidance on evaluation questions and design.
A small team can still produce valuable evidence without answering every causal question. It should use language that matches its design. Saying participants demonstrated improvement during the programme is different from saying the programme alone produced that improvement.
If a funding or expansion decision depends on a strong causal claim, it may be time to seek an evaluation partner. That decision deserves proper design, appropriate expertise and attention to ethics. It should not be improvised by denying help to people simply to create a comparison group.
A real evaluation shows why time and comparison matter
WorkAdvance offers a public example of the difference between delivering training and studying its effects. Launched in 2011, the model combined sector-focused training with an emphasis on advancement and was implemented by four nonprofit providers.
MDRC's September 2025 report examined a randomized controlled trial ten years later. At St. Nicks Alliance, the programme increased average earnings by 32 percent compared with its control group and increased the share earning at least $45,000 by 7 percentage points in Year 10. The evaluation found no impact for the other three programmes on either of those two prespecified outcomes in that year. All four had increased earnings at some point during the evaluation. Year 10 covered 2021 through 2023 across cohorts, including pandemic-related disruption. MDRC's ten-year WorkAdvance findings.
Those findings should not become a promise that another training programme will produce the same gains. They illustrate why the provider, comparison, outcome and follow-up period belong beside a result.
For a smaller organisation, the practical lesson is not to copy the scale of a decade-long trial. It is to resist treating one positive moment as the whole story. A skill demonstrated on the final day and a skill used months later are related questions, but they are not identical.
Collect only evidence you can responsibly use
More information is not automatically better information. Before adding a question, ask what decision its answer will support and whether a less intrusive measure could serve the same purpose.
A team evaluating a short learning programme may need examples of a task completed, not detailed accounts of family finances or personal hardship. It may need a way to match an initial and later assessment, not a public spreadsheet containing names and individual scores.
These are practical design recommendations rather than a substitute for the privacy, safeguarding or research requirements that may apply to a particular programme. Sensitive work needs appropriate expertise.
Explain what will be collected, who will see it and how it will be used. Keep access limited, decide how long records are needed and avoid publishing small-group details that could identify someone indirectly. A moving personal story is not automatically available for a public report merely because it was shared with a facilitator.
Also consider the burden on participants. A long survey can consume the time the programme was meant to protect. A short, well-chosen task may tell the team more than a questionnaire that asks about everything and clarifies little.
Give the findings a job
Before collecting data, decide when the team will review it and what decisions it could change. Otherwise measurement risks becoming an annual search for attractive numbers.
For the illustrative presentation programme, the review might show that participants explain the main idea clearly but struggle to make a specific request. The next course could include more practice on that step. If participants perform well during sessions but report few opportunities to use the skill afterwards, the team could examine whether the programme's practical expectations were realistic.
A compact internal summary could include the number who started, the number who completed comparable tasks, the observed changes, the missing information and the next adjustment. It need not be a polished dashboard. It needs to be understandable enough that somebody can act on it.
Disappointing findings should remain in the conversation. They may reveal an unsuitable measure, a delivery problem or a theory that needs revision. Removing them would make the report tidier while making the programme harder to improve.
Ultimately, honest measurement is a way of taking the work seriously. People gave their time. Staff and volunteers invested effort. Supporters contributed resources. All of them deserve a description of what is known that does not hide behind either a crowded room or an inflated claim.
What is one change your programme hopes to make, and what modest piece of evidence would help you see whether it is happening?
