What are impact survey questions?
Impact survey questions collect evidence about people's experiences, outcomes and the contribution an initiative may have made. They help distinguish what was delivered, what changed and what still needs explanation. A survey can inform an impact evaluation; a participant's answer alone does not establish causality.
This bank includes 26 adaptable questions for programs, training, services and member initiatives. Each question has a suggested response format and timing. Select the items that serve your decisions rather than sending the entire bank to everyone.
The examples are not a validated scale. Test them with the intended respondents, use suitable established instruments where needed and follow any instrument's administration rules. Keep the purpose, language and collection burden clear.
Choose questions from your outcomes, not from the size of the bank
Start with the change you hope to support and the evidence needed to examine it. A training initiative may need to distinguish confidence, demonstrated skill and later application. A membership initiative may need to understand access to opportunities and whether local support is useful. These questions do not necessarily require the same design.
Scroll horizontally to see all columns →
| Evidence need | What a survey can contribute | What it cannot establish alone |
|---|---|---|
| Participation and delivery | What respondents say they received and barriers they experienced | The accuracy of every administrative count |
| Outcomes | Reported circumstances, perceptions or practices at a defined point | That self-reported confidence equals demonstrated skill |
| Persistence | Whether respondents report a change holding or being used later | The experience of people who did not respond |
| Contribution | Respondents' explanations of the program's role and other influences | A definitive estimate of what would have happened without the program |
Use administrative or observational evidence when it is more appropriate than asking respondents to recall information. Sensitive information, such as income or household circumstances, should be collected only when necessary and with suitable arrangements. If income is needed, specify the currency, reference period and measure rather than mixing hourly pay and annual salary in one field.
Participation and access: what was received?
Scroll horizontally to see all columns →
| Example question | Response format | When it may fit |
|---|---|---|
| Which program activities did you take part in during [period]? | Defined activity list; allow more than one where appropriate | During or exit |
| How many sessions did you attend during [period]? | Count; allow unsure; use reliable attendance records instead when available | During or exit |
| Did you complete [clearly defined requirement]? | Yes / No / In progress / Unsure, as appropriate | Exit or relevant milestone |
| Did you receive [specific material or service] during [period]? | Yes / No / Unsure | During or exit |
| Did you meet with a mentor or coach during [period]? | Yes / No; add frequency only if needed | During or exit |
| What, if anything, made it difficult to take part? | Optional open text | During or exit |
| Which access arrangements would make participation easier? | Relevant options and optional additional response | Before or during |
| What was useful or not useful about [specific activity]? | Optional open text | During or exit |
Outcomes: what is different or possible now?
Scroll horizontally to see all columns →
| Example question | Response format | When it may fit |
|---|---|---|
| How confident are you that you can [specific task] at present? | Not at all / Slightly / Moderately / Very / Extremely confident; separate unable to judge | Baseline and suitable follow-up |
| How much support do you currently need to [specific task]? | Clearly defined support levels; separate task not applicable | Baseline and suitable follow-up |
| How clear are you about your next step toward [defined goal]? | Not at all / Slightly / Moderately / Very / Completely clear | Baseline and suitable follow-up |
| Do you currently have someone you can ask for help with [specific task]? | Yes / No / Unsure; optional prefer not to answer | Baseline and suitable follow-up |
| During the past 30 days, on how many days did you use [specific skill] outside the program? | 0–30 days / Unsure / No opportunity to use it | When the recall period is meaningful |
| Are you currently doing paid work, including self-employment? | Yes / No / Prefer not to answer; ask hours separately if needed | Baseline and relevant follow-up |
| Are you currently enrolled in education or training outside this program? | Yes / No / Prefer not to answer; clarify the definition | Baseline and relevant follow-up |
| Have you obtained [specific credential]? | Yes / No / Awaiting result / Not applicable | Relevant milestone or follow-up |
| What, if anything, has changed in your ability to [specific task] during [period]? | Optional open text; permits no change or decline | Exit or follow-up |
| What has helped or made it difficult to use [specific skill]? | Optional open text | Follow-up |
Confidence, support needs and observed performance are different constructs. Use an appropriate task or assessment when the question requires demonstrated ability. The 30-day application question also distinguishes no opportunity from no use; those findings may call for different responses.
Longer-term change and perceived contribution
Scroll horizontally to see all columns →
| Example question | Response format | When it may fit |
|---|---|---|
| Are you still using [specific practice] in a relevant setting? | Yes / No / No current opportunity / Unsure | Timing suited to the outcome |
| Has [defined outcome] improved, stayed the same or worsened since [reference point]? | Improved / Same / Worsened / Unable to judge | Relevant follow-up; label retrospective judgment |
| If your work situation changed, what changed? | Optional relevant categories and open text | Follow-up, with appropriate routing |
| What role, if any, did the program play in the change you described? | Optional open text | After the respondent describes a change |
| What other people, opportunities or circumstances influenced that change? | Optional open text | Follow-up |
| What do you think might have happened without the program? | Optional open text; allow unable to judge | Where useful for contribution review |
| Were there any unexpected positive or negative effects of taking part? | Yes / No / Unsure with optional explanation | During, exit or follow-up |
| What additional support, if any, would be useful now? | Optional open text or suitable options | Relevant follow-up |
These questions capture perceived contribution and context. They do not turn respondents into observers of a real counterfactual. Combine accounts with the wider evidence and an evaluation design suited to the causal question. BetterEvaluation explains the role of causal reasoning and methods in impact evaluation.
Write response options people can interpret
Use plain language and a specific reference period. Ask one thing at a time. Avoid overlapping choices, such as “employed” and “employed part-time” in a single-choice list, or “once” and “monthly” within a one-month frequency question.
For a single-choice item, make the choices cover the reasonable responses without overlap. Keep “not applicable,” “unsure” or “prefer not to answer” separate where those are meaningful. Do not silently treat them as zero or as the midpoint of a scale.
The best scale depends on the construct. There is no rule that every question needs five agreement points. A count, a factual category, a clearly labelled frequency scale or an established instrument may be more suitable. Pew Research Center's question-writing guidance explains how wording and response options influence answers.
Read survey design for the broader questionnaire process and qualitative questions for eliciting useful explanations without leading respondents.
Plan the timing and matching deliberately
A baseline and later measurement can support analysis of within-person change when the records can appropriately be matched. A further follow-up may help examine persistence. The necessary timing depends on the outcome; three waves are not mandatory for every study.
Not every question belongs at every wave. A question about the usefulness of a completed session does not belong before it takes place. A current-state measure may work at baseline and follow-up. A question about using a skill outside the program needs a meaningful opportunity and recall period.
Keep wording, response categories, mode and administration comparable where the analysis depends on comparability. Record changes to the instrument and assess their implications. A necessary improvement should not be prohibited forever; it may require a new baseline, bridging work or a break in the trend.
Use appropriate identifiers when within-person analysis is required. Anonymous repeated surveys can still answer group-level questions, but changes may reflect different respondents. Do not claim that each person improved from two unrelated group averages. The practical methods are covered in pre- and post-surveys.
Worked example: report the group you actually observed
Fictional example. A program has 120 eligible participants. Ninety answer at baseline, 75 at follow-up, and 60 have a usable response at both points. The paired comparison covers 60 people, or 50% of the eligible cohort.
If 42 of those 60 show an increase on the defined measure, report 70% of the matched group. Do not describe this as 70% of all 120 participants. Report the missing and unmatched responses, how the measure was interpreted and whether other available evidence suggests that respondents differ from nonrespondents.
Some follow-up comments may describe lack of opportunity, practical support or changes outside the program. Those accounts can help investigate the pattern. They do not automatically explain every score or prove the cause of the increase.
Ask for explanations where they add value
One or two well-placed open-ended questions may be more useful than requiring a written explanation after every rating. Use routing where a particular answer needs clarification and allow people to skip optional narrative questions. Balance interpretive value with burden, especially on mobile devices.
As responses grow, use a defined codebook to make themes reviewable. Sopact's approach keeps human control of those definitions while supporting application across eligible responses and analysis alongside the relevant numbers. When a definition changes, review the revised analysis with its version and sources rather than quietly mixing old and new categories.
The benefit is less repeated coding and spreadsheet joining, with evidence available for review. It is not a guarantee that an AI interpretation is correct. See the full visual, video and illustrated effort comparison in qualitative and quantitative analysis, or the study-design guidance in mixed-method surveys.
Map questions to frameworks without pretending all frameworks are the same
Reuse evidence where the definitions align. A new reporting request does not always require a new survey, but it may expose a genuine measurement gap. Check the required population, period, unit, source and calculation before mapping one answer into several reporting rows.
Scroll horizontally to see all columns →
| Framework or tool | How the questions may help | Additional work |
|---|---|---|
| Theory of change | Explore outcomes, conditions and perceived contribution | Examine the reasoning and other influences |
| Logic model | Supply evidence about specified outputs and outcomes | Keep resources, activities and outcomes distinct |
| IRIS+ | Contribute data to relevant defined metrics | Follow the metric's actual definitions and calculation guidance |
| Five dimensions of impact | Inform questions about what changed, for whom and how much | Assess contribution and risk using the wider evidence |
| SROI | Inform outcome descriptions and stakeholder perspectives | Substantiate valuation, duration and adjustments with appropriate evidence |
IRIS+ describes its metrics and related guidance; a metric is not necessarily a survey question to copy verbatim. Confirm how the requested measure is defined and assembled. Start with your theory of change and the five dimensions when choosing the learning questions.
Allow local questions while preserving a shared core
Across schools, chapters, programs or partners, identify the few fields that need valid comparison. Record them in a common dictionary with their eligibility, timing, units and missing-value rules. Keep local questions that support local decisions.
Separate stable registration data from changing circumstances and repeated outcome measures. Collect stable details once where appropriate and provide a way to update them. Do not impose one complete questionnaire on every stakeholder simply to make aggregation easier.
If two sites use incompatible outcome definitions, report them separately or develop an appropriate harmonization method. A shared label in a dashboard does not make the underlying evidence equivalent.
Turn the bank into a working instrument
- Select the decision. Write what the team will do differently with the answers.
- Choose the necessary items. Remove questions that duplicate reliable existing information or have no planned use.
- Define sources and timing. Specify the relevant person, organization or group, collection window and matching design.
- Test comprehension. Check wording, translations, response options and routing with suitable respondents.
- Prepare the review. Agree definitions, access, missingness reporting and who interprets uncertain findings.
- Return the findings. Explain what was learned and what will happen next.
Existing survey tools can support repeated collection and contextual data in different ways. Evaluate the full workflow rather than assuming they cannot connect respondents. Sopact's distinction is the combination of recurring collection, record context, governed analysis and source review that a team can manage without repeated manual assembly.
Test one cycle with a changed question, an unmatched response, a revised theme and a restricted record. Reproduce a reported result and inspect its sources. Compare staff time spent configuring, reconciling, coding, checking and correcting the report; do not judge implementation effort by the form-building step alone.
Continue from questions to evidence and reporting
The Impact Measurement & Reporting course connects the measurement plan to analysis and reporting. Use survey-data analysis for the next analytical steps, then the impact-report writing guide and report examples to communicate the findings and limits.
Watch: collecting the rating and the explanation together
This companion video discusses how quantitative responses and open-ended evidence can remain connected for analysis. The survey still needs clear questions, an appropriate design and human review.
Frequently asked questions
What is a social impact survey?
It collects evidence about social outcomes and relevant experiences, such as access to opportunity, learning or changes in circumstances. Its design should follow the initiative's questions and population rather than a universal template.
What are examples of impact survey questions?
Examples include how often someone uses a specific skill, what changed in their circumstances, what barriers remain and what role they think the program played. Choose a suitable reference period and response format, and distinguish self-report from verified outcomes.
Are satisfaction questions useful?
Yes. They can inform experience and delivery improvements. Satisfaction is different from learning, behavior or longer-term outcomes and should not be used as a substitute for those measures.
Do impact surveys always require three waves?
No. Timing depends on the outcome and evaluation design. A baseline and follow-up may support a matched comparison; further waves may examine persistence. Some questions appropriately belong at one point only.
Can a participant tell us what would have happened without the program?
They can offer a perspective, which may be useful evidence. It is not an observed counterfactual or definitive causal estimate. Interpret it alongside other evidence and an appropriate evaluation method.
How many questions should I ask?
Ask the smallest set that serves the decisions and necessary measures. Test completion burden rather than applying a universal number. Not every rating needs its own mandatory open-ended response.
