What questions should a stakeholder survey ask?
A stakeholder survey should ask about the experience, priorities and decisions relevant to each group you need to hear from. Start with a clear purpose, choose a small shared core where comparison is useful, and add questions appropriate to the person’s role. Members, staff, partners and customers do not necessarily need the same questionnaire.
Use structured questions when you need a defined measure and open questions when the respondent’s own account will help. Neither format is inherently more valuable. The test is whether the answer will inform a decision and whether the team has a plan to review it.
The examples below are adaptable starting points, not validated instruments. Select and pilot the questions that fit your context; do not send the entire bank to every stakeholder.
A six-question starting set
Imagine a network reviewing how it communicates with participating organizations. A short starting set might include the following. Change the activity and period to match your survey.
Scroll horizontally to see all columns →
| Purpose | Question | Response approach |
|---|---|---|
| Relevant experience | Which network activities did your organization take part in this quarter? | A defined list with multiple selections and “None” |
| Information | How clear was the information about the activities you used? | A labeled clarity scale; route only to relevant respondents |
| Access | What, if anything, made participation difficult? | Optional open response |
| Value | Which activity was most useful to your organization, and how? | Activity selection with an optional example |
| Priority | Which one improvement should the network consider first? | Open response or a tested list with another-option route |
| Missing context | Is there anything else we should understand before planning the next quarter? | Optional open response |
If the respondent answers on behalf of an organization, say so. A representative’s account is not automatically the view of every individual member. Record their role only where it helps interpretation and is appropriate.
Question examples by stakeholder group
Members and association participants
- Which membership service have you used most in the past year?
- What made it easier or harder to take part?
- What would make the membership more useful for your work?
Separate people who did not participate from those who participated and had a poor experience. Their answers may suggest different next steps.
Partners and suppliers
- How clear are the reporting requirements for this period?
- Which handoff or request needs clarification?
- What information would help us plan the next delivery or activity together?
Retain the relevant organization, period and activity context. Avoid asking a supplier to judge work outside its experience.
Customers and service users
- What were you trying to accomplish when you contacted us?
- How easy was it to find the information you needed?
- What remains unresolved?
Explain whether the survey is a route for individual help. If it is not monitored for that purpose, provide the appropriate contact route.
Employees and internal teams
- How clear are the responsibilities for this process?
- What currently makes the work harder than it needs to be?
- What change would help your team work more effectively?
Make confidentiality arrangements clear and avoid combinations of role or location fields that identify someone unnecessarily in a small team.
Community members and program participants
- How did the timing and location work for you?
- What would make participation more accessible?
- What should the team understand about your experience?
Use wording people recognize rather than internal program labels. Offer appropriate alternatives for people who cannot comfortably complete the survey in its main format.
Funders, sponsors and decision partners
- Which part of the current reporting is most useful for your decisions?
- Where do you need clearer evidence or definitions?
- Which question should the next reporting cycle help answer?
Do not let the funder’s reporting preferences replace the questions that matter to people receiving the service. Use both perspectives for their proper purpose.
Write questions without leading the answer
“How much do you value our excellent support?” invites agreement with the organization’s claim. “How useful was the support you received?” allows a wider range of answers. Include a suitable option for someone who did not receive support, or route them past the question.
Ask one concept at a time. “Was communication timely and clear?” combines two things that may differ. If both matter, separate them. Name the period or event rather than asking vaguely about “recent experience.”
For a priority question, distinguish “select one” from “select all that apply.” A list of all relevant concerns does not reveal which one matters most. If you need a ranking, consider the effort involved and use a manageable set.
See question types and open-question design for more examples.
Keep the survey proportionate to the task
There is no universally correct number of questions. A quick event check-in and an annual organization-level return require different effort. Remove questions that do not support a decision, repeat information already available or ask for details respondents cannot reasonably know.
Open questions also create effort. One broad essay question is not necessarily easier than several clear response choices. Test the actual time and experience with intended respondents rather than estimating only from the number of questions.
Plan the review capacity before collecting a large text set. AI-assisted organization can help, but it does not remove the need to check interpretations or make decisions.
Does the whole network need one survey?
No. In a federated network, local teams may need different questions for their own activities. Agree on the limited shared fields that support cross-network questions, and keep local detail where it serves a real purpose.
For example, collect stable organization details through registration and confirm information that changes at an appropriate interval. A quarterly activity return can then focus on the current period rather than repeatedly asking for the same profile.
Define the shared measures in a data dictionary: field meaning, eligible group, reporting period, answer options, units and missing-value rules. A “participant count” must distinguish unique people from attendance entries before it can be aggregated.
Local survey wording may differ where context requires it. Do not assume two questions are comparable merely because their fields share a name. Review mappings and keep the original question with the response.
How do you compare results over time?
Keep important measures sufficiently consistent in meaning, scale and reference period. If wording needs improvement, record the version and assess whether the change affects comparison. Not every minor edit destroys a trend, but a changed construct or population can.
Distinguish the same individuals or organizations followed over time from different respondents in successive periods. Both designs can be useful. Do not claim individual change from a changing group sample.
Report who was asked and who responded in each group. A network-wide average can conceal a local issue or a change in participation. Compare like with like before combining the results.
A worked interpretation
A fictional survey receives 120 responses: 80 from members and 40 from delivery partners. Fifty-six members and 20 partners rate communication as clear or very clear.
- Members: 56 of 80, or 70%.
- Partners: 20 of 40, or 50%.
- Combined: 76 of 120, or 63.3%.
The combined result is useful for a broad view, but the partner result suggests a different review question. It does not prove communication caused another outcome, and the groups may have received different kinds of information.
If 15 partners leave comments and six mention changing deadlines, report that theme among those 15 comments. Check the deadlines and communications before deciding the remedy. Do not infer that all 40 partners experienced the same concern.
Tell contributors how the findings will be used
Before sending the survey, name the owner of the review and the decisions it will inform. Explain whether feedback is anonymous, who can see it and whether personal follow-up is possible.
After review, share an appropriate account of what was heard, what will change and what cannot change now. You do not need to promise to implement every request. A clear explanation is more useful than a vague claim that all feedback is valued.
Keep findings, source context and actions connected. For the full review process, see stakeholder feedback. For presentation, use survey report examples.
Make the collection process maintainable
Sopact’s relevant approach combines collection, contextual analysis and governance for teams managing evidence across contributors and periods. A pilot should include local questions, shared measures, a revised field and a reporting view for each relevant audience.
Check whether the team can trace a result to its definition and source, correct an interpretation and control what is shared. The value is a workflow the team can manage, not a requirement to impose one questionnaire on everyone.
Frequently asked questions
Should all stakeholder groups answer the same questions?
No. Use a shared core only where comparison is meaningful, and add role-specific questions. Keep the question context with each response.
Are open questions better than rating scales?
They serve different purposes. Ratings describe responses to a defined measure; open questions can reveal context or unexpected ideas. Choose based on the decision.
How many questions should we ask?
Use the set needed for the purpose, then pilot its burden. Remove duplication and questions without a clear use rather than following a universal maximum.
Can we change questions between cycles?
Yes, when necessary. Record the change and assess whether the measure remains comparable. Major changes may require a new baseline or separate reporting.
Do we need to identify every respondent?
No. Group-level learning can often use anonymous responses. Individual follow-up or longitudinal matching requires an appropriate design and access controls.
