What is a survey report?
A survey report explains why a survey was conducted, how the evidence was collected, what the responses show and what conclusions can reasonably follow. It should let a reader trace a finding to its question, sample and analysis rather than relying on a chart without context.
The four examples below use illustrative data, not customer results. Each includes a question, a finding, an interpretation and an action. They cover training, customer experience, employee feedback and partner reporting.
How to write a survey report
- State the decision or question the survey is intended to inform.
- Describe the population, recruitment, dates, mode and number of responses.
- Check the data, including missing answers, duplicates and score direction.
- Present findings with their denominators and relevant comparisons.
- Use qualitative evidence to explain, qualify or challenge the numbers.
- Separate interpretation from observed results.
- Finish with practical actions, owners and the next review point.
AAPOR’s reporting guidance emphasizes transparency about how a survey was conducted, including question wording, sampling, recruitment and other methodological details. Make those details available even when the main report is short. Read the disclosure standards.
Example 1: a training follow-up report
Question: are learners using the skill after training, and what is preventing application? One hundred people enrolled, 80 completed and 60 completers answered the follow-up. Of those 60, 42 reported using the skill.
| Report element | Illustrative wording |
|---|---|
| Finding | 42 of 60 follow-up respondents (70%) reported using the skill. Follow-up covered 60 of 80 completers (75%). |
| Qualitative detail | Review open responses for opportunity to practice, equipment, time and support; report actual coded evidence rather than assuming a barrier. |
| Limitation | Application is self-reported and the status of 20 nonresponding completers is unknown. |
| Action | Investigate barriers and agree a follow-up with the delivery lead before changing course content. |
Do not write “70% of all learners now apply the skill.” The figure describes the people who responded, not every enrollee. If the report includes pre/post scores, show the matched sample separately.
Example 2: a customer experience report
Question: where is the service failing to resolve customer issues? Suppose 200 recent service users were invited and 120 answered. Ninety said their issue was resolved and 30 said it was not.
| Report element | Illustrative wording |
|---|---|
| Finding | 90 of 120 respondents (75%) reported resolution. |
| Comparison | Compare resolution by issue category only where category definitions and group sizes support interpretation. |
| Qualitative detail | Use comments to distinguish a delayed response, an incomplete fix and an unclear explanation. |
| Limitation | Respondents may differ from customers who did not answer. Closed support cases are not necessarily customer-confirmed resolutions. |
| Action | Assign the largest verified unresolved issue category to an owner and review subsequent resolution evidence. |
If a customer’s earlier contacts are linked with permission, the report can examine whether repeated issues preceded an unresolved response. That is a useful pattern to investigate, not automatic evidence of the cause of churn.
Example 3: an employee feedback report
Question: do employees have the information they need during a process change? In this illustration, 90 of 150 invited employees respond. Fifty-four agree that responsibilities are clear, 18 are neutral and 18 disagree.
| Report element | Illustrative wording |
|---|---|
| Finding | 60% of respondents agreed that responsibilities were clear; 20% were neutral and 20% disagreed. |
| Qualitative detail | Review comments for specific unclear handoffs, instructions or decision rights. |
| Privacy | Use group reporting with an appropriate minimum reporting threshold; remove identifying details from quotations. |
| Limitation | Do not attribute a group result to a named employee or infer individual intent from a short comment. |
| Action | Clarify the relevant handoff and repeat the question after people have experienced the change. |
A threshold alone does not guarantee confidentiality. A role, location or distinctive incident may identify someone even without a name. Review the full presentation before sharing it.
Example 4: a partner reporting survey
Question: are partners able to provide consistent quarterly evidence? Suppose 30 partners are expected to submit, 24 respond and 18 provide all required fields. Supporting documents need a separate quality review.
| Report element | Illustrative wording |
|---|---|
| Coverage | 24 of 30 expected partners submitted (80%). |
| Completeness | 18 of 24 submissions were complete (75%); that is 60% of all expected submissions. |
| Qualitative detail | Examine explanations of missing metrics or documents and distinguish unavailable data from unclear definitions. |
| Limitation | Completeness does not establish accuracy. Evidence may still use inconsistent periods, units or populations. |
| Action | Resolve definition conflicts, request missing evidence and keep a record of corrections. |
Keep each submission tied to the partner, reporting period and metric definition. A financial report or audit should remain available as source evidence rather than being reduced to an unchecked summary.
A survey report format you can reuse
| Section | Include |
|---|---|
| Title and scope | Topic, population and period. |
| Executive summary | The main findings, their limits and the decision they inform. |
| Method | Recruitment, sample, mode, dates, wording and analysis approach. |
| Results | Counts, denominators, relevant distributions and qualitative findings. |
| Interpretation | What the evidence may mean and other explanations to consider. |
| Limitations | Missingness, selection, measurement and comparison limits. |
| Actions | Owner, next step and review date. |
| Appendix | Questionnaire, definitions and detailed tables where needed. |
Keep the first page useful to a busy reader. Put detail in an accessible appendix rather than removing it. A good short report is selective about presentation, not selective about evidence.
How to report qualitative findings
Describe the analysis approach and use a consistent unit: respondents with a theme, responses or coded mentions. Those are not interchangeable. If one response can receive several codes, say so and do not imply the categories sum to the full sample.
Use a quotation to illustrate a finding, with appropriate permission and de-identification. Include a contrary or qualified account when it matters to interpretation. A striking quotation should not replace evidence about the pattern.
AI can help organize comments and draft a synthesis, but reviewers should check quotations, counts and source links before release. Preserve the difference between what a person said and what the analyst concluded.
Make the next reporting cycle easier to explain
Keep a small reporting record alongside the finished document: the question version, included population, reporting period, calculation, source records, reviewer and release date. A chart may be copied into a presentation, but these details should remain available to the person who needs to check it.
Consider a customer report in which 30 of 120 respondents describe an unresolved issue. At the next check-in, 12 of those 30 reply and nine report resolution. You can say that nine of the 12 follow-up respondents reported resolution. You cannot say that 75% of all 30 original issues were resolved: 18 customers have no follow-up response. Keep completed staff actions separate from customer-confirmed resolution.
If locations collect different local questions, agree on the limited shared measures before aggregation. Record their units, eligible populations and time windows in a data dictionary. Keep local findings visible where no fair shared comparison is possible. Standardizing a report's appearance does not make different measures equivalent.
Give each proposed action an owner and a review date. At the next report, revisit the original finding, the action and the new evidence together. That makes reporting useful during the work, instead of requiring a fresh explanation from disconnected files each time. The CSAT software evaluation guide shows what to test when choosing tools for that customer workflow.
Final checks before sharing
- Every percentage has an identifiable denominator.
- Charts and prose use the same numbers and definitions.
- The period and population are clear.
- Matched, unmatched and subgroup analyses are labeled.
- No observational association is described as proven causation.
- Quotations and small groups have been reviewed for privacy.
- Recommendations follow from findings rather than a predetermined product pitch.
For the design behind these reports, continue to mixed-method surveys or pre and post surveys.
Continue learning in the Academy
Use these existing Academy guides for the practical next step. They are suggested companion readings; follow each guide’s course navigation for the full sequence.
Use the practical guides to connect the model to a collection plan, review new evidence and decide what to improve.
Frequently asked questions
How long should a survey report be?
Long enough to answer the reporting question transparently. Use a concise summary with methods and detailed tables available in an appendix.
What belongs in the executive summary?
The question, the most important findings, material limitations and the action or decision the evidence supports.
Should every question appear in the main report?
No. Prioritize the questions relevant to the decision, but make the full questionnaire and appropriate supporting results available.
Can I use a margin of error for any survey?
No. A conventional sampling margin of error depends on the sampling design. Do not apply it automatically to an opt-in or convenience sample.
Can a survey report include interviews and documents?
Yes, if their role and methods are clear. Identify which source supports each finding and do not present unlike evidence as if it came from the same survey.
