A survey gives a team a way to ask many people the same questions, but the questionnaire is only the most visible part of the work. Deciding whose answers are needed, reaching those people and interpreting who actually replied can matter as much as polishing the form. A beautifully worded survey sent to the wrong audience will still answer the wrong question.

Survey research collects information from a defined group through questions, usually with enough consistency to compare responses. It can examine reported behaviour, experiences and attitudes, and may combine fixed response options with space for people to answer in their own words.

Decide what a survey can resolve

Surveys are useful when the team can state a question that respondents understand and can answer from their experience. They can establish how responses are distributed within a sample, explore relationships and track a measure over time. They are less suited to uncovering a complicated sequence of actions that people barely noticed while carrying it out.

Imagine a fictional team investigating abandoned report exports. A survey might ask recent users what they were trying to produce and whether they encountered a problem. Watching someone attempt the export in usability testing could reveal an overlooked control that the person would never think to mention. The two methods contribute different evidence; neither needs to carry the entire investigation.

Questions about future behaviour need particular restraint. People can report an intention to subscribe, but that answer is not an observed purchase. Treat the distance between the two as something to investigate rather than remove through optimistic wording.

Define the audience before choosing the distribution channel

An in-product invitation reaches people who are still using the product. An email to paying administrators may miss the colleagues doing the daily work. A panel can reach beyond the customer base, but eligibility and recruitment still determine which population the results can describe.

Specify the audience and the selection process before fieldwork. If the study needs an estimate for a wider population, plan the sampling and analysis accordingly; increasing the number of convenient respondents does not automatically create a representative sample. AAPOR’s best practices offer a reference for planning the survey process, including recruitment, questionnaire testing and transparent reporting.

Build and test the instrument

Good questionnaire design starts with the decision the study will inform. Use clear reference periods, avoid combining separate ideas in one question and provide response options that fit the intended audience’s circumstances. An open question can reveal a missing explanation, but a long series of text boxes may demand more effort than respondents are willing to give.

Pilot the whole journey, from invitation through submission. Check how people understand the questions, whether routing works and how the form behaves on a phone. Ballpark’s survey tools can support collection, while the pilot establishes whether the study you have built is ready to send.

Report the result with its boundaries

Keep the question wording, fieldwork dates, recruitment method and respondent count alongside the findings. For each percentage, make clear who is included in the denominator, especially when routing or skipped answers change it. Examine missing responses rather than quietly treating them as neutral opinions.

A survey result should leave readers understanding both the finding and the people it describes. That clarity makes it easier to decide whether to act, investigate a particular group or gather a different kind of evidence before committing to a change.