Surveys are the fastest way to get structured answers from a lot of people.
Setting one up
Describe what you want to measure and the study builder drafts your questions. Edit anything that doesn't fit.
Add a screener so only the right people get through.
Choose your audience and launch.
The screener matters more than most people expect. A tight screener on a smaller audience beats a loose one on a large audience every time, because you end up analysing answers from people who actually have the problem.
Question types
The usual range: single and multiple choice, dropdowns, open text, rating scales, ranking, and image choice. Grid questions for asking the same thing about several items. Net promoter score if you track it.
Mix closed questions for measurement with a small number of open ones for colour. Open text is where you'll find the quotes worth sharing.
Keep it short
Shorter surveys get finished, and finished surveys give better data. Quality falls off noticeably once people feel they're being kept.
A useful test: could you answer this yourself, attentively, in one sitting? If not, cut it. Ask what you'll act on and drop what's merely interesting.
Writing questions that work
Ask one thing at a time. "Was the checkout fast and clear" can't be answered by someone who found it fast but confusing.
Avoid leading. "How helpful was the new dashboard" assumes it was helpful.
Offer an escape. If someone genuinely has no opinion, let them say so rather than forcing a guess that becomes noise in your data.
While it runs
Responses arrive as people complete the survey. You review them, and approving a response releases that participant's payment.
Reviewing promptly matters, participants are waiting on you.
Reading the results
Answers are summarised automatically, with themes pulled from your open-text questions.
Look at the open text first. Closed questions confirm what you suspected; open text tells you what you didn't think to ask.