Closed questions limit answers to a fixed set of options, like yes/no, multiple choice, or a rating scale. In UX research they power surveys, since responses are quick to give and easy to count, and they show up as clarifying follow-ups during usability testing. Their opposite, open questions, invites people to answer in their own words.
Use cases
Lean on closed questions at the wrong moment and you get clean data that tells you nothing. People pick the nearest option, you chart it, and you miss the reason that mattered.
The survey at scale: You need to know which of five features 2,000 users want most. A closed multiple-choice question gives you countable, comparable answers. An open one would bury you in text.
The screener: Recruiting for a study, you ask "How many times a week do you use [product]?" with set ranges. Fast, consistent, and it filters people in or out cleanly.
The interview that flatlines: A new researcher fires off "Do you like it? Did that work? Was it easy?" The user answers yes, yes, yes, and the session ends with nothing learned.
How it's used in practice
Use them for quant: Surveys, screeners, and rating scales, anywhere you need to count and compare responses.
Cover the answer space: Make options complete and balanced. Missing or lopsided choices push people into answers that aren't true, and that bias is invisible later.
Add an escape hatch: Include "Other," "None," or "Prefer not to say" so nobody's forced into a box that doesn't fit.
Go open first when exploring: If you don't yet know the possible answers, ask it open, then build the closed version from what you hear.
🪄
Pro-tip: "Do" and "did" questions almost always come back closed and shallow. Swap the stem to "how" or "what" and the same question opens up. "Did the checkout work?" becomes "What happened when you checked out?" and suddenly they're telling you the story.
Challenges & limitations
Option bias: Your answer choices cap what people can say. If you didn't think of a reason, they can't report it, and your data looks tidier than reality.
False precision: Clean percentages feel authoritative even when the question was leading or the options were weak. Quant can hide a flawed instrument.
Commonly used tools:
SurveyMonkey — the reliable workhorse. Huge template library, solid for general-purpose closed questions.
Maze — my pick for UX-native work. AI can flag biased phrasing, run dynamic open follow-ups, and it links surveys to prototype tests. Best if you're already testing designs.
Typeform — one-question-at-a-time, genuinely pretty, higher completion rates. Great for consumer-facing surveys, less so for dense research batteries.
