Survey question types, and when to use each
A guide to Versive survey question types, from AI questions to ratings, rankings, and structured tasks, and when to use each.
Choose a survey question type based on what you need to learn and how you plan to analyze the answer. Versive's builder has more than 20 types across three broad groups: qualitative questions for motivations and explanations, quantitative questions for comparable results, and structured exercises for observing a participant complete a task.
Qualitative questions for motivations and context
Use these to collect explanations and context in a participant's own words.
AI Question starts with one open-ended question. The AI interviewer then asks adaptive follow-ups based on the participant's response instead of using a scripted list. Use it for motivations or reasoning you cannot fully anticipate, such as "why did you cancel your subscription?" You can cap the number of follow-ups for a more predictable length or leave it unlimited. See AI follow-up questions: probing without a moderator for details.
Exploratory questions use a research goal and a turn budget instead of one fixed question. The default budget is eight turns, and the conversation continues until the goal is covered or the budget runs out. Use this for broad topics such as a participant's full onboarding experience.
Open Text collects a long-form written or spoken answer with no AI follow-up. Use it when you want a verbatim quote or a paragraph of context but do not need an adaptive conversation, or when you are running a study in Survey mode, where the conversational types are not available.
Short Text is for a single line: a job title, a product name, or one word that describes your app. Use it for brief labels and identifiers.
Quantitative questions for comparable data
Structured types give every participant the same set of options, allowing you to chart, filter, and compare results directly.
Choice, ranking, and sorting
Multiple Choice supports single- or multi-select options, an optional "Other" free-text option, and randomized order to control for bias. With randomized order on, you can anchor options such as "None of these" so they keep their place while the rest shuffle. For long lists, such as every US state, show the options in two or three columns and add a search box. Use it for a known, finite answer set, including screening questions that check a specific condition before letting someone into a study. See How to write screener questions that actually screen.
Image Choice works like Multiple Choice, with a picture for each answer: up to 6 pictures, random order by default, and a "None of these" answer that always stays last. Use it to compare packaging, logos, or concepts. In a survey, follow it with an Open Text question that shows the picture they picked to ask why.
Card Sort asks participants to sort cards into categories. Use this standard information architecture method to test whether your naming and grouping match participants' mental models. See Card sorting: a practical guide for setup guidance.
Ranking has participants order a list of options by preference, either by drag-and-drop or a dropdown. Use it to measure relative priority. Ranking requires participants to make trade-offs that a rating scale does not. With a long list, turn on Rank top N so participants rank only their favorites, for example their top 3 of 15.
Matrix Ranking extends that to multiple criteria at once, so participants might rank the same features by importance and separately by satisfaction in a single question. Use it when you need two dimensions of priority without running two separate questions.
Allocation has participants distribute a fixed budget of points, 100 by default, across a set of options. It fits questions like "how would you split your budget across these features," where relative weight matters more than a simple rank order. Switch it to sliders, one per option with a running total, for a breakdown such as a usage share that must add up to 100%. Turn on the N/A option to give each option a checkbox labeled with your own text, such as "Unable to assess", that participants tick instead of entering a value.
Conjoint runs choice-based conjoint analysis: participants repeatedly choose between product concepts assembled from attributes you define, such as price, features, or support tier, and Versive computes attribute importance and level utilities from the results automatically. Use it for pricing and feature trade-off research when direct questions do not reveal relative priorities.
MaxDiff shows participants a few items at a time, such as features or claims, and asks them to pick the most and the least important in each set. Versive builds the sets so each item appears about equally often, then ranks every item by a preference share and a best-minus-worst score. Use it to prioritize a long list, from 6 to 30 items, when a rating scale would rate everything as important.
Tree Test measures findability in a navigation structure. Participants get a task and locate a node in your hierarchy without visual design cues. Use it to validate an information architecture before you build it. See Tree testing: a practical guide.
Matrix Selection asks participants to pick one answer for each row, with the same answer choices for every row. Show it as a grid of radio buttons, or as a dropdown per row when the answer text is long. Use it for a batch of labelled choices, such as how often someone uses each of five tools, without writing five separate questions.
Ratings and scored feedback
Rating Scale is a Likert-style scale, five points by default, with labels you set for the low, middle, and high ends. Use it for general agreement or satisfaction questions you want to track over time. Switch it to a slider for wide or fine ranges, such as a 0–100 % likelihood, with an optional N/A answer.
Star Rating presents an adjustable rating scale with familiar icons such as stars, hearts, or paws. Use it where a consumer-style rating fits the tone better than a numbered scale.
NPS is the standard Net Promoter Score question on the fixed 0 to 10 scale, with customizable end labels. Use it specifically for loyalty and referral likelihood, and keep the wording standard if you want to compare scores against a benchmark over time.
Matrix Rating Scale lets participants rate several items on the same scale in one question, with an optional N/A column and randomized row order. Use it to keep a batch of similar ratings, such as rating five features on the same scale, from turning into five separate questions.
Structured exercises for observing tasks
These types record task behavior or collect a specific artifact.
Prototype Task sends participants into a Figma prototype and tracks their path automatically against a goal node you define, recording each attempt as not started, in progress, completed, or gave up. Use it to observe whether participants complete a flow.
Website Task tracks task completion on a live website, either in a new tab or an embedded browser inside the study. Optional voice recording lets participants think aloud as they browse. Use it to test a shipped experience instead of a prototype.
Website Link redirects a participant to an external URL, such as a partner survey or booking page. Use it as a handoff step in the study flow.
Media Upload lets participants attach images or videos, with a setting for which file types are allowed; file count and size are capped automatically. Use it for diary-style evidence, such as a screenshot of an error or a photo of a product in use at home.
Email captures a valid email address, useful when you need to follow up or send an incentive.
Number collects a numeric answer with optional minimum and maximum validation, for anything from a budget figure to a usage count. Turn on the N/A option to give participants a checkbox labeled with your own text, such as "I do not plan to", that they tick instead of typing 0.
Calendar embeds a Calendly scheduling page inside the study, for booking a follow-up session such as a deeper interview.
Survey mode compatibility
Versive's Survey engine, the structured no-AI mode, hides the two conversational types: AI Question and Exploratory. For a fixed-format survey, use Open Text, Short Text, and the choice, ranking, rating, and utility types above. Adaptive follow-ups are available only in AI-moderated studies.
A study can mix all three groups. For example, use multiple choice for screening, AI questions for explanations, and rating or ranking questions for comparable results. You can choose each type from the question library or ask the AI assistant to generate a mix for you to review.
Frequently asked questions
How many question types does Versive support?
Versive offers 20+ question types, spanning open response, choices and sorting, ratings and feedback, and task-based types such as prototype and website tasks.
What is the difference between an AI question and a multiple choice question?
An AI question lets the interviewer ask adaptive follow-ups based on what a participant says, while types like multiple choice, rating scale, and NPS collect a fixed, structured answer with no follow-up.
Are AI questions available in Survey mode?
No, Survey mode is a structured, no-AI engine that hides the AI question and exploratory types and uses only the structured question types.
Full reference
Question types
Keep reading
AI follow-up questions: probing without a moderator
How AI follow-up questions work in a Versive AI-moderated survey: how the interviewer probes, how to steer it, and when to cap follow-ups.
AI-moderated usability testing with real participants
Run Figma and website tasks with real participants in Versive, capture think-aloud feedback and screen recordings, and use AI follow-up questions.
How to write screener questions that actually screen
Screener questions examples, the principles behind good ones, and how to wire screen-out logic in Versive so screeners actually screen.
