Survey Question Types: The Complete Guide to 14 Question Types with Examples (2026)

The complete list of survey question types

Survey questions fall into two parents and fourteen children. The two parents are open-ended (qualitative — respondents write in their own words) and closed-ended (quantitative — respondents pick from defined options). Every survey question type is a variation on those two ideas, optimized for a specific data goal.

Here is the complete map, ranked by how often each shows up in real research:

The rest of this guide walks through each type — when to use it, how to write it well, the pitfalls, and how AI-native research platforms like Koji collapse this entire taxonomy into a single conversational flow with 6 structured question types (open_ended, scale, single_choice, multiple_choice, ranking, yes_no — see our structured questions guide).

1. Open-ended questions

Definition: Respondents answer in their own words — no predefined options.

Examples:

• "Walk me through the last time you tried to do X."
• "What is the single biggest frustration with your current workflow?"
• "If you could change one thing about [product], what would it be and why?"

When to use: Discovery, root-cause exploration, capturing unexpected language. Open-ended is the only question type that surfaces what you did not already think to ask about.

Pitfalls:

• Low response rate. In traditional surveys, only 20–40% of respondents answer optional open-ended questions, and most answers are 1–5 words.
• Hard to analyze at scale without dedicated qualitative coding.
• Easy to get vague hypotheticals ("I would love better dashboards") instead of behavioral evidence ("I rebuilt the same dashboard three times last quarter").

Modern approach: AI-moderated interviews — like Koji's voice and text interviews — fix the response-rate and depth problems by probing follow-up questions in real time. See how Koji's AI follow-up probing works and open-ended interview questions: 100+ examples.

2. Dichotomous (Yes/No) questions

Definition: Two mutually exclusive options — usually Yes/No, True/False, or Have/Have not.

Examples:

• "Have you used [product] in the last 30 days?" (Yes / No)
• "Are you the primary decision-maker for [category] software in your team?" (Yes / No)

When to use: Screening, eligibility, simple behavioral facts. Use dichotomous when there are genuinely only two states — don't force a binary on something nuanced (satisfaction is not Yes/No).

Pitfalls:

• Forces false dichotomies on continuous concepts ("Do you like the product?" — many users feel "kind of").
• Loses gradient information you can never recover.

3. Single-choice (multiple choice, one answer)

Definition: Respondent picks exactly one option from 3+ choices.

Examples:

• "Which of the following best describes your role?" → IC, Manager, Director, VP+, Founder
• "What is the primary reason you signed up?" → Specific list

When to use: Demographics, primary intent, mutually exclusive categorization.

Pitfalls:

• Order bias. Options at the top get picked more often (primacy effect). Randomize answer order when option meaning is symmetric.
• Missing options. Always include "Other (please specify)" and a "None of the above" or "I prefer not to say" exit.
• Overlapping options. "Engineer," "Software Engineer," and "Frontend Developer" all in one list = chaos.

4. Multi-select (multiple choice, multi-answer)

Definition: Respondent picks all options that apply.

Examples:

• "Which of these tools do you currently use? (Select all that apply)"
• "What features matter most to you? (Select up to 3)"

When to use: Capturing behavior or attributes where multiple states are simultaneously true.

Pitfalls:

• List length matters. Respondents tire by item 8 and skim the rest — keep multi-select lists under 10 items where possible.
• Single-select disguised as multi-select. If most people will logically pick one, use single-choice — multi-select adds noise.
• No cap = no signal. "Select all that apply" without a cap often results in 4–7 selections that are hard to prioritize. Adding "Select your top 3" forces clarity.

5. Likert scale questions

Definition: Statement + a balanced agreement scale, typically 5 or 7 points: Strongly Disagree → Strongly Agree.

Examples:

• "The onboarding process made it easy to find the features I needed."
  • Strongly Disagree / Disagree / Neutral / Agree / Strongly Agree

When to use: Attitudes, beliefs, satisfaction, perceptions. Likert is the workhorse of attitudinal research.

Pitfalls:

• Acquiescence bias. People drift toward "Agree" if the question is framed positively. Balance with reverse-coded items.
• 5 vs 7 points. 5-point is faster; 7-point captures more nuance for analytic teams. Pick one and stick to it across the survey.
• Neutral midpoint. Including a "Neither agree nor disagree" option respects the respondent but invites fence-sitting. Forced-choice (no neutral) increases polarization in data but irritates respondents.

See our Likert scale research guide for a full breakdown.

6. Rating scale questions (numeric, star, smiley)

Definition: Numeric or visual scale, typically 1–5 or 1–10.

Examples:

• "How likely are you to recommend us to a colleague?" (0–10)
• "Rate your overall satisfaction" (1–5 stars)
• "How easy was it to complete this task?" (1–7)

When to use: Sentiment intensity, performance, ease metrics like CSAT, CES, and SUS.

Pitfalls:

• End-aversion. Many respondents avoid the extremes (1 and 7), compressing the scale.
• Cultural variation. Respondents in different countries use scales differently — direct comparisons across geographies are risky without normalization.
• See Customer Effort Score and System Usability Scale for two standardized rating scales worth adopting.

7. Ranking questions

Definition: Respondent orders a list of items by preference, importance, or frequency.

Examples:

• "Rank these features from most to least important to your team."
• "Order these channels from most to least frequently used."

When to use: Forcing trade-offs. Unlike ratings — where everything can be "very important" — ranking forces relative priority.

Pitfalls:

• Cognitive load. Ranking 4 items is fine; ranking 10 is exhausting and produces noisy data. Cap at 5–7 items.
• Top-of-list bias. Respondents rank the first few items carefully and randomize the rest.
• No "tie." Force-ranked lists hide genuinely equivalent items. For high-stakes prioritization, supplement with rating scales.

For a deeper dive, see our choice and ranking questions guide.

8. Matrix / grid questions

Definition: Multiple related Likert or rating items in a grid, sharing the same response scale.

Examples:

• "Rate the following on a 1–5 scale: Onboarding, Pricing, Support, Documentation, Reliability"

When to use: Efficient bulk attribute rating, especially when items share a comparable scale.

Pitfalls:

• Straight-lining. Respondents pick the same column for all rows without reading — accounts for 15–30% of matrix data quality issues in large surveys.
• Mobile experience. Matrix grids break on small screens and inflate drop-off.
• Hidden survey length. A matrix of 10 rows × 5 columns is technically one question but feels like 10 — and behaves like 10 in fatigue analysis.

9. Semantic differential

Definition: A bipolar scale anchored by opposing adjectives.

Examples:

• "How would you describe our brand?"
  • Innovative ←→ Traditional
  • Friendly ←→ Cold
  • Expensive ←→ Affordable

When to use: Brand perception, emotional response, aesthetic ratings.

Pitfalls:

• Anchor choice matters. "Affordable" vs "Cheap" measure totally different concepts — choose anchors precisely.
• Cross-respondent comparability. What counts as "Innovative" varies across people. Best paired with open-ended follow-ups.

10. Slider questions

Definition: A continuous slider input, typically 0–100.

Examples:

• "Drag the slider to indicate the percentage of your week spent on manual reporting."

When to use: Continuous numeric estimates where the gradient matters.

Pitfalls:

• Default position bias. Sliders pre-set at 50 will be left at 50 by lazy respondents — randomize the start position.
• False precision. Sliders create the illusion of high precision on data that may be a rough guess.

11. Net Promoter Score (NPS)

Definition: "How likely are you to recommend us to a colleague?" on a 0–10 scale, segmented into Detractors (0–6), Passives (7–8), and Promoters (9–10).

When to use: Tracking loyalty and word-of-mouth intent over time. NPS is most useful as a trend within your own customer base, not as a cross-industry benchmark.

Pitfalls:

• The number is not the insight. A 35 NPS without follow-up "Why?" tells you nothing actionable.
• Cultural scale variation. US respondents use the top of the scale much more freely than European or Japanese respondents — comparing global NPS scores without controlling for this is misleading.

For the right way to use NPS, see our NPS survey guide and NPS follow-up interviews — pairing the score with a follow-up "Why?" is where the value comes from.

12. Demographic and firmographic questions

Definition: Categorical questions that segment respondents — age, gender, role, company size, industry, geography.

Best practices:

• Ask only what you will use. Every demographic question costs response rate. If you will not segment by it, do not ask.
• Put them at the end of consumer surveys (they feel intrusive upfront) but at the start of B2B surveys (so you can branch logic based on role/company size).
• Include "Prefer not to say" for sensitive demographics — required by privacy regulations in many jurisdictions.
• Use ranges, not free text for age, income, and company size to reduce drop-off and improve comparability.

13. Constant sum questions

Definition: Respondent allocates a fixed total (often 100 points or $100) across multiple options.

Examples:

• "Allocate 100 points across these 5 features based on how important each is to you."

When to use: Forced budget allocation — pricing research, feature prioritization, time allocation.

Pitfalls:

• High cognitive load — respondents drop off fast.
• Math errors — many surveys fail to validate that allocations sum to the target.
• Better suited to motivated, high-context respondents (e.g., customer panels, expert reviews) than cold outreach.

14. Image, video, and visual questions

Definition: Respondents react to an image, video, mockup, or design.

Examples:

• "Which of these landing pages feels more trustworthy?"
• "Watch this 30-second concept video and tell us what you think."

When to use: Concept testing, brand and creative validation, prototype testing.

Pitfalls:

• Stimulus quality matters — a low-fidelity sketch will be judged on the sketch, not the idea.
• Always ask "Why?" after a visual reaction question — the reason is the insight.

See our concept testing methodology and prototype testing concept validation for more.

The two-question taxonomy: open vs closed

Underneath all 14 types is a simple distinction:

• Open-ended questions are written, qualitative answers in the respondent's own words. They surface the unexpected but require manual or AI-assisted analysis.
• Closed-ended questions use predefined response options to produce structured, quantitative data — measurable, comparable, fast to analyze.

The best surveys mix both. As one comprehensive analysis from Kantar puts it, closed-ended questions offer measurable, comparable data — researchers can calculate percentages, averages, and trends, and cross-tabulate responses by demographics or behaviors to uncover meaningful patterns. Open-ended questions tell you why.

How AI-native research collapses the taxonomy

The 14-type taxonomy above is a legacy of paper and clipboard surveys. Modern AI-moderated interviews — like Koji — collapse the distinction by running an adaptive conversation that uses structured question types when they are the right tool and switches to open-ended probing when depth is needed.

Koji uses 6 structured question types that map cleanly onto the most-used legacy types:

For a deeper breakdown of each, see structured questions in AI interviews.

The difference from a static survey: Koji can ask a Likert question, see a low score, and automatically probe with an open-ended follow-up — collecting the why in the same conversation. Traditional surveys force you to pick a type up front and live with the limits.

Traditional survey vs Koji-style adaptive interview

For a side-by-side, see Koji vs SurveyMonkey and Koji vs Typeform.

A modern survey question template

Use this template when designing your next study. The order matters — it minimizes fatigue and drop-off.

1. Screener (Yes/No or single-choice): "Are you a [target user]?"
2. Behavioral anchor (open-ended): "Tell me about the last time you [did X]."
3. Closed quantification (scale or Likert): "How easy was that to do?"
4. Adaptive probe (open-ended): "What made it hard? OR What made it easy?"
5. Prioritization (ranking): "Rank these 4 improvements from most to least valuable."
6. Demographics (single-choice, at the end): Role, company size, etc.
7. Optional open-ended (open): "Anything else we should know?"

This pattern — anchor with behavior, quantify with a scale, probe with an open-ended, prioritize with ranking — is the spine of high-signal research. Koji automates this entire pattern with intelligent moderation.

Common mistakes across all question types

1. Double-barreled questions: "How satisfied are you with our pricing and support?" forces one answer to two things. Split them.
2. Leading questions: "How much do you love our new redesign?" assumes the answer. See avoiding bias in interviews and research bias guide.
3. Loaded language: "Should we continue our excellent customer service?" — biased adjective.
4. Asking about hypotheticals when behavior is available: "Would you use a feature that does X?" is far weaker than "Have you done X before, and how?"
5. Burying the headline: Putting your most important question on page 4, after fatigue has set in.
6. Asking what you cannot act on: If you cannot do anything with the answer, do not ask the question.

Related Resources

• Structured Questions in AI Interviews — the 6 Koji question types that replace 14 legacy types
• Survey Design Best Practices — end-to-end principles for high-quality surveys
• Likert Scale Research Guide — deep dive on the most-used rating type
• Open-Ended Interview Questions: 100+ Examples — the qualitative companion
• Choice and Ranking Questions Guide — capturing preference data at scale
• How to Write Great Interview Questions — applies to surveys too
• How to Analyze Open-Ended Survey Responses with AI — what to do with all that free-text data
• Surveys vs. Interviews — when to use each method