Single Choice Questions in AI Interviews: When Radio-Style Questions Beat Free-Form

When you need to know which one option a customer chose — not all the options that applied, not their preference ordering, but the single thing they did or believe — you need a single_choice question. It is the structured-question equivalent of "pick one and only one," and it is one of the most common question patterns in customer research.

In Koji AI interviews, single_choice questions work like you'd expect from a radio-button widget in any survey tool, but with an important twist: after the participant makes their choice, the AI immediately probes for the reasoning. This guide covers Koji's single_choice question type — when to use it, how it differs from multiple_choice and yes_no, and how to design options that produce reliable, comparable data across every interview.

What Is a Single Choice Question in Koji?

In Koji's structured question system, single_choice is one of six question types — alongside open_ended, scale, multiple_choice, ranking, and yes_no. It presents the participant with a list of options and captures exactly one selection. The data type Koji stores is string — a single option label.

What makes single_choice in Koji different from a radio button on a survey form:

• AI probing on the chosen option: After the participant picks one, the AI generates a contextual follow-up: "You picked Option B — what specifically led you to pick that over the others?"
• Voice-mode conversational delivery: In voice mode, the AI reads the options conversationally and parses the participant's natural-language response into the correct option.
• Optional "Other" with thematic coding: When allowOther is enabled, free-form responses are automatically clustered for analysis.
• Confidence-graded extraction: When ambiguity exists (e.g., "I'd say maybe A or D"), the AI either probes for clarification or flags the answer as low-confidence in the report.

The Six Question Types: Where Single Choice Fits

Koji supports six structured question types, each chosen for a different research goal:

Single_choice is the right type when the options are mutually exclusive — only one can logically be true for any given participant. Use it whenever you'd reach for a radio-button widget in a traditional survey tool.

When to Use Single Choice Questions

1. Mutually Exclusive Demographics or Roles

"Which role best describes your current position?" with options like "Engineer," "Product Manager," "Designer," "Researcher," "Other" is a classic single-choice use case. Even if a participant's actual title spans multiple roles, the question forces a primary identification — useful for segmentation.

2. Primary Use Case Identification

"Which of the following best describes how you primarily use our product?" forces a single answer for analysis purposes. AI probing then captures the secondary use cases ("You picked A. Do you ever use it for B or C too?") without losing the clean primary segmentation data.

3. Pricing Tier and Plan Selection

"Which pricing tier are you on?" or "Which plan are you considering?" with mutually exclusive options is a textbook single_choice question. The AI follow-up captures the reasoning: "What made you go with the Pro tier over Starter?"

4. Decision Maker vs Influencer Roles

In B2B research, "Which best describes your role in this purchase decision?" with options like "Final decision maker," "Strong influencer," "Recommender," "Evaluator only" is essential for understanding buying-committee dynamics. AI probing then drills into the actual decision dynamics for richer context.

5. Win/Loss Outcome Classification

"Which best describes what happened with this deal?" with options like "We chose your product," "We chose a competitor," "We did nothing," "Still evaluating" is the foundation of any win/loss research program. Single_choice keeps the outcome data clean for aggregation.

6. Stage-of-Journey Identification

"Which stage best describes where you are right now?" with discrete journey stages lets you segment the rest of the interview by lifecycle position — and gives you frequency data on how participants are distributed across the journey.

How Single Choice Works in Text Mode

In Koji's text (chat) interview mode, a single_choice question renders as a list of radio-button options. The participant taps one, and that selection is captured immediately as the structured value.

What happens next:

1. The selected option (e.g., "Option B") is stored as the question's structured value the moment the participant taps.
2. The AI reads the selection in context of the study brief and the participant's earlier answers.
3. Based on probing configuration, the AI generates a follow-up: "You picked Option B. Walk me through what made you choose that over Option A or Option C."
4. The participant's open-text follow-up is captured as a qualitative answer linked to the structured selection.

This gives you the clean categorical data needed for charts plus the qualitative why behind the choice — captured in a single conversational moment rather than requiring a separate open-ended question.

How Single Choice Works in Voice Mode

In voice mode, there are no radio buttons — everything is conversational. The AI reads the question and the options conversationally: "There are four options I'll read out. Pick the one that best fits. The options are: A, B, C, or D."

The participant responds in natural speech: "B, definitely" or "I'd say A, the second one you mentioned" or "It's really between A and C, but if I had to pick one, A."

The AI parses the response into the correct option, handling natural-language responses that wouldn't fit a radio button form. If the response is genuinely ambiguous, the AI asks for clarification: "Just to confirm — you're going with Option A?" before moving on. This makes voice-mode single-choice questions feel like a natural moment in the conversation, not a survey interruption.

With tools like Koji, you don't have to abandon structured data collection to get conversational depth. The platform handles both at once.

The "Other" Option in Single Choice

For single-choice questions where you're not sure your options cover the answer space, enable allowOther and the participant sees an "Other" option that, when selected, prompts a short open-text response.

Why this matters in single_choice specifically: if a meaningful share of participants pick "Other" and describe the same concept, you've discovered a gap in your option list. Koji's analysis clusters those "Other" responses thematically and surfaces them in the report as recommended options to add to your next study.

For early, exploratory studies: include "Other." For tracking studies where comparability matters: exclude it.

AI Probing Behavior for Single Choice

The probing behavior is configurable per question:

• maxFollowUps: 0 (no probing), 1 (one follow-up, the default), or up to 3 (deep probing for critical questions).
• instructions: Custom guidance, like "Ask specifically about why they picked their option over the closest alternative" or "If they picked Option D, probe for the disqualifying factor with the other options."
• anchor: When true (the default), the AI's follow-up references the specific option they chose.

Good probing strategies for single_choice:

• Probe the rejected alternatives. "You picked B. What about A or C didn't work for you?" surfaces the decision boundary — often more interesting than the chosen option itself.
• Probe the strength of preference. "How close was it between B and your second choice?" reveals whether the selection is a strong preference or a marginal one.
• Probe the future consideration. "Is there anything that would have made you pick a different option?" surfaces the conditions under which preferences could shift.

Single Choice Answers in Your Research Report

In the Koji research report, single_choice questions are visualized as frequency bar charts — showing what percentage of participants selected each option, ordered from most to least selected.

Below the chart, the report includes:

• A qualitative synthesis from the AI's follow-up exchanges, organized by selected option
• The "Other" responses thematically clustered with verbatim quotes
• Cross-question segmentation — automatically slicing other questions in the study by single_choice option (e.g., "How did Engineers rate the product on a scale of 1-10 vs Designers?")
• Participant-level detail so you can click any option to see exactly which participants selected it

This segmentation capability is one of the most powerful uses of single_choice questions in Koji — because each single_choice question becomes a dimension along which you can slice the rest of your research data.

Writing Strong Single Choice Questions

Verify mutual exclusivity. If a thoughtful participant could honestly pick more than one option, you should be using multiple_choice, not single_choice. Workshop your option list with at least three team members to test for overlap.

Cover the answer space. If your options miss a common case, participants will either pick the closest-but-wrong option (distorting your data) or select "Other" (forcing manual analysis). Cast a wide enough net.

Use parallel construction. Each option should be grammatically and conceptually parallel. Mixing "Engineer," "Product team," and "I make decisions" creates inconsistent options that bias the answer.

Order options thoughtfully. Random ordering is often the safest choice to avoid order bias. For numeric-progression options ("Less than 1 year," "1-3 years," "3-5 years," "5+ years"), keep the natural sequence.

Keep the list short. Five to seven options is a comfortable maximum. Long lists in single_choice mode produce decision fatigue and randomized selection. If you need more options, consider grouping into categories.

Match the participant's mental model. If you're asking about a familiar concept, use the language your audience uses. If you're introducing a new concept, define it briefly in the question stem.

When Not to Use Single Choice

Single_choice is not always the right answer. Avoid it when:

• Multiple options are honestly true. Forcing participants to pick one when they'd pick three distorts your data. Use multiple_choice.
• The answer exists on a spectrum. "How important is this?" with options like "Very important," "Important," "Neutral," "Unimportant," "Very unimportant" is better as a scale question — easier to aggregate and visualize.
• You want to discover the option space. Start with open_ended to see what participants generate naturally, then build a single_choice question for your next wave.
• The answer is a binary. Use yes_no — it's cleaner and visualizes as a pie chart.

Single Choice vs Yes/No vs Scale: How to Decide

These three question types frequently get confused. The deciding factors:

• Yes/No — Use when there are exactly two options and they're binary. "Have you used this feature? Yes or no."
• Single Choice — Use when there are three or more mutually exclusive options. "Which feature did you use most? A, B, C, D, or E."
• Scale — Use when the options exist on a spectrum and you want to aggregate numerically. "How satisfied are you with this feature on a scale of 1-10?"

The difference matters because each type produces a different report visualization and supports different downstream analysis. Picking the right type at design time saves you analysis pain later.

Combining Single Choice with Other Question Types

Single_choice questions are most powerful when used as segmentation dimensions in a larger study. A typical structure:

1. Single_choice: "Which role best describes your position?" — for segmentation.
2. Scale: "On a scale of 1-10, how satisfied are you with your current process?" — measured per role.
3. Multiple_choice: "Which of these tools do you use? Select all that apply." — frequency per role.
4. Open-ended: "Walk me through your biggest workflow frustration." — qualitative depth per role.
5. Single_choice: "Which of these would be most impactful to improve?" — prioritization per role.

In the report, every chart is segmentable by the single_choice answers — so you can compare "what Engineers said" against "what Designers said" across all the structured and open-ended questions without re-running the analysis.

Single Choice in Koji vs Traditional Surveys

In a traditional survey tool, a radio-button question captures the selection and that's it. The participant clicks, the form advances, and you find out the reasoning weeks later when you read the open-text "Other comments?" responses at the end.

In Koji, the moment the participant picks their option, the AI is already probing for the reasoning. By the end of the interview, you have:

• A clean string value for the structured chart
• A qualitative paragraph explaining the choice
• Verbatim quotes captured for use in the report
• Cross-segmentation automatically applied to every other question

The same single_choice mechanic becomes a fundamentally more useful research tool — exactly the kind of step-change that AI-native research platforms enable. Platforms like Koji turn one-shot survey questions into living research artifacts.

Related Resources

• Structured Questions Guide: All 6 Koji Question Types
• Multiple Choice Questions in AI Interviews
• Open-Ended Questions in AI Interviews
• Scale Questions in AI Interviews
• Yes/No Questions in AI Interviews
• How Koji's AI Probing Works