Text Interview Experience

Text interviews offer participants a familiar chat-style interface where they type their responses and read the AI interviewer's questions in real time. But Koji's text mode goes far beyond a simple chatbot — it embeds interactive structured question widgets directly in the conversation flow, capturing quantifiable data without breaking the natural interview feel.

Starting a Text Interview

When a participant arrives at the interview landing page, they see two separate start buttons — Start Voice Chat and Start Text Chat. The participant clicks Start Text Chat to begin.

If your project is configured for text-only mode, the voice option does not appear and participants go directly to the chat.

The Chat Interface

The text interview looks and feels like a modern messaging app. Here is what participants see:

Layout

• Message bubbles display the conversation. The AI interviewer's messages appear on one side, and the participant's responses appear on the other.
• A text input field sits at the bottom of the screen where participants type their responses.
• A send button submits the message. Participants can also press Enter to send.
• A Done button in the header lets participants end the interview at any time.

Streaming Responses

When the AI interviewer replies, the text streams in word by word rather than appearing all at once. This creates a natural, conversational rhythm that feels more like chatting with a person than reading a pre-written block of text.

Participants can start reading the response as it appears and begin thinking about their answer before the interviewer finishes typing.

Structured Question Widgets

Unlike traditional surveys or basic chatbots, Koji embeds interactive question widgets directly in the conversation flow. When the AI interviewer reaches a question that calls for structured data, an interactive widget appears inline — the participant responds using the widget, and the AI then follows up conversationally to explore the reasoning behind their answer.

This hybrid approach captures the structured, quantifiable data that stakeholders need for decision-making, while the conversational AI provides the qualitative depth and context that surveys miss entirely.

Scale Widgets

When the interview calls for a numeric rating — such as an NPS score, satisfaction rating, or likelihood assessment — a scale widget appears inline. Participants interact with a slider or tap buttons to select their rating on a customizable scale (for example, 0-10 for NPS or 1-5 for satisfaction).

After the participant selects their rating, the AI interviewer automatically follows up to understand the reasoning behind the number. For example, if a participant rates their onboarding experience as a 4 out of 10, the interviewer might ask what specific aspects drove that low score and what would need to change to improve it.

Scale widgets support:

• Customizable minimum and maximum values
• Labeled endpoints (e.g., "Not at all likely" to "Extremely likely")
• Anchor probing — the AI references the specific number in its follow-up question

Choice Widgets

For questions with predefined answer options, choice widgets render directly in the chat:

• Single-select questions display radio buttons, letting the participant pick one option from the list
• Multi-select questions display checkboxes, letting the participant select all options that apply

After the participant makes their selection, the AI follows up to understand why they chose those options. An optional "Other" free-text field can be included for answers that do not fit the predefined choices.

Ranking Widgets

When you need to understand priorities, a ranking widget presents a drag-and-drop interface where participants reorder options by preference. This captures priority data that is difficult to extract from open-ended conversation alone.

After ranking, the AI interviewer typically asks about the reasoning behind the top and bottom choices, adding qualitative context to the quantitative ranking data.

Yes/No Widgets

For simple binary decisions, clean yes/no buttons appear inline. The participant taps one, and the AI interviewer follows up to explore the reasoning behind their choice. These are useful for screening questions, feature validation, or any situation where a clear binary answer is needed.

How Widgets Fit the Conversation

Structured widgets appear seamlessly within the chat flow. The AI interviewer asks a question conversationally, the widget appears, the participant responds using the widget, and then the conversation continues naturally. From the participant's perspective, it feels like a conversation with occasional interactive elements — not like filling out a form.

The data captured by these widgets is stored as structured responses alongside the full transcript. In your research report, widget responses are aggregated into charts, averages, and distributions — giving you both the numbers and the stories behind them.

For a complete guide on designing effective structured questions, see the Structured Questions Guide.

Conversation Flow

The text interview follows the same overall structure as a voice interview, combining open-ended conversation with structured data collection:

1. Opening greeting. The AI interviewer introduces itself and the topic, then asks the first question.
2. Questions and responses. The participant types their answer or interacts with a widget, and the interviewer responds with a follow-up question or a new topic.
3. Probing and clarification. If a response is brief or vague, the interviewer asks for more detail. If the participant shares something interesting, the interviewer explores it further.
4. Natural wrap-up. When enough topics have been covered, the interviewer thanks the participant and ends the conversation.

The AI interviewer adapts its questions based on what the participant writes, just as it does in voice mode. The research brief you configured guides the overall direction, but the specific follow-up questions are shaped by each participant's unique answers.

Typing at Your Own Pace

One advantage of text mode is that participants can take their time. There is no pressure to respond immediately — they can pause, think, and compose a thoughtful answer. This often leads to more considered and detailed responses, especially for complex topics.

The AI interviewer waits patiently. There are no timeouts during the conversation, so participants who step away briefly can pick up right where they left off.

Ending the Interview

Automatic Completion

The AI interviewer tracks coverage of your research topics. When it determines that sufficient ground has been covered, it naturally wraps up the conversation. The text interface also supports an automatic completion signal that triggers the transition to the completion flow.

Manual Completion

Participants can click the Done button in the header at any time to end the interview on their own terms.

How It Compares to Voice Mode

Both modes produce the same output for researchers — a transcript, quality score, and AI-generated insights. The key differences are in the participant experience:

Text mode is uniquely suited for research that requires structured data collection alongside qualitative depth. The structured question widgets are only available in text mode, making it the best choice when your study includes scale ratings, multiple-choice questions, or ranking exercises.

Accessibility and Mobile

Text mode works well on mobile devices. The chat interface adapts to smaller screens, and participants can type using their phone's keyboard. Structured question widgets are fully touch-friendly — scale sliders, choice buttons, and drag-and-drop ranking all work on mobile.

For participants using screen readers or other assistive technologies, the chat interface follows standard web accessibility patterns.

What Researchers See

As the study owner, completed text interviews appear in your project dashboard just like voice interviews. Each includes:

• The full conversation transcript
• Structured question responses with their numeric or categorical values
• A quality score reflecting the depth and relevance of the responses
• AI-generated insights, themes, and summaries

Structured widget responses are aggregated across all interviews in your project report, giving you charts and statistics alongside qualitative findings.

Tips for Encouraging Good Text Responses

When sharing your interview link and recommending text mode, consider these suggestions:

1. Set expectations. Let participants know the interview is a conversation, not a survey. Encourage them to write naturally.
2. Mention the interactive elements. Participants appreciate knowing that some questions will have quick interactive widgets, making the experience feel lighter than pure typing.
3. Reassure on privacy. If participants know their responses are analyzed by an AI and not read word-by-word by a human team, they may feel more comfortable being candid.

Next Steps

• Structured Questions Guide — design effective scale, choice, ranking, and yes/no questions
• Voice Interview Experience — how voice mode works
• Interview Landing Page — what participants see first
• Interview Completion Flow — what happens when the interview ends

Further reading on the blog

• AI-Moderated Interview Platforms Compared: Which One Actually Works? (2026) — Not all AI interview platforms deliver real qualitative depth. This guide compares the top AI-moderated interview platforms in 2026 — Koji,
• Best AI Customer Interview Tools in 2026: The Complete Buyer's Guide — AI has fundamentally changed how product teams conduct customer research. Here are the best AI customer interview tools in 2026 — ranked by
• Best Customer Churn Interview Tools (2026): The Top 8 Compared — A 2024 study found exit surveys match the real churn driver in only 31% of cases. The right interview tool fixes that. Here are the 8 best c

<!-- further-reading:blog -->