The first 3 interviews are free · Start free or see the prices

Voice or Text Interviews: Which Format to Choose for AI Research

Voice gives stories, tone and clips. Text gives more answers that are easy to scan and compare. Video with screen share lets people show a task. Choose by the answer you need.

Overview

The format changes the data, not only the experience. People who talk tell stories in the order that things happened, with hesitation and emotion. People who type write shorter, more edited answers. Both are useful for different questions.

User Evaluation has an AI interview, which is voice only by default or video with screen share. It also has a text survey, where people answer in writing and the AI asks follow-ups. In a voice interview, participants can also type an answer instead of speaking.

In a video study the AI moderator only hears the participant. It does not see the camera or the shared screen. It asks people to say what they see and do. The recording keeps the screen and the camera for you to watch later.

Rules

  1. 01

    Choose voice when you need the story behind a decision.

    Decisions such as a purchase, a switch or a cancellation have a sequence of events. People recall and tell a sequence more easily when they talk.

  2. 02

    Choose text when you need many short answers to compare.

    Text answers are faster to read and code. A text survey also costs less for each response, so you can ask more people the same few questions.

  3. 03

    Choose video with screen share when the participant must use something.

    You see what they do in the recording. Because the moderator cannot see the screen, write questions that ask the participant to describe their actions.

    Do not

    What do you think of this page?

    Do

    Tell me what you see on this page. What would you click first, and why?

  4. 04

    Write text questions shorter than voice questions.

    A participant reads a text question and can skip half of it. In voice, a longer question with a short example is easy to follow. In text, it becomes a block that people skim.

    Do not

    Thinking about the last few months, and the different tools you tried, which one did you end up keeping, and what made you decide on it over the others?

    Do

    Which tool did you keep after your trial? What made you choose it?

  5. 05

    Ask for one specific detail in text follow-ups.

    "Tell me more" in text gets one more short sentence. Ask for one concrete detail, such as a date or a step. It gets a better answer for the same effort.

    Do not

    Can you tell me more?

    Do

    What was the last step you did before you gave up?

  6. 06

    Check text answers for pasted and AI-written text.

    Text is easier to fake than voice. Watch for perfect structure, lists, generic phrasing and the same phrasing in different sessions.

  7. 07

    Let voice participants type when they need to.

    Some participants are in a shared space, have a speech difficulty or want to give a detail in writing. Tell them in the guide introduction that typing is fine.

  8. 08

    Think about what the report needs to show.

    Voice and video give clips that your team can watch. A clip from a text answer is a card with the quote. A short video of a customer often persuades a team more than a quote.

Checklist before launch

  • You wrote down whether you need stories, short comparable answers or observed tasks.
  • The format matches that need.
  • Video studies ask participants to describe what they see and do.
  • Text questions are short and ask one thing.
  • Text follow-ups ask for one specific detail.
  • You know how you will check text answers for pasted or AI-written text.
  • Voice studies tell participants that they can type an answer.

Questions

Not in general. Voice gives more detail and emotion. Text gives more answers at a lower cost that are faster to analyze. Choose by the question you must answer.

In User Evaluation, yes. Participants can type an answer in the interview room instead of speaking.

Not in User Evaluation. The recording keeps the camera and the shared screen, but the moderator only hears the participant. It asks them to describe what they see and do.

Voice is harder to fake than text, because a person must speak in real time. Text answers can be pasted from another tool.

Ask real people your question

The AI moderator interviews screened people by voice or text. Each claim in the study report links to its clip. The first 3 interviews are free.

Start free