Where talking beats typing
Typing a long answer on a phone keyboard is slow, so people shorten it to «it was fine». Given a record button, the same person explains what happened at the checkout and how it made them feel. Product teams use it after a low rating, language schools use it for a short speaking sample, and customer research uses it for «walk me through the last time you…» prompts.
Use one voice question per survey, not five. Listening to answers takes you far longer than reading them.
Voice prompts that get a real answer
- In a minute or less, tell us what nearly stopped you from ordering.
- Describe the last time the app let you down. What were you trying to do?
- Read the sentence above aloud so we can assess your pronunciation.
Recording controls, time limit and file format
The only setting is the maximum length in seconds: 60 by default, 5 at the lowest. The recording stops by itself when time runs out.
The respondent gets record, pause, stop and re-record buttons, so a false start costs nothing. The browser asks for microphone permission first. Audio is saved as webm/opus or mp4 depending on the device and uploads with a progress bar.
Results count recordings and show an extension breakdown; you listen to each answer in the response view.
What you can set
- recording length limit, seconds
Making voice answers workable on your side
Plan the listening before you add the button. Two hundred one-minute answers are more than three hours of audio. Voice works best for small qualitative studies, for interviews you would otherwise run by phone, and for the most important open question in a larger survey.
Offer a way out. Some people are on a bus or in an open office. Keep the voice question optional, or put a long text question next to it for those who prefer to type.
Keep the prompt narrow. «Tell us anything» produces rambling; «What nearly stopped you from buying?» produces a story you can act on.