Run a video interview study

In a video interview, participants answer your questions on camera, one question at a time, and you review each answer as a video clip with its transcript. Everything else about the study, from the interview guide to going live, works as described in Create a study. Before you start

• The study must be a draft. Going live locks the interview type.
• Participants need a camera, a microphone, and a recent version of Chrome, Safari, or Firefox. Other browsers show a message asking them to switch.

​

Choose the interview type

The builder’s first question is “How should participants take the interview?”. You can also switch later, until the study goes live, with Interview type at the top of the study page.

• Video interview: participants answer each question on camera, and you watch each answer as a clip with its transcript. The interview opens on a welcome page instead of an outreach message, and shows no agent name, avatar, or AI notice.
• Text chat: participants type or send voice messages to a named agent, and you read each interview as a transcript.

In a video study, the interview opens directly on the first question of your guide.

​

Set up the welcome page

The Welcome page section holds the first screen participants see, before they turn on their camera. The builder drafts it when you choose video.

• Title and subtitle: in the subtitle, {{minutes}} shows the estimated interview length, which follows the study depth and the number of guide items. An empty field shows the placeholder text from the study page, in the participant’s language.
• Recording consent: off by default. When it is on, participants must tick “I agree to the recording and processing of my interview data for research purposes as described in the Privacy Notice” before they can continue. The link opens Pillow’s Privacy Notice, and the time of consent is saved with the interview.

When the study has more than one language, the welcome page also lets participants pick theirs.

​

Set skipping, read-aloud, and branding

The switches are in Advanced settings on the study page, under Study depth, and all are off by default. Formal address is there too (see Customize your agent).

• Question skipping: participants get a Skip this question button next to the record button. A skipped question shows as “Skipped this question” in the interview.
• Read questions aloud: a voice reads each question as it appears. Participants can turn it off with the Read aloud switch, and the voice stops as soon as they start recording. The voice speaks Arabic, Bulgarian, Chinese, Croatian, Czech, Danish, Dutch, English, Filipino, Finnish, French, German, Greek, Hindi, Hungarian, Indonesian, Italian, Japanese, Korean, Malay, Norwegian, Polish, Portuguese, Romanian, Russian, Slovak, Spanish, Swedish, Tamil, Turkish, Ukrainian, and Vietnamese. In any other language, participants read the questions on screen and don’t see the switch.
• Blind study: participants never see or hear who commissioned the study. The interview shows no logo or company name, the study link uses the neutral pillow.so address instead of your organization’s, and the agent doesn’t name your company. Keep the welcome page and closing generic. Questions and screener answers can still name brands and products as research subjects, including your own.

​

What participants do

1. Participants read your welcome page, tick the consent box if you require it, and click Continue.
2. They allow camera and microphone access on “Let’s check your setup” and click Start interview.
3. For each question, they record an answer and submit it. Recording starts only while their face is fully in frame and close enough to the camera; until then, the screen asks them to face the camera. Images, a video, or a Figma prototype attached to the guide item show with the question. For a prototype question, they click Start recording, then explore the prototype. Pillow records their camera, microphone, cursor movements, clicks, and prototype screen changes. They talk through their reactions, then submit their answer to finish. For a choice question, they pick from the options instead of recording. On phones, the question appears above the prototype in the same scrolling area.
4. After the last question, they see the agent’s closing message and that the interview is finished.

Each answer is checked when it is submitted. An answer that doesn’t address the question, such as a microphone test or unrelated talk, is turned away, and so is another answer of a second or two from a participant who keeps answering that briefly. The participant is invited to say more and records the answer again; the turned-away answer stays out of the interview. The third answer turned away ends the interview, and that interview is left out of the insights. If a prototype answer reaches the recording limit, recording pauses and a notice asks the participant to submit the answer. The recorded clip and interactions are kept. If an answer fails to send, participants choose Try again to resend the same recording or Cancel recording to discard it. If they reload the page mid-interview, they skip the welcome page and return to the setup check. If the study offers a reward, participants enter their identifier in a form at the end. See Offer a reward.

​

Preview a video study

In the builder, click Preview, then Open test link to take the interview in a new browser tab with your own camera. The preview restarts whenever you change the welcome page, skipping, or read-aloud settings. See Launch a study for going live.

​

Review video answers

Open a video interview from the study’s Interviews list. One player plays every answer clip back to back; click a question in the transcript to jump to its answer. When you translate the interview, the player’s subtitles show the translation, timed to the participant’s speech. Prototype answers show a reconstructed replay in the 16:9 player, with the camera recording in a square overlay. Playback keeps the same prototype loaded across follow-up answers and preloads upcoming prototypes. The entire recorded view fits inside the player by default, keeping the participant’s viewport proportions without cropping. In the playback controls, Fill prototype width zooms in and crops taller recordings vertically; Show full prototype restores the fitted view. Seeking the video restores recorded screens and component states, with a blue dot and pulses showing recorded pointer movements and clicks. These stay aligned in both views and when you resize the player. Scrolling, overlays, and animations may differ from the original session. The prototype must remain accessible, and edits in Figma can change what the replay shows. For everything else in the interview reader, see Review interviews.

​

Watch the highlights video

Once the study ends, its Summary opens on a highlights video. It walks through the study’s key findings, one chapter each: each chapter opens on its finding alone on screen, then short clips of different participants saying it play with captions, the finding staying on as the title. Answers in which a participant shared personal details are left out.

• Each clip names its interview, such as INT-42. Click it to open that interview at the answer.
• Click Download video in the player’s controls to save the video as an MP4, with the chapter titles and captions included. The first download is prepared on request, and the file saves once it’s ready.

The video follows the same findings as the written Summary, and is rebuilt when the Summary is. See Read your study’s insights. You can also play the participant clips embedded in the Summary to hear the words behind a finding. Click a clip to open its player, or its interview reference to read the transcript. Shared Summaries and PDFs present these clips as quotes.

​

Ask Pillow for video evidence

Ask Pillow can embed an important moment as a video extract, or play several participants’ extracts in sequence to support a finding. Ask it to show the moments worth watching, then use the player’s play button, captions, and timeline to review them. Shared answers, copied text, print, and Notion exports show the extracts as quotes. The interview references let you trace each quote to its source.

​

Ask for clips in ChatGPT or Claude

With Pillow MCP connected, ask your assistant to show video evidence for a finding. It can retrieve links to extracts from participants’ answers, or several clips that support the same point, using get_video_clips. Each result includes the quoted words, playback timestamps, an interview link, and a video link that expires after 24 hours. Ask the assistant for fresh links when they expire. Anyone with a video link can open the full answer recording until it expires; the timestamps select the quoted moment in players that support them. Several results are separate clips, not a combined video file. Clips need usable transcript timings and a quoted span of 2 to 30 seconds. Answers whose transcripts contain redacted personal data are excluded.

​

Related