Latest US poll: 63% say gas prices are affecting their household budget a great deal or a moderate amount
See the pollInterviews
Ask a question on video and people answer on camera, in their own words, as part of a normal survey. An AI interviewer can ask up to three follow-up questions, and every answer is transcribed, checked and grouped into themes.
“Our vet suggested something gentler for his stomach.”
“She kept scratching, so we tried a grain-free one.”
“I read the label. Corn was the first ingredient.”
“A friend with two labs swore by it.”
“Honestly, it was the price. The old bag went up again.”
“Our usual shop stopped selling the old one.”
Why it matters
A typed answer in an open box is often one short line. A video interview used to mean a recruiter, a calendar and a handful of people. Video questions sit inside your survey instead, so you hear many people explain their answers on camera, next to the numbers from the rest of the questionnaire.
Step by step
Add a video question in the CatalyseAI builder. Name the topic the answer must cover, show a picture above the recorder if it helps, and set how long answers can be.
Choose zero to three follow-ups and write a short note on what to dig into. After each answer, the interviewer asks about what the person just said.
A quick camera check confirms a live person is in front of the camera, in enough light. Each recording is then checked for length, words and topic. If it fails, the person sees why and records again.
Each answer arrives with its transcript and its follow-up thread. Themes, the word cloud and the AI analyst all work from the transcripts.
Example
A US pet food brand wants to know how dog owners choose a new food. The survey screens for adults who changed their dog's food in the past six months and asks how long they have had a dog. Then one video question: “Think about the last time you changed your dog's food. What made you switch, and how did you pick the new one?” Answers can run to 90 seconds, and the AI interviewer may ask two follow-ups, briefed to “find out what they compared and who they trusted.” The team reads the themes, plays the answers behind the biggest one, and asks the analyst whether first-time owners give different reasons from long-time owners.
The story below follows this study, one answer at a time.
US adults who changed their dog’s food in the past six months
How long they have had a dog
“Think about the last time you changed your dog’s food. What made you switch, and how did you pick the new one?”
Answers up to 90 seconds
Two follow-ups, briefed to “find out what they compared and who they trusted”
Read the themes, play the answers behind the biggest one, and ask whether first-time owners give different reasons
The video semantic analysis
Following one answer: Dana, 38 · DenverQ3 · Think about the last time you changed your dog’s food. What made you switch, and how did you pick the new one?
Step 01 / 09
People answer on camera inside the survey. Before recording starts, a quick camera check confirms a live person is in front of the camera, in enough light.
Camera check, before recording starts
Step 02 / 09
Each recording is checked for length, words and topic. If it fails, the person sees why and records again.
Attempt 1 Sent back
Reason: Too darkWe can’t see you clearly. Find a brighter spot and record again.
Failed attempts never count as answers. The report shows how many there were, and why.
Attempt 2 Accepted
Step 03 / 09
Every accepted recording is transcribed, follow-ups included, so you can search every answer for any word.
Max started getting an upset stomach every few days, so our vet suggested we try something gentler.
I looked at a few brands online and compared the ingredients, and honestly the price mattered too.
We went with one that has real meat first and fewer fillers.
A friend swears by it, and he’s been so much better since.
Step 04 / 09
Up to three follow-ups, each built from what the person just said and guided by your brief. Skipping one ends the follow-ups for that question.
“Find out what they compared and who they trusted.”
Follow-upsMain answer · 0:46 · accepted
AI follow-up 1You said you compared the ingredients. What were you looking for on the label?
Reply · 0:21 “Real meat as the first ingredient, and no corn or wheat. The old food listed corn first.”
AI follow-up 2Whose advice counted most when you picked the new food?
Reply · 0:17 “The vet, for sure. The reviews just helped me narrow it down to two.”
Step 05 / 09
AI insights reads the transcripts and groups what people said by meaning, so different words for the same point land in the same theme.
Max started getting an upset stomach every few days, so our vet suggested we try something gentler. I looked at a few brands online and compared the ingredients, and honestly the price mattered too. We went with one that has real meat first and fewer fillers. A friend swears by it, and he’s been so much better since.
Vet advice
vet suggested+ 71 more answers
Stomach and skin problems
upset stomach+ 60 more answers
Ingredients on the label
compared the ingredientsreal meat first and fewer fillers+ 50 more answers
Price and value
price+ 37 more answers
Reviews and friends
friend swears by it+ 24 more answers
Step 06 / 09
Each sentence is scored from the words people use. This answer starts worried, ends relieved. Across the study, every theme carries a sentiment bar.
Step 07 / 09
Themes are sized by how many people raised them, with the tone of each. Click a theme to read and play the answers behind it.
212 accepted video answers · share of answers that raise each theme
PositiveNeutralNegative
Step 08 / 09
Filter recordings by question, theme or a word in the transcript, then play them one after another.
Dana, 382 follow-ups
Owen, 471 follow-up
Keisha, 352 follow-ups
Luis, 580 follow-ups
Grace, 262 follow-ups
Tom, 411 follow-up
Step 09 / 09
The AI report writes up the study: summary, sections, quotes and recommendations, downloaded as an HTML file. Ask the study answers what comes next, and lists any number or quote it can’t trace back to your data.
Most owners switched after a health problem, and the vet’s advice decided the new food. They check the label for real meat first and no corn; price matters, but rarely decides on its own.
“The vet, for sure. The reviews just helped me narrow it down to two.”Dana, 38 · follow-up 2 · 0:17
Do first-time owners give different reasons from long-time owners?
Yes, in two places. First-time owners lean more on reviews and friends; long-time owners lean on the vet.
Based on 64 first-time and 148 long-time owners.
Every number and quote traced to your data
What the analysis gives you
All of it sits in the report next to the rest of your survey, and all of it works from the transcripts.
AI insights · Themes
PositiveNeutralNegative
Voices
“The vet, for sure. The reviews just helped me narrow it down to two.”
Dana, 38 · Follow-up 2 0:17
“She kept scratching, so we tried a grain-free one. It stopped within a month.”
Leah, 52 · Main answer 0:57
“I’d pay a bit more if I knew it was actually better for him.”
Marcus, 44 · Follow-up 1 0:24
Words and phrases
Share of people. Each person counts once, even if they said it again in a follow-up.
Audience groups
Would buy the new food again
Switched within a week of deciding
In the report
Everything you take away
Keep scrolling: the reel runs sideways.
Watch each answer in the report, with the follow-up questions and replies underneath.
Every accepted recording is transcribed, follow-ups included. Search them for any word.
AI insights groups the answers to each video question into themes, with a sentiment bar. Click a theme to see the answers behind it.
Filter recordings by question, theme or a word in the transcript, then play them one after another.
Compare up to five audience groups, or make a group of everyone whose transcript mentions a word.
Ask the AI analyst a question in plain words. It reads the transcripts alongside the rest of your data.
The raw CSV holds the file link, transcript, relevance score, length and follow-ups for every answer. The AI-written report downloads as an HTML file.
Inside any study
A video question is one more question in a CatalyseAI survey, so it fits wherever you need the why: next to a concept score, a pack choice or an ad rating.
How likely are you to buy this oat drink?
4 · Probably would buy
What made you give it that score?
Camera check passed · accepted
“It sounds healthier than what I buy now, and I like that it’s made from oats. I’d want to taste it before I switch.”
AI follow-up 1What would you want to find out when you taste it?
Reply · 0:12 “Whether it’s too sweet. Most oat drinks are too sweet for my coffee.”
The score and the reason, from the same person, in one report.
FAQ
Up to three per video question, and you set the number. Each one comes from the answer the person just gave, guided by your note on what to dig into. People can skip a follow-up, which ends the follow-ups for that question.
It confirms that a live person is in front of the camera and that the light is good enough, before recording starts. You can switch it off for a question, for example when you want people to show you something in their home rather than talk to the camera.
You set it per question, from 1 second up to 10 minutes. If you leave it, answers must be at least 3 seconds and at most 90. Shorter prompts usually get more natural answers.
The person sees a short reason and records again. Failed attempts never count as answers, but the report shows how many there were and why they failed.
Recordings are collected for research. Using a clip in an ad or on social media needs the person's separate consent, so tell us before fieldwork if you might want that.
Related
Tell us what you want to hear from people. We'll show you how the question, the follow-ups and the report would look for your study.
Nearest office: Orlando, FL (corporate office)