Twins Studio
Interview once. Keep asking.
Every interview you run on Dipio leaves a twin of that person: a living model of your user. When the next question comes up, you ask them again, without booking another call.
How twins work
The interview ends. The questions don’t.
A new feature idea, a price to try, a step where people stall: the next question never waits for the next study. With twins, you put it to the people you’ve already heard from.
Ask
now
What’s a twin?
Every interview you run on Dipio leaves a twin of that person: a living model of them, built from what they said and what they did in your product, read through behavioural science.
The follow-up you forgot to ask on the call? Put it to their twin.
Without twins, every question waits for the next study.
The best follow-up always comes after the call. A new idea, a price to try, a step where people stall: today each one means another round of recruiting and scheduling.
With twins, you ask again.
Put the next question to the people you’ve already heard from, without booking another call. When new kinds of users arrive, interview them too.
Interview
Interview once.
Dipio interviews your real users and reads what they say against what your product analytics shows they do.
Twin
Their twin takes shape.
Every interview leaves a twin of that person, a living model built from what they said and what they did. Twins group into the same segments as your real users, so you can put a question to a whole segment at once.
Ask
Ask again.
Put follow-up questions to their twins whenever they come up. What they answer points to who to interview next, so the loop keeps learning.
Most synthetic users are made up.
A demographic, a prompt, a guess.
Yours are built from your real users.
What they told you in the interview. What they did in your product.
A living model of each of them.
- Cheaper than running the study with real users
- 144–381x
- Cheaper than running the study with real users
- Match with real people, across Dipio’s twin experiments
- 91%
- Match with real people, across Dipio’s twin experiments
Cost: Dipio’s internal estimate, depending on the model a twin runs on. Match: scored against how well one group of real people predicts another (Dipio twin experiments, 7 studies).
Our first paper, in press at Behavioural Public Policy, compared twins against 373 real people: in aggregate, their answers tracked closely. It also found the limit. Matching what people say is not yet matching how they think, and that gap is the work we’re on now.
Who asks, and who answers
A researcher interviews your users, Dipio interviews your users, or Dipio asks their twins. Twins exist only after a real study, so they answer the follow-up questions, not the first ones.
Qualified interviews
Dipio’s time is the median of 26 studies that reached 20 qualified interviews.
Researcher asks users
Dipio asks users
Dipio asks twins
Dipio asks users
3.6x
Faster than a researcher
40x
Less researcher time
100%
Match with real people
Dipio asks twins
~2,900x
Faster than a researcher
120x
Less researcher time
91%
Match with real peopleDemographic personas: 63%
Faster than a researcher
Less researcher time
Match with real people
Dipio asks users
3.6x
40x
100%
Dipio asks twins
~2,900x
120x
91%
Demographic personas: 63%
A researcher asking users assumes one person: 5 days of planning, recruiting that grows with the study, about 4 interviews per working day, and about 3 hours of analysis per interview. Dipio asking users uses the real median time from publish across 79 Dipio studies (1,828 qualified interviews). Dipio asking twins assumes about 15 minutes for a follow-up question, after a Dipio study. Matching is scored against how well one group of real people predicts another (Dipio twin experiments, 7 studies). Multipliers are against a researcher asking users.
Twins cover the users you’ve met.
Each twin is a model of someone you interviewed, as they were when you talked. Keep asking them follow-up questions, including about what you build next.
As your product and market move, you attract new users with new motivations, and segments shift. Nobody has interviewed them yet, so there are no twins of them. That’s when you run new interviews, and every new user you talk to leaves a twin too.
Be first to ask your users again.
Twins of the users you interview are in early access. Tell us the first question you’d put to them.
Join early access