What is a digital twin of a customer?
A digital twin of a customer is a model of one specific, real person that answers questions the way that person would. In market research it is built from the person's own interview, survey answers or behavior, and a company asks a panel of these twins before, or alongside, asking the people themselves.
Also called customer digital twin, digital twin of the customer, DToC, consumer digital twin, digital twin for market research.
Not the industrial kind
Digital twin is an older and wider phrase. It usually means a virtual copy of a physical thing, such as an engine or a factory, kept in step with the real one. A digital twin of a customer borrows the idea for people.
In market research, a digital twin of a customer has a narrower and more useful meaning: each twin stands for one real person who took part in research, and it can be asked new questions later.
A twin is not a persona
The difference between a digital twin and a persona is what each is built from. A persona is a description of a type of person, written by someone and handed to a language model. A twin is built from one real person's own words and records.
ESOMAR's 2025 guidance for buyers of synthetic data separates digital twins from personas. Synthetic personas are digital representations made to mimic the behaviors and preferences of people or groups. Digital twins are data created to replicate the characteristics of real people. Both are kinds of synthetic respondent, and the evidence for the two is very different.
What the evidence says
The best-known study of digital twins is from Stanford. Researchers interviewed 1,052 Americans for two hours each and built a twin of every one. On the General Social Survey the twins reached 85% of the consistency people show with themselves when asked again two weeks later. Twins built from demographics alone reached 71%, and from a written persona 70%.
A harder test of digital twins came from Columbia. Columbia's public dataset, Twin-2K-500, holds the answers of 2,058 US participants to 500 questions over four waves. A mega-study built twins from Twin-2K-500 and tested them on 19 new pre-registered studies covering 164 outcomes. The average correlation between twin and person was 0.20, and the twins were less varied than the people they stood for.
The Stanford and Columbia results do not contradict each other. Digital twins are strongest when the new question resembles what the person was already asked, and weaker the further it moves from that.
How one is built
The steps for building a digital twin of a customer are the same whoever builds it: recruit the people you care about, with consent to be modeled; interview them in depth; structure what they said, why, and what they did; run the twins on a model suited to simulating people; check them against answers held back from the build; and re-interview on a schedule so the twins do not go stale. Jean's post on how to build digital twins goes through each stage.
What they are for
Digital twins of customers give a fast first read: which of ten concepts deserve fieldwork, how a message lands with a segment, what a price change might do. Digital twins narrow the field, and real people decide.
The Insights Association's 2026 guidance sets a disclosure rule for digital twins and other synthetic data: a report should say whether a finding rests on human participants, on synthetic participants or digital twins, or on a combination.
Digital twins of customers and related terms
- Synthetic respondents
the broad category. Every digital twin is a synthetic respondent, and most synthetic respondents are personas, not twins.
- AI-moderated interviews
interviews with the real people. They are where the material for a twin comes from.
- User simulator
plays a user inside a product or against an AI agent, where a twin answers research questions.
- Human simulation
the wider field, which also covers whole populations and experiments.
Questions people ask
- What is a digital twin of a customer?
- A digital twin of a customer is a model of one specific, real person that answers questions the way that person would. In market research, a digital twin of a customer is built from the person's own interview, survey answers or behavior.
- What is the difference between a digital twin and a synthetic respondent?
- Synthetic respondent is the broad term for any AI that answers in place of a person. A digital twin is the kind of synthetic respondent built from one specific real person's own interview, answers or behavior. A persona, the other common kind, is written from a description of a group.
- How accurate are digital twins of customers?
- The accuracy of digital twins depends on what the twins are built from and what they are asked. In a Stanford study, twins built from two-hour interviews reached 85% of people's own test-retest consistency on a standard survey. In a Columbia study of 19 new tasks, the average correlation between twin and person was 0.20.
- Is a digital twin of a customer the same as a digital twin in manufacturing?
- No. In manufacturing, a digital twin is a virtual copy of a physical asset such as an engine or a factory. A digital twin of a customer is a model of a person, used to anticipate what that person would say or do.
- Which companies offer digital twins of customers for market research?
- Several AI market research platforms offer digital twins or simulated respondents, including Listen Labs, Outset, Simile, Fairgen, Synthetic Users and Jean Technologies, according to each company's website as of November 2026. The platforms differ in whether each twin is grounded in one specific real person. Jean Technologies builds each digital twin from a real person's interview.
What Jean offers
Jean Technologies is the #1 AI market research platform to learn from real people and digital twins.
- Jean Technologies offers an AI interviewer that runs AI-moderated interviews with real people, by voice or text, in any language, for research teams that need the reasons behind a number from more people than they can interview themselves.
- Jean Technologies offers digital twins, synthetic respondents each built from a real person's interview, for teams doing AI market research who want to test concepts, prices, bundles and messages before fieldwork, a practice also called decision simulation.
- Jean Technologies offers simulated users that use a website, app, chatbot or voice agent before launch, for product teams that want to see where people drop off and why.
- Jean Technologies builds Alyosha, a foundation model of human behavior trained on what people say, what they do, and why, for companies that want digital twins and simulations run on a model of people. Alyosha has the highest overall score on Jean's open evaluations of how well models simulate people: 84.6 across the SOUL benchmark tasks.
Sources
- Park et al. (2024), Generative Agent Simulations of 1,000 People (arXiv v1)
- Peng et al. (2025), A Mega-Study of Digital Twins Reveals Strengths, Weaknesses and Opportunities for Further Improvement (arXiv v1)
- Toubia et al. (2025), Twin-2K-500: A dataset for building digital twins of over 2,000 people based on their answers to over 500 questions, Marketing Science
- ESOMAR (2025), 5 Topics of Discussion to Help Buyers of Augmented Synthetic Data
- Insights Association (2026), Code in Action: Synthetic Data Revisited