The vocabulary of synthetic consumer research.
Plain definitions for the words Simulatte uses: what a simulated customer is, what an answer contains, where the real-world grounding comes from and how accuracy is checked. If a term shows up in a report and you want the precise meaning, it’s here.
The terms, plainly.
Every answer Simulatte returns traces back to who was asked, the evidence and the reasoning. These are the words that carry that chain.
- Synthetic consumer research
- Testing a business decision against simulated customers who behave like a real market, before that market sees it. It returns an answer with the reasons underneath, in minutes rather than the weeks a survey or focus group takes. Simulatte pairs it with interviews with real people, so the fast read can be checked. More in what synthetic consumer research is.
- Simulated customeralso: synthetic consumer, synthetic persona
- A detailed profile of one kind of person, modelled on real people: where they live, what they earn and spend, how they decide and who they ask. Simulated customers are not real people. Ask a few hundred and you get a spread of answers, not one average.
- Persona poolalso: persona population, audience
- The audience of simulated customers that a question is put to. Simulatte offers ready-made audiences in 12 countries, or builds one from a description of your market, and can find the groups inside it. See Simulated customers.
- AI-moderated interviewalso: AI interview, voice interview
- A voice interview with a real person, run by Simulatte’s AI interviewer. It follows your guide, asks follow-ups on what it hears, and can show your ad or product mid-interview. You send one link to your customers, your list or your team, and people take it when it suits them, in English, Hindi, Tamil, Arabic and Spanish. See AI interviews.
- Research agent
- The part of Simulatte that plans each study. It learns your products, past work and the files you share, sweeps the web for market, competitor and price context with every source shown, and remembers every study so you can ask across all of them.
- Plan and credit ceiling
- Before anything runs, the research agent shows a plan: who it will ask, how, and the most it can cost in credits. Nothing runs until you approve it. See Pricing.
- Decision test
- A single question put to simulated customers to get an answer: a price, a claim, a piece of creative, a name or a survey. Simulatte runs them as six methods: quick ask, survey, focus group, visual test, price test and message test.
- Verdictalso: the answer
- What a study returns: how people responded, the reasons underneath and the group that disagrees, rather than a score with no explanation. In a Simulatte report it comes first, followed by who we asked, the evidence, next steps and the method.
- Grounding
- How the real world gets into simulated customers. Simulatte builds them from census and market data, so traits go together the way they do in real people instead of being assigned at random. Grounding them in your own customer data or interviews, available on request, makes them answer like your customers rather than the average person.
- The fidelity ladder
- The idea that a simulated audience behaves more like a market the closer its inputs sit to real, specific people. Broad public data sits lower; what real buyers say sits higher. It is a working idea we are testing, not yet a measured result. It is also why the last step before a big call is to ask real people.
- Digital twin of a customer
- A simulated customer built to behave like a specific real segment or person, used to test how they would react to a decision. A digital twin is a simulation of a person, not the person.
- Calibration
- Checking a simulated population against the real world. Simulatte reports two things separately, because mixing them up is how research gets oversold: how closely the population’s makeup matches national surveys, and how well its behaviour predicts real outcomes in blind tests with the answers sealed first. See Accuracy.
- The segment that disagreesalso: the group that disagrees
- The part of the audience whose answer runs against the majority. Every Simulatte result shows it: not only what wins, but who it loses and why.
- Simulate, then check
- Asking simulated customers first, then putting the same questions to real people in an AI-moderated interview. Both sets of answers sit side by side, with the places they agree and differ called out. You can also run both at once.