Explainer

What are synthetic respondents?

Synthetic respondents are AI stand-ins for real survey participants. Here's what they are, how a credible one is built, and the limits that keep the results honest.

Synthetic respondents

Synthetic respondents are AI-generated stand-ins for real survey participants: language models conditioned to react like a specific slice of a market. Each one reads a concept and responds in character, giving a free-text reaction that can be aggregated into a purchase-intent distribution across a panel.

At Reverb we call them Echoes, to keep the honesty clear: an Echo reflects your market, it isn't a real person. A room of Echoes gives you a directional read in minutes, a way to hear enthusiasts, skeptics, and value-seekers react before you spend on fieldwork.

How a credible synthetic respondent is built

The difference between a useful synthetic respondent and a party trick is in how it's briefed and how its answer is measured:

  • Conditioned on behaviour, not demographics. A good Echo is defined by category usage and buying behaviour (how someone actually shops the category) rather than a demographic label. Behaviour is what predicts a purchase decision.
  • Reacts in words, not numbers. Ask a model for a 1–5 rating and it clusters on the middle. Ask what the respondent would actually say and you get a genuine reaction, with drivers and objections you can read.
  • Scored by meaning. Each reaction is mapped to a purchase-intent distribution using semantic similarity to calibrated anchor statements, a method validated in peer-reviewed research (Maier et al., 2025, arXiv:2510.08338).

How synthetic respondents are validated

A synthetic respondent is only as trustworthy as the evidence behind it. The credible approach is a calibration loop: every time you validate a concept with real, verified humans, you measure how close the synthetic prediction was and feed that back in. Over time the synthetic layer gets measurably sharper in your categories, and you should be able to see that calibration score on every report, not take it on faith.

The honest limits

Synthetic respondents are a directional screen, not ground truth. They're weaker in unfamiliar categories and non-Western markets, and their answers should never be cut by gender or ethnicity as if they were statistical signal. Every serious synthetic study should ship with a plain-English methods-and-limitations note.

How teams use synthetic respondents

  • Screen many concepts or claims at once and rank them before committing to any.
  • Sweep a price ladder to see where intent falls off.
  • Pressure-test messaging: which claim lands, and why.
  • Interrogate a reaction: ask a follow-up and rerun at near-zero cost.

Then validate the calls that matter with real people. Synthetic respondents make research something you do continuously, not once a quarter. See what is synthetic market research for the bigger picture.

Frequently asked questions

What are synthetic respondents?

Synthetic respondents are AI-generated stand-ins for real survey participants. Each is conditioned on a market segment and reacts to a concept in words, which are aggregated into a purchase-intent distribution, giving a fast, directional read that teams validate with real people.

Are synthetic respondents accurate?

For screening, they can be: a peer-reviewed elicitation-and-similarity method recovered about 90% of human test–retest reliability on purchase intent. Accuracy depends on category and market, so they're used as a directional screen, not ground truth.

How are synthetic respondents different from personas?

A persona is a marketing sketch of a customer. A synthetic respondent is an active participant that reads a concept and reacts, producing measurable output. Reverb conditions respondents on category behaviour rather than demographic persona detail.

When should you use real respondents instead?

Whenever you need ground-truth numbers to bet on: a launch decision, a pricing commitment, or an unfamiliar category. The honest pattern is to screen with synthetic respondents, then validate the survivors with verified humans.

See what your market thinks in the next five minutes.

Run your first Echo Pulse free. No fieldwork, no waiting, no black box.

Start a free Pulse
Keep reading

What is synthetic market research?

Read

Synthetic personas in market research

Read