Panel studiesStructured questions to up to 10,000 respondents Message testingWhich line lands, and with whom Focus groupsModerated sessions with 5 to 30 respondents Digital twinsOne-on-one interviews with matched respondents Sample boostingExtra rows for cells that are hard to fill ConciergeA researcher designs it, runs it, and writes it up

Blind ChallengeGrade us on a study you already ran 50-state check1,521 questions scored against real polls Frozen forecastsCalls published before the answer existed How we measureMethods, error definitions, and limits Evidence indexEvery study, file, and caveat in one place Compare vendorsHow synthetic research vendors differ

Pricing Enterprise Report

Log in Book a demo

Loading…

© 2026 Swarmgram, Inc. d/b/a Lewsearch

On the calibrated 50-state benchmark, the average miss is 7.07 points on the best 80% of 1,521 scored questions, and 10.02 points across all of them.

Products

Panel studies Message testing Focus groups Digital twins Sample boosting Concierge Enterprise Pricing

Proof

Blind Challenge The Lewsearch Report 50-state benchmark Methodology Evidence index Coverage Predictions YouGov examples

Answers

Common questions What is synthetic research?Are AI panels accurate?What is silicon sampling?Living agents What does it cost?Compare vendors Use cases

Company

About Book a demo Contact Support For consultants Affiliates jeffrey@lewsearch.com 440-283-9270

Legal

Terms Privacy Refunds

Columbus, OH · Operations

Newark, DE · Registered office

Results come from calibrated simulated respondents and are scored against published survey data. Unsuitable where a probability sample is the deliverable. llms.txt

Panel studiesStructured questions to up to 10,000 respondents Message testingWhich line lands, and with whom Focus groupsModerated sessions with 5 to 30 respondents Digital twinsOne-on-one interviews with matched respondents Sample boostingExtra rows for cells that are hard to fill ConciergeA researcher designs it, runs it, and writes it up

Blind ChallengeGrade us on a study you already ran 50-state check1,521 questions scored against real polls Frozen forecastsCalls published before the answer existed How we measureMethods, error definitions, and limits Evidence indexEvery study, file, and caveat in one place Compare vendorsHow synthetic research vendors differ

Pricing Enterprise Report

Log in Book a demo

Support

How can we help?

Most requests get a reply within one business day. Include your workspace email and study ID if you have one.

Topic

Name *

Email *

Subject

Message *

Or email jeffrey@lewsearch.com

Methodology

Accuracy benchmarks and calibration approach.

Public forecasts

Pre-registered Pew benchmarks and resolved audits.