How accurate is a synthetic audience?

On iMario's public benchmark, synthetic audiences reproduce real survey distributions 89% of the time, measured across 11 populations, 922 questions and about 47,000 real respondents from Pew, CGSS and Stack Overflow. Two real samples of the same population agree about 93%, so this is 95% of what a human rerun achieves. With a recent real survey to reference, the score passes 90%. The method, datasets and raw files are published at imario.ai/synthetic-audience-benchmark.

What questions can iMario answer?

Anything you would put to a panel: which concept, message or price wins, why customers churn, how a segment reacts to an ad, a landing page or a new feature, and what objections a sales pitch will hit. You define the audience, ask through a survey, an interview, a focus group or a quick test, explore every answer and transcript, and get a report in under an hour. Most teams run synthetic first, pick the winner, then confirm it with a smaller real sample.

What makes iMario different from other synthetic audience tools?

Three things. First, every audience stands on a modelled world population, 5.4 billion people across 59 markets built from official statistics, not on a prompt, so distributions and minority views are real. Second, audiences are fitted to your own interviews and survey data and refresh as new market data arrives. Third, the accuracy is public: iMario publishes its benchmark method, datasets and raw files, while most vendors quote a number on data you cannot see. Each Synthetic Individual also holds its position, giving the same answer to 12 rephrasings 95% of the time.

Can I build a synthetic audience from my own customers or interviews, and is my data safe?

Yes. Upload interview transcripts, survey exports or CRM records and the audience is fitted to them, so it answers in your customers' own words and positions. Building an audience is free; you pay only for the questions you ask. Your data stays private to your workspace and is never used to train models for other customers.

What is "The Human Layer of AI"?

It is the idea that AI systems need real human perspective, not an average voice. iMario turns real people into reusable Synthetic Individuals that any workflow can call: an insights team surveys them, a product team tests with them, a sales team rehearses against them, and an AI agent can consult them through MCP or the API before it acts. Each individual keeps its own memory, so the person you asked about a concept this week refers back to it when you ask about price next month.

Which markets and languages does iMario cover?

59 markets and 5.4 billion people, each built from that country's official census and survey statistics, including the US, Canada, the UK and EU, China, Japan, Korea, India, Southeast Asia, Brazil, Australia and the Gulf. Ask your questions in the language you write in and respondents answer in it. Any narrower audience, from lapsed premium members in Texas to B2B IT buyers in Germany, is built on demand from the same foundation.

How much does it cost, and how do I start?

Sign up and you get 1,000 free credits with no card. One credit is one respondent answering one question, so that is a 100-person, 10-question survey. After that, credits are $0.05 each or $0.04 on a monthly pack, they never expire, and audiences are free to build. Full rates and worked examples are on the pricing page.

The Human Layer of AI

Reach anyone.Test anything.

From your own data to 5.4 billion people across 59 markets, iMario builds Synthetic Individuals that reveal how any audience thinks before you decide.

See the accuracy

Xenia Hendriks

36 · Product Manager · Germany

Xenia Hendriks

36 · Product Manager · Germany

Xenia Hendriks

36 · Product Manager · Germany

Xenia Hendriks

36 · Product Manager · Germany

Trusted by teams at

Understand humans. Before you decide.

Synthetic Individuals are built from your real people data, extended by a built-in simulated world population. You can ask them, test on, and rehearse with them. Every stage from first market read to after-sales.

Discover

See this use case

Simulated

Outcome

How It Works

Signal, not noise.

Layer 01
The world foundation
59 markets, 5.4 billion people, built from official statistics. Not an LLM prompt.
Layer 02
Your real people data
Bring interviews, surveys, and transcripts. Every audience answers in your customers' own voice.
Layer 03
Autonomous refresh
Autonomous agents train on new behavioral, macro, and policy data, and run real people interviews to reinforce the audience.

Platform

From your customers to billions of people worldwide. Ask them anything, anytime. Know before you decide.

0102030405/ 05

Define your audience

Bring interviews, surveys, and customer records, or start from any of 59 markets. iMario turns it into an audience you can return to anytime.

Ask your questions

Surveys, interviews, focus groups, creative tests, live web pages, even your own AI agents. Pick the format and the size. The whole audience answers.

Explore the answers

See the distribution, cross it with any question, and see who lands in each cell. Read their words, talk to one, or follow up.

Get the insights

Findings, charts, and quotes in one report. Every number links back to the answers behind it.

Validate with humans

Send the same questionnaire to real people and compare side by side. What they say goes back into the audience, so the next study starts closer.

Define your audience
Bring interviews, surveys, and customer records, or start from any of 59 markets. iMario turns it into an audience you can return to anytime.
Northwind /Audience/New audience12,480
Create a new audience
Model your audience from your own data, or build one from iMario’s population data.
Audience Name
Lapsed premium members
From your data From iMario
Core Audience Dataset
Upload Connect Filter sources
customer_interviews_q2.vtt18 MB2 min ago
48 transcripts
ProcessingVERBATIMS
member_survey_2026.csv1.2 MB1 min ago
2,310 respondents
ProcessingSURVEY
Salesforce · Contactsjust now
18,400 rows
SyncingDEMOS Connected
Describe New Segment
Create from
US Population
US Population
UK Population
Japan Population
Germany Population
Brazil Population
Middle-aged, high-income, men
Creates a segment bounded to what US Population actually covers.
Create audience
Ask your questions
Surveys, interviews, focus groups, creative tests, live web pages, even your own AI agents. Pick the format and the size. The whole audience answers.
Northwind /New task12,480
Hello, I’m iMario.
What do you want to find out?
Who should answer?
Lapsed premium members1,200 people
Premium members8,400 people
US PopulationFrom iMario
I want to understand how [audience] thinks about [topic]…
1,200 credits 1,200 respondents
Add files and photosUpload from computer
DefaultAsk your audience a simple question
SurveySeveral structured questions
InterviewGuided, in-depth conversation
Focus groupModerated live group discussion
Preference testCompare options, pick a winner
Creative testReactions to an ad or design
URL feedbackFeedback on a webpage or link
Agent testStress-test an assistantSoon
Attach a survey
New surveyMembership tier surveyReady · 6 questionsOnboarding drop-offReady · 9 questionsBrand tracker · Q2Ready · 12 questions
Explore the answers
See the distribution, cross it with any question, and see who lands in each cell. Read their words, talk to one, or follow up.
Northwind /Membership tier study12,480
Run the membership tier survey with lapsed premium members.
Survey · 6 questions
Objective: Should we launch a $6.99 membership tier?
Lapsed premium members · 1,200
02
Would you switch to the $6.99 tier?
1,187 responses
Switch to the $6.99 tier
41.0%
Stay on premium
33.0%
Not sure
17.0%
Cancel
9.0%
Summary
Two in five would move down to $6.99, and almost all of them are on premium today. The cheaper tier mostly re-prices existing members rather than adding new ones.
DetailsValidate with real peopleFeedback
Explorer
Selected audiences
Switch to the $6.99 tier
Break down byCurrent plan✕AgeRegion
Crosstab
swap axesremove
Would you switch to the $6.99 tier?
and
Current plan
Would you switch… Premium Basic Trial
Switch to $6.99 34.2% 4.8% 2.0%
Stay on premium 27.9% 3.6% 1.5%
Not sure 12.1% 3.1% 1.8%
Cancel 4.6% 3.0% 1.4%
Who they are
Agebreak down by this
18–34 · 17335–54 · 23455+ · 102
↑ This group · ↓ All18–34 173 · 34% 1.2×
509 selectedFollow-upAdd Segment
Get the insights
Findings, charts, and quotes in one report. Every number links back to the answers behind it.
Northwind /Membership tier study/Report12,480
Document DashboardMarkdownPDFPPT
Executive Summary
Background & Context
Research Objectives
Key Findings
Detailed Findings
Per-Question Analysis
Implications
Limitations
Respondents1,200 AI-simulatedQuestions6Answers7,122
01
Executive Summary
What you set out to decide
Should we launch a $6.99 membership tier?
The study’s answer
Not at $6.99. The tier mostly re-prices members you already have: 41% of sign-ups would be premium members trading down, and net revenue peaks at $8.99.
Top findings
1
At $6.99, two in five members would trade down rather than new members trading up.
RecommendationLaunch at $8.99: net revenue peaks there and trade-down falls to 14%.
2
Free shipping is the only premium benefit most members can name.
RecommendationKeep shipping on premium only; move early access to the cheaper tier.
Net revenue by price point, indexed
$5.9962
$6.9971
$8.99100
$9.9984
02
Background & Context
Premium grew 4% year on year, but cancellations at renewal have doubled since Q3. Finance asked whether a cheaper tier would hold the members who leave on price, or simply discount the ones who were going to stay.
What we already knew
Renewal churn
11.4%
up from 5.8%
Cited price
62%
of exit surveys
Premium ARPU
$11.90
blended
Exit surveys tell you who left and what they said on the way out. They cannot tell you what the members still paying would do if a cheaper option appeared next to the one they hold.
03
Research Objectives
1
Size the demand
How many lapsed members would take a $6.99 tier, and how firm is that intent?
2
Separate trade-down from new demand
Of those who would take it, who is already paying us more today?
3
Find the peak
Which price point maximises net revenue once trade-down is netted out?
How it was fielded
AudienceLapsed premium members · 1,200
InstrumentSurvey · 6 questions
Anchors187 real responses, fielded in parallel
Validate with humans
Send the same questionnaire to real people and compare side by side. What they say goes back into the audience, so the next study starts closer.
Northwind /Membership tier study12,480
Run the membership tier survey with lapsed premium members.
Survey · 6 questions
Objective: Should we launch a $6.99 membership tier?
Lapsed premium members · 1,200
02
Would you switch to the $6.99 tier?
1,187 responses
Switch to the $6.99 tier
41.0%
Stay on premium
33.0%
DetailsValidate with real peopleFeedback
Validate with real people
Field the same questionnaire to real humans; results feed calibration
1 Recruiting profilefull panel
WhoLapsed premium members×1,200
Target n200 completed responses
2 Questionnaire
Which plan are you on today?
Would you switch to the $6.99 tier?
3 Distribute
imario.ai/validate/8k3f Copy
0 / 200 responsesClose & save
4 Real vs synthetic
Would you switch to the $6.99 tier?
OptionSyntheticRealΔ
Switch to $6.9941%38%-3
Stay on premium33%35%+2
Not sure17%17%0
Cancel9%10%+1
n = 187 real responses

Would you switch…	Premium	Basic	Trial
Switch to $6.99	34.2%	4.8%	2.0%
Stay on premium	27.9%	3.6%	1.5%
Not sure	12.1%	3.1%	1.8%
Cancel	4.6%	3.0%	1.4%

Testimonials

“Going into Indonesia, what we did not have was a feel for how local men actually think about skincare. iMario got us that in time to change one strategy before the campaign went live.”

1 / 4

FAQ

Your questions,answered.

Reach out to our sales team for guidance on how to integrate iMario into your workflow.

A synthetic audience is a group of simulated people who answer questions the way a real, specific population would. Its basic unit is the Synthetic Individual: one simulated person with a demographic profile, attitudes and a memory of what they have said. iMario builds audiences from official population statistics for 59 markets, fits them to your own customer data, and lets you survey, interview or focus-group them as often as you need.A synthetic audience is a group of simulated people who answer questions the way a real, specific population would. Its basic unit is the Synthetic Individual: one simulated person with a demographic profile, attitudes and a memory of what they have said. iMario builds audiences from official population statistics for 59 markets, fits them to your own customer data, and lets you survey, interview or focus-group them as often as you need.

Reach anyone.Test anything.

Understand humans. Before you decide.

Discover

From your customers to billions of people worldwide.From your customers to billions of people worldwide. Ask them anything, anytime. Know before you decide.Ask them anything, anytime. Know before you decide.

Define your audience

Create a new audience

Ask your questions

Hello, I’m iMario.

Explore the answers

Explorer

Selected audiences

Crosstab

Who they are

Get the insights

Executive Summary

At $6.99, two in five members would trade down rather than new members trading up.

Free shipping is the only premium benefit most members can name.

Background & Context

Research Objectives

Size the demand

Separate trade-down from new demand

Find the peak

Validate with humans

Validate with real people

Built for rigor, not just speed.

Your questions,answered.

Know anyone, anytime.

From your customers to billions of people worldwide.From your customers to billions of people worldwide. Ask them anything, anytime. Know before you decide.Ask them anything, anytime. Know before you decide.

Define your audience

Create a new audience

Ask your questions

Hello, I’m iMario.

Explore the answers

Explorer

Selected audiences

Crosstab

Who they are

Get the insights

Executive Summary

At $6.99, two in five members would trade down rather than new members trading up.

Free shipping is the only premium benefit most members can name.

Background & Context

Research Objectives

Size the demand

Separate trade-down from new demand

Find the peak

Validate with humans

Validate with real people

Built for rigor, not just speed.

Your questions,answered.

From your customers to billions of people worldwide. Ask them anything, anytime. Know before you decide.

From your customers to billions of people worldwide. Ask them anything, anytime. Know before you decide.