Choose a place.
Search a US city or town, or any county, metro or state. Explore twins of the people who live there or the businesses that operate there, then narrow to the audience you care about.
Choose a place → Define an audience+A place. A population. Your question.
Test it ona place beforeyou decide.
Pick any US place and ask its population twins: simulated residents and businesses modeled on the real make-up of that place. See which option is ahead, with whom, and how sure the reading is, then interview the twins behind it.
12,759US places6,701 cities · 3,152 counties · 2,462 areas · 393 metros · 51 states/DC
Residents & businessestwo audiencesAsk one, the other, or both side by side
1,500twins per readingIn the app, by default
How it works
One conversation,from question to reading.
Every step happens in one thread. You see what will run before it runs, and nothing spends until you press Run.
Ask. Then change something.
Ask a yes/no or multiple-choice question. Compare two to five messages, or add a hypothetical event and re-ask the same twins.
Questions · Message tests · What-ifsFollow the differences.
See responses on the map and across groups. Dig into the ranges and the reasons behind them, then export results and breakdowns as CSV.
Map · Breakdowns · CSV exports01 · Ask
Type the decision in a sentence. Paste the message, the offer or the options you are weighing.
Place · Oakland, CAWhich of these taglines lands better for a new neighborhood bus line?
- AGet there sooner.
- BThe whole city within reach.
One sentence is enough. The options can go in the same line, separated by slashes.
Interviews
Ask why,in their own words.
Ask the twins first, then the people. Twin interviews are instant: open the twins behind a reading, ask them why, and ask a follow-up.
From a reading.
One click opens an interview with twins from the group you just read.
Up to five at a time.
Pick the twins you want to hear from. Nothing is spent until you press twice.
Follow-ups keep the thread.
Ask again and the twin answers with the conversation so far in mind.
Business decision-makers too.
On a businesses reading, the modeled decision-maker at a business in the panel answers. A business name is shown as a fact, never as the speaker.
Interview
Resident · early forties, owns a small house at the edge of the city, drives to a warehouse job
- You
Why would you skip the new bus line?
- Twin
It doesn’t go where I work. I’d need two transfers, and my shift starts before the first bus of the day.
- You
What would change your mind?
- Twin
A direct route out to the industrial park. Even one bus an hour would do, if it ran early enough.
Real people
Real people,when you need them.
Twins are fast and cheap to ask. When a decision needs real answers, we run interviews and short studies with the public and with credential-checked experts.
- By arrangement
The public
Interviews and short studies with residents of the places you care about. Answers come back as real-people results, separate from twin readings.
- By arrangement
Experts
Credential-checked before they answer. Tell us who you need to hear from.
- Business owners and managers
- Clinicians
- Lawyers
- Engineers
- People in local government
- Others on request
How it works
Scope it with us.
On a call, we agree who you need to hear from, what to ask, and whether it is interviews or a short study.
Who · What to ask · FormatWe recruit and interview.
We find the people, check each expert’s credentials, and run the interviews or the study, with consent.
Recruit · Check · InterviewYou get transcripts and results.
Everything comes back labelled as real people, and kept apart from any twin reading.
Transcripts · Results · LabelledIn practice
One idea.A different point of view.
Compare two to five messages against the same question. See which version is ahead in which group, and where it is too close to call.
Marketing & brand
Which version speaks to which audience?
Put two versions side by side and see which lands with whom.
Audience: all residents
Message comparison
- 01Ask a question
- 02Twins respond
- 03See which is ahead
All residentsWhich version is ahead
Reads asAll residents: B ahead of A
Range
Your own question. Your own audience. Ask a place
Explore a different outcome
Change the situation.See what shifts.
Add a hypothetical event, then ask the same twins again. Compare the responses with your baseline.
Hypothetical bulletin
The city adds buses every 10 minutes on major routes.
Scenario comparison
Would you support a small local tax to improve bus service?
Responses Support
BaselineSupport
The band is the range
No black box
See exactly whateach twin is asked.
Every twin reads the same question, grounded in who it is and where it lives. From a reading, you can open twins from the group you read and interview them.
What one twin is given
Who
34, rents a one-bedroom
Works in health care, commutes by bus
Lives in a dense, mixed-income neighborhood
Local news, as of the reading’s date
One dated block of local headlines, the same for every twin in the place
The question
Which of these taglines lands better for a new neighborhood bus line?
The choices
A · Get there sooner.
B · The whole city within reach.
Its reason, in one line
“Sooner is nice, but the bus already gets me to work. What I can’t reach is anything else.”
A reading is many of these, one per group of similar twins, each answered the same way and weighted back to the place. Nothing about you or your customers is put in front of a twin.
- Who the twin is, and where it lives
- What the local news said, as of the reading’s date
- Your question, word for word, and your options
- Its answer, and one line on why
Use cases
One question.Any decision that depends on a place.
Messages, products, prices, policies, business buyers, and what happens if something changes. Bring the question; the population is built for you.
Messages & positioning
“Which of these three taglines lands in Oakland?”
B ahead of A among renters; homeowners split
Product concepts
“Meal-kit box or pantry refill: which would a household buy?”
Concept 2 ahead with families; no separation among retirees
Pricing
“Which of four monthly prices would you still pay?”
Interest falls away above the middle tier
Policy & public affairs
“A peak-hour fee to drive downtown: support or oppose?”
Opposition runs further in outer districts than downtown
Business buyers
“Would a small restaurant try an after-hours answering service?”
Food service ahead of retail
What-if
“The same question, after a fare cut is announced.”
Support shifts after the announcement; check the interval first
Interviews
“Why would you skip the new bus line?”
One twin’s reasons, in its own words
Creative & ads
“Two posters for a transit campaign: which lands with whom?”
Poster B ahead with households without a car
Features & flows
“Sign up with a phone number, or with an email address?”
Email ahead with older households; too close among younger ones
Any question. Any US place.
From a tagline to a ballot measure, a storefront idea to a price. If it depends on who lives or does business somewhere, ask it.
Ask a place- Find the message that moves a place.
- Pick the concept worth building.
- Find where a price stops being interesting.
- Read a policy before it is announced.
- Know which local businesses would try it.
- See how a positioning reads with each group.
- Play out a change before it happens.
- See how one announcement lands with every group.
- Ask why, in their own words.
- Talk to the people who know.
- Test creative and ads before you spend.
- Test features, flows and copy.
For teams building AI
Populations you canreach from code.
What your team can use today: from code, in the app, or with us.
- Keys on request
An API
Polls, backtests and message tests run as jobs you collect; interviews answer directly. All of it is described by an OpenAPI spec. Keys are issued on request.
- Keys on request
A read-only connector for AI assistants
Look up places and collect the results of jobs you submitted through the API, from an AI assistant. The connector cannot run or pay for a reading.
- In the app
Check readings against known results
Ask a past question and compare the reading with a result you already know, and see where it diverged. A public result may already be known to the model, so the check flags that.
- By arrangement
Human data from experts and the public
Judgments, rankings and interviews from the public and credential-checked experts, collected with consent and delivered as data, labelled as real people and kept separate from twin readings.
- By arrangement
Custom work, scoped with you
Places, question sets and runs built around what your team is working on. Tell us what you are building.
Quality
A number alone is not a reading.You get the interval, and what is missing.
Grounded twins.
Every twin is drawn from the real make-up of its place and weighted back to it.
Answers or nothing.
A reply that does not match the twin it was asked of is thrown out, never guessed onto someone else.
One re-ask, then disclosure.
A broken batch is asked again once. Whatever stays unanswered is counted and shown on the result, in so many words.
Intervals priced on what is real.
Twins are not independent people, so the interval is priced on how much independent information the reading really has.
A grade on every result.
Incomplete, Directional only, Comparable, Tight read. Each one says what it means for the decision.
Past questions, checked.
Ask a past question and compare the reading with a result you already know. A public result may already be known to the model, so the check flags that; a private study of your own avoids it.
Failures are loud. If part of the population did not answer, the result says so.
What not to ask it
- What one named person will do.
- A twin resembles nobody; the finest thing a reading reports on is a group.
- What someone did, or when.
- A twin has no personal history to recall, so a recalled answer would be invented.
- Whether a brand name is known here.
- The model cannot tell known in this place from known to the model.
- Anything that must stand as a survey.
- Filings, published toplines, anything where the method is the deliverable — for that, ask us about a short study with real people (#people).
- A cut too small to read.
- It is shown, greyed, with its authority removed, never hidden and never dressed up.
Method
How a reading is built.
- Place
- Twins
- Lean & turnout
- Cells
- Weighted estimate
- Interval
How the population is built.
Residents and businesses for each place are modeled on its real make-up: age, household, work, housing, industry and size. There is no panel to recruit.
How a question is read.
The population is grouped into cells that matter for the answer. Reads go where answers vary most, and each group’s answer is weighted back to the place.
How sure it is.
Every number carries an interval and a grade. You can score any past question against a result you already know, and see where the reading diverged.
What it is not.
A simulation, not a survey. Directional: it shows orders and gaps, not absolute levels. US places only.
Where it is weakest
- Small groups. Effective sample shrinks faster than the twin count suggests, so a thin cut gets a wide interval.
- Recalled behaviour and brand awareness. It will answer, but the answer is not grounded in the place, so don’t ask it.
- Businesses, which have a shorter track record than residents.
- Levels. Read which option is ahead and by roughly how much, not the exact share.
Keep looking
The topline isjust the beginning.
San Francisco, CA37.77° N / 122.42° W
A twin01 / 03
- Age
- 34
- Education
- Bachelor’s degree
- Housing
- Renter
Would more frequent buses make you more likely to ride?
“A shorter wait matters more to me than a new route. I already know how to get there.”
Interview a twin
Open a resident, or a business’s modeled decision-maker, and ask why. The answer draws on its household, work and neighborhood.
Cut by group or neighborhood
Break a result down without asking again: the cut reads answers you already have.
Compare places
Ask the same question in two places and see where they part ways.
Track a question
Ask it again when you choose, with the same wording, and see what moved.
Change the situation
Add a what-if, a date or a piece of news, then re-read the same twins.
Take it with you
CSV, a printable report and a share link for the people who were not in the room.
Bring us the decision on your desk.
Tell us the place and the question, or book a call and we will run it with you.