Understand the neighborhood
Explore surrounding businesses and population patterns to understand the market a location serves.
Product sheet, full listing
compiled 2026-09 · rev. A
Product
A working model of the U.S. economy, one layer at a time.
Explore a location, understand the businesses around it and test an idea before committing to it. Follow the evidence behind the estimates, then examine how a scenario changes customer activity and operating results.
See the place behind the numbers
Move from a national view to a neighborhood, with businesses, buildings and local context on one globe. The U.S. is our current product focus; detail and source coverage vary elsewhere.
Read the physical setting
Use imagery, terrain and buildings to see how a proposed site fits its surroundings.
Inspect the business
Open a business card for its activities, customer focus and supporting evidence. See where information is reported or modeled.
Compare an idea
Describe a business and its operating assumptions, then examine modeled revenue, costs and customer activity over time.
Business cards that explain what a company does
Go beyond a broad industry label. Website-derived cards describe products, services, customer types and market reach, with quoted evidence you can inspect. If a card describes a brand rather than that specific location, it says so.
Reported company financials and modeled location estimates have different scopes. The card keeps those distinctions visible, alongside available filings and source details, so a corporate revenue figure is not mistaken for one storefront’s sales.
Find the businesses that fit
The separate Business Data product turns requests such as “insurance premium finance companies in the U.S.” into a reviewable list. Refine the matches, inspect the evidence and distinguish website-supported results from broader category matches. Review the price before purchasing a saved CSV or PDF export.
The AI analyst & Monte Carlo simulation
The analyst queries business records, local statistics, benchmarks and simulation tools. Describe the business, staffing and operating hours, then compare modeled revenue, costs, uncertainty and customer activity. Peer-cohort and coverage limits remain part of the result. Saved playback shows modeled customers by period.
Follow the customers in a saved scenario
Customer playback shows synthetic people and business buyers represented in the scenario’s saved purchase records. Select a day or month to inspect the activity for that period; those records are modeled transactions, not observed purchases by real individuals.
Expected demand, sampled estimates and posted synthetic purchases are labeled separately. Where a scenario includes a purchase ledger, its graph follows those saved buyer-to-business relationships. It covers that scenario, not every transaction in the economy.
Supplier, income and financing relationships are still being developed. Missing funding remains explicit; a revenue target does not create a customer with an invented budget.
Design principle, the synthetic population
The businesses are real. The people never are.
Business facts come from public commercial information, government filings, public registers, and what companies publish about themselves. Every resident, household and personal detail in the simulation is fully synthetic: statistically representative agents generated from aggregate census statistics, based on no real individual. Account information and saved research are covered in our data practices.
The calibration story
The difference between this and a confident guess is that we grade ourselves against filed reality, government records held out of the model's training, and publish the grade. Physically impossible answers are hunted deliberately: a modeled total revenue below a venue's own filed alcohol sales, or modeled employment below a carrier's filed driver count, is a caught contradiction, not a rounding error.
| test | published benchmark |
|---|---|
| held-out filings | 42% of employment estimates within 2× on 339,174 filings the model never saw, published as-is |
| earlier tract simulator | actuals fell inside its 10–90% band 86.9% of the time (target 80%), on 297 backtested sites. This does not validate individual customer switching. |
| falsification gates | contradiction tests compare modeled totals with filed Texas beverage revenue and federal carrier records |
The data backbone
More than 2,500 distinct source entries are recorded in the canonical manifests. They cover census statistics, labor data, business registers, public filings and commercial datasets. Each source retains its vintage, access conditions and license notes. U.S. coverage is the current product focus; coverage elsewhere varies.
- One bounded site simulation each month with a verified account. Try it free.
- Developing an integration? Explore API access.
- Questions about the model or the data? Contact us.