Transparency

Every figure names its source

A plan whose numbers cannot be checked is worth less than one with fewer numbers. Here is every dataset we use, which vintage, what it is used for, and — the part most sites leave out — what it does not tell you.

County Business Patterns

3,144 counties

U.S. Census Bureau · 2023

What it gives
Number of establishments, employment and annual payroll by NAICS code and county.
How your plan uses it
The competitive picture in your Market Analysis, and the average size of an operator in your county.
What it does not tell you
Census suppresses any cell small enough to identify an individual business. Where that happens the plan says the figure is not published rather than estimating one.
Go to the source ↗

Population Estimates Program

3,144 counties

U.S. Census Bureau · 2024

What it gives
County population as of 1 July.
How your plan uses it
Market sizing and the establishments-per-10,000-residents comparison against your state.
What it does not tell you
Estimates between decennial censuses; revised annually.
Go to the source ↗

Small Area Income and Poverty Estimates (SAIPE)

3,144 counties

U.S. Census Bureau · 2023

What it gives
Median household income by county.
How your plan uses it
Customer purchasing power in your Market Analysis.
What it does not tell you
Model-based estimates with confidence intervals; a county-level median hides a lot of local variation.
Go to the source ↗

FOIA 7(a) and 504 loan data

505,888 loans indexed · 1,192 industry codes

U.S. Small Business Administration · Fiscal years 2023, 2024, 2025, 2026, as published 2026-06-30

What it gives
Every approved 7(a) and 504 loan: lender, borrower state, NAICS code, gross approval, term, initial rate.
How your plan uses it
The lenders section — how many lenders funded your industry in your state, how many loans, the median and quartile approval amounts, and the most active lenders by name.
What it does not tell you
Approvals, not disbursements or approval rates: the data records loans that were approved, and says nothing about applications that were declined. The current fiscal year is incomplete until 30 September — plans quote the latest COMPLETE year, FY2025.
Go to the source ↗

Quarterly Census of Employment and Wages (QCEW)

All U.S. counties

U.S. Bureau of Labor Statistics · Most recent published quarter

What it gives
Average weekly wage and employment by industry and county.
How your plan uses it
Pricing your staffing plan at the real local wage rather than a national average, in Organization and Management.
What it does not tell you
Published roughly five to seven months in arrears. Where a county figure is unavailable the plan falls back to state then national, and says which level it used.
Go to the source ↗

Places API

Where Google has coverage

Google · Live, cached 30 days

What it gives
Named businesses near your location matching your industry.
How your plan uses it
The named-competitor paragraph in Market Analysis, when available.
What it does not tell you
The only paid source we use, and the only optional one. If it is unavailable the competitive section relies on Census counts alone — the plan never invents a competitor.
Go to the source ↗

Industry ranges for assumption bounds

All industries, with dedicated bands for the most-financed codes

Derived from IRS Statistics of Income and Census CBP/SUSB · Latest published aggregates

What it gives
Plausible ranges for gross margin, payroll share, occupancy cost, marketing spend, net margin, revenue per employee and working-capital days, by industry.
How your plan uses it
Bounding every assumption before it reaches the model. An 85% gross margin on a restaurant is refused and corrected, and the correction is printed in the plan.
What it does not tell you
These are wide bands meant to refuse the absurd, not a substitute for RMA Annual Statement Studies, which is a paid subscription we do not resell.
Go to the source ↗

And the part that is not data at all

The written sections are drafted by a language model. It never calculates: it receives the figures the financial engine computed and may only use those. Every draft is scanned, and any dollar amount, percentage or ratio that does not appear in the computed figures is rejected and the section is rewritten. Anything still flagged after that is shown to you rather than quietly published.

No U.S. government agency endorses this site or its use of their data.