Business Benchmarks Methodology

Data Standards

Sources come in a strict order of preference. Official government and regulatory data comes first: the US Small Business Administration, Census Bureau, IRS, Bureau of Labor Statistics, ISED Canada, and the Australian Taxation Office. Franchise disclosure documents, labeled per brand, come next, followed by industry association figures labeled clearly as association estimates. Nothing is invented, and no figure appears without a named source.

Every figure carries a source, a reference period or an explicit "vintage not stated" label together with the date we accessed it, and a clearly defined scope. When data exists only for a broader category than the business in question, the label says so in plain words: sector-level means the sector, not the specific business.

Definitions are kept apart on purpose. Revenue is not profit. Receipts are not income. SBA financing is not startup cost. Business density is not market saturation. Employer businesses and owner-operator businesses without employees are different populations and are never merged into one number.

Profiles start as unpublished drafts. A profile is published only after automated checks pass (complete source metadata on every metric, labeling rules, metadata length and uniqueness, and a similarity gate that blocks profiles too close to an existing one) and a human review approves it. Statistical thresholds are set before results are seen, for example a minimum of 50 loan records before any median is displayed.

If operator-contributed figures are collected in the future, they will be displayed in their own clearly separated section with sample sizes shown, and never blended into public-source figures.

Corrections are logged and published. If you spot an error on any profile, reach us through the site contact page and we will review it against the original source.

Primary sources. Every published figure traces to one of the sealed primary sources below, archived with excerpts at retrieval time:
Australian Taxation Office: Small business benchmarks
Coin Laundry Association: Industry Overview for investors
IRS Statistics of Income: Statistics of Income for nonfarm sole proprietorships
US Bureau of Labor Statistics: Business Employment Dynamics establishment age tables
US Census Bureau: Economic Census data API
US Census Bureau: County Business Patterns
US Census Bureau: Nonemployer Statistics
US Small Business Administration: Loan program FOIA datasets

All published figures live in the Business Revenue Benchmarks hub, where each card states its value, population, source and reference period.

Survival figures come from two different federal programs and the pages say which one they are using. Where a profile shows a 5-Year Industry Continuation Benchmark, the source is the Census Bureau's Business Dynamics Statistics, specifically the table of 4-digit NAICS industries crossed by establishment age from the 2023 release. Where a profile shows a Sector Survival Benchmark instead, the source is the Bureau of Labor Statistics, Business Employment Dynamics, Table 7.

The continuation benchmark is arithmetic BusinessNES performs on published counts: the number of establishments aged five in a given year divided by the number aged zero in the same industry group five years earlier. A five-cohort reference repeats that calculation for the cohorts that first reported paid employment from 2014 through 2018 and reports the range between them.

This is deliberately not called a survival rate. Census measures establishment age from the first year an establishment reports paid employment and increases it every year afterwards; an establishment that reports nothing for a period can appear again later, and industry classification can change between years. Across the 285 usable industry groups, meaning those with complete counts and a nonzero age-zero denominator, 70 rise between two ages and 7 exceed their own age-zero cohort, which cannot happen in a true survival curve.

Because of that, each benchmark carries a data quality tier. Standard means a complete series, a large enough age-zero cohort, a curve that falls from every age to the next and cohorts that agree with each other. Limited means the reference group is thin or the cohorts disagree by more than ten percentage points, and the page says so under the chart. Separately, when the displayed cohort does not fall from every age to the next, the card shows only the start and the five-year endpoint instead of drawing a path through inconsistent points. Where a group is suppressed, tiny, or exceeds its own cohort, no benchmark is published at all.

The Business Dynamics Statistics file publishes 2017 vintage NAICS codes, while every profile on this site uses NAICS 2022. The two are not interchangeable: 37 four-digit codes in that file do not exist in NAICS 2022, most of them in the retail restructure. Each profile therefore requires an approved entry in a crosswalk registry recording that its 2022 code maps to a 2017 group with the same scope. A profile without one shows no continuation benchmark.

Primary sources for this page: the 2022 Economic Census for revenue, establishment counts and payroll; the Census Business Dynamics Statistics, 4-digit NAICS by establishment age, for the industry continuation benchmark; and BLS Business Employment Dynamics for the sector survival series used as a fallback.

Business data and benchmarks by BusinessNES.