Where to find market data you can cite

Market research data sources you can cite: a US guide

Your first market research data sources should be free US public records: Census Bureau for businesses, revenue and retail sales; BLS for jobs, wages and household spending; BEA for output and consumer spending; SEC EDGAR for company filings. Buy a report only for the cut public data cannot make, and cite each figure's publisher, date and what it counts.

What makes a market figure citable?

A market figure is citable when it can carry a full source line: who published it, under what title, on what date, and what exactly it counts. The partner will ask where the number came from, and the client will ask again in the meeting. A number without a source is worse than no number at all, because it invites the question and cannot answer it.

Most numbers a search engine returns fail on the last part. A market-size headline travels from a report summary to a press release to a blog post, shedding its definition at each step, until only a dollar figure, a year and a growth rate survive. The original may have counted software alone, software plus services, or one region; the copy no longer says. Paywalls compound the problem: the summary is free, the methodology page is not.

US government statistics are built the other way around. Census Bureau, BLS and BEA tables are free, dated, revised on a published calendar and documented down to which businesses are left out. They rarely answer the brief in its own words. They are, however, the figures nobody in the room can wave away, which makes them the foundation.

Build from what is counted, then add what is estimated.

Which free US sources count what?

Ten free market research data sources answer most US market questions. Each counts something specific, at a specific level of detail, on a specific lag. The table records all three as each publisher's page stood on October 17, 2026.

Free US market data sources and what each counts (as of October 17, 2026)
Source and publisherWhat it countsDetail and latest releaseUse it for
Economic Census (U.S. Census Bureau)Establishments, employees, payroll, and sales, shipments or revenue, for more than 950 industriesEvery five years, for years ending in 2 and 7; the 2022 edition finished releasing in March 2026 and covers nearly 21,000 geographic areasThe base-year size of an industry, nationally and by state or metro
Annual Integrated Economic Survey (U.S. Census Bureau)Business revenues, expenses, payroll and employment; it replaced seven annual surveys, including the Service Annual SurveyAnnual; 2024 time series released September 10, 2026Revenue growth between census years
County Business Patterns (U.S. Census Bureau)Establishments with paid employees, mid-March employment, and first-quarter and annual payrollAnnual; 2023 is the latest data year; county, metro and ZIP code detail; excludes crop and animal production, rail, postal service, private households and public administrationCounting buyers or competitors in a territory
Quarterly E-Commerce Report (U.S. Census Bureau)Retail e-commerce sales and their share of total retail salesQuarterly; second quarter 2026: $340.2 billion, 17.1% of retail sales, seasonally adjusted (released August 18, 2026)Channel shift in a retail category
Quarterly Census of Employment and Wages (BLS)Employment and wages reported by employers, covering more than 95% of US jobsQuarterly; first quarter 2026 released August 28, 2026; county, metro, state and national, by industryIndustry growth and labor cost, down to county level
Consumer Expenditure Surveys (BLS)Spending, income and demographics of US consumer unitsAnnual; 2024 average spending of $78,535 per consumer unit (released December 19, 2025)Spending per household in a bottom-up consumer sizing
Personal consumption expenditures (BEA)The value of goods and services bought by or on behalf of US residentsMonthly, national; August 2026 data released September 30, 2026; BEA cautions that its more detailed estimates rest on less reliable source dataNational consumer category trends
GDP by industry (BEA)Each industry's value added, with compensation, operating surplus and taxesQuarterly, with the third GDP estimate; latest release September 30, 2026An industry's growth against the whole economy
EDGAR (U.S. Securities and Exchange Commission)Company filings, with full-text search of electronic filings since 2001Continuous; a 10-K must describe revenue-generating activities, segments and competitive conditionsCompetitor revenue, segment size and market position
FRED (Federal Reserve Bank of St. Louis)Economic series gathered from many publishers, browsable by sourceUpdated as each source releases; running since 1991Charting a series fast, then citing the original publisher

95%+

of US jobs are counted every quarter in the BLS Quarterly Census of Employment and Wages, by industry and countyU.S. Bureau of Labor Statistics, Quarterly Census of Employment and Wages (first quarter 2026, released August 28, 2026)

Two cautions apply across the table. Census business statistics are organized by industry code (NAICS), which files an establishment under its main activity, not under every product it sells, so a product market rarely matches one code exactly. And the lag differs by series: the 2022 Economic Census finished releasing in March 2026, County Business Patterns runs to 2023, and the quarterly series reach the first half of 2026. This means one slide can carry a 2022 base and a 2026 trend, provided each figure is labeled with its own year.

Coverage is stated, not hidden. County Business Patterns counts only establishments with paid employees, and says so; a market made up largely of sole proprietors is therefore undercounted there. That stated limit is what makes the figure usable: the footnote can name the gap.

The usual objection, that public data is outdated, misreads the problem. A 2022 census count with a stated scope can be rolled forward with a quarterly series and explained in one footnote. A 2025 headline with no stated scope cannot be explained at all.

How do you build a citable market figure from public data?

The sequence below works for any US market that maps to an industry or a consumer category. It is the sourcing half of a market sizing; the arithmetic of sizing a market top-down or bottom-up sits on top of it.

  1. Write the definition before searching: the product or service, the customer, the geography, the year and the unit (revenue, units or establishments). Every source is judged against this sentence later.
  2. Find the six-digit NAICS code that matches the definition, or the two or three codes that together contain it, and note what each includes and leaves out.
  3. Take the base from the 2022 Economic Census: establishments, payroll and sales or revenue for those codes, nationally and for the states or metro areas in scope.
  4. Roll the base forward with an annual or quarterly series: the Annual Integrated Economic Survey for revenue, QCEW for employment and wages, the retail and e-commerce reports for consumer channels. State the roll-forward method in one line.
  5. Build the bottom-up check: County Business Patterns for the number of buyers or competitors in a territory, and Consumer Expenditure Survey spending per household for consumer categories.
  6. Search EDGAR for the listed companies in the market and read each 10-K's business description, segment revenue and competition discussion. Where those segments fall inside your definition, their US revenue added together is a floor the market cannot sit below.
  7. Only then open the published market reports. Record each one's definition and base year beside the public figure, and mark every gap as definitional or genuine before using either number.
  8. Write the source line for each figure as it is collected, not the night before the meeting.

Step 6 is also the answer to a competitor briefing. Under the SEC's Regulation S-K, Item 101, a listed company's 10-K must describe its revenue-generating activities, its reportable segments and the trends in market demand and competitive conditions it faces. That is a rival's position, in its own words, in a document it is legally accountable for. EDGAR's full-text search reaches back to 2001, so a product name can be followed through every filing that mentions it, competitors' included.

Private competitors leave no such record. For them, the defensible limit is what they disclose themselves plus the establishment counts in County Business Patterns; anything beyond that is an estimate and should be labeled as one.

When is a paid market report worth buying?

Public data stops at four boundaries: a product cut finer than an industry code, market shares among private companies, forecasts, and most markets outside the US. A category manager reviewing one grocery subcategory, a fund screening a niche software segment and an agency pitching into an unfamiliar industry all reach the first boundary on day one.

That is where published market reports earn their price, on one condition: the scope note. A serious report defines what it counts (which products, channels, countries and base year), usually in a methodology section the free summary leaves out. Without that page, the figure is a headline, not a source. Compare market research pricing against the single segment you need before buying a library of reports.

When two reports disagree, they are usually measuring different things. Line up the definitions first; a gap that survives identical scope and base year is a genuine disagreement and belongs on the slide as a range, never as an average. The method to reconcile two estimates that disagree starts from exactly that comparison.

The remaining sources are citable on the same terms, with one more field: who paid. Trade association figures are often the only count of a niche, and they are funded by the industry they count. Supplier-commissioned research tends to favor the supplier's own range, and a board knows it; cite it as the supplier's view, not as the market. Sell-side notes carry the bank's name, date and conflicts. None of these is unusable. Each needs its sponsor in the source line.

How do you cite market data in a deck, memo or board report?

A source line has five parts, and a slide footnote can hold all five.

The five parts of a source line, using a Census Bureau figure
PartWhat it answersExample
PublisherWho stands behind the numberU.S. Census Bureau
TitleWhich release a reader should openQuarterly E-Commerce Report, second quarter 2026
DateWhich vintage, since series are revisedReleased August 18, 2026
BasisWhat exactly is countedRetail e-commerce sales, seasonally adjusted, not adjusted for price changes
FigureThe number as published, with its unit$340.2 billion; 17.1% of total retail sales

The date is part of the figure. The Census Bureau marked its second-quarter 2026 e-commerce estimates as no longer the most current as of September 28, 2026, with revised figures due in the third-quarter release on November 19, 2026. A footnote that names the release can be updated in a minute; one that says only "Census, 2026" cannot.

Aggregators are a route, not a source. FRED, run by the Federal Reserve Bank of St. Louis, gathers series from many publishers and lists each one's source; cite the BLS or BEA release behind the chart, and mention FRED only as where the series was pulled. The same rule covers a figure found inside another report or an article: follow it back to the original publisher and cite that, or cite it as quoted, with both names and both dates.

The format does not change with the audience, only the placement. In an investment memo the line sits under each market figure, and the market section of an investment memo leans on these sources more than any other page. In a board report it moves to an appendix. In a dissertation or an MBA assignment, APA, MLA or Chicago reorders the five parts without removing any of them.

Where does Caspr fit?

Running the sequence above by hand for every new sector is the desk research that crowds out the analysis. Caspr is an analyst: a single-line prompt returns a structured, cited report, as a PDF or an editable PPTX, in under 15 minutes, with each figure cited to its publisher. The Market Research Study ($80) behind the worked example below went further: it published the order in which it trusted its sources, for a sector priced off SEC filings, and marked the one figure it could not source.

The decision on which number goes on the slide stays with the person who presents it.

Worked example

How a Caspr Study ranked its sources for vertical SaaS valuations

In August 2026 a Caspr Market Research Study priced vertical SaaS for investors and published the order in which it trusted its sources. SEC filings and issuer disclosures ranked first, rated "highest for company-reported facts, but limited to issuer scope"; Toast's 10-K and a Samsara earnings release were taken from EDGAR. Specialist valuation datasets came next, trusted "provided methodology and update cadence are disclosed." They supplied the public anchor: a median of 3.7x EV/TTM revenue for vertically focused public SaaS through 2Q26, against 3.2x for the broader SEG SaaS Index. Market-size reports sat among the secondary sources, used cautiously because definitions of vertical software, vertical-specific software and vertical SaaS vary across publishers. Gartner's 1Q26 forecast put the worldwide vertical-specific software market at $543 billion by 2030; Future Market Insights put vertical market software at $194.2 billion in 2026. Those are two scopes and two years, not one market in dispute: the Study named Gartner's figure as its anchor and kept the other beside it rather than blending them. Where no transparent source published a vertical-only NTM median, it said so and used named public comparables instead. The result is the standard this guide sets: every multiple carries its basis, and the one figure that could not be sourced is marked as missing rather than filled.

Frequently asked

What are the best free market research data sources in the US?

For US businesses, start with the Census Bureau's Economic Census, Annual Integrated Economic Survey and County Business Patterns. For jobs and wages, the BLS Quarterly Census of Employment and Wages. For consumers, the BLS Consumer Expenditure Surveys and BEA's personal consumption expenditures. For companies, SEC EDGAR filings. They're all free, dated and documented, which makes them citable.

Where can I find a reliable growth rate for an industry?

Calculate an industry's growth rate from a public series, not a borrowed forecast. BEA's GDP by industry gives quarterly value-added growth; QCEW gives employment and wage growth by industry and county; the Annual Integrated Economic Survey gives revenue by year. A historical rate from a named series holds up. A forecast needs its publisher's assumptions beside it.

What if my market is too niche for government data?

Government data can still bound a niche market. Find the industry code containing the niche and take its establishments and revenue as a ceiling. Narrow it with company filings, trade association counts or a published report's scope note. Footnote the narrowing step. A clearly bounded estimate is more defensible than a precise figure with no stated scope.

Can I cite a market size I found quoted in an article?

Cite the original report, not the article that quoted the market size. Find the report's publisher, title and date, and read how it defines the market. If the original is behind a paywall, cite it as quoted in the article, with both names and both dates. Treat the figure as unexamined until you know its scope.

How do I cite market data in a dissertation or an MBA assignment?

Stick with primary statistical releases for market data. Census, BLS and BEA tables carry the publisher, title, release date and stable address every academic style requires. Cite the specific table or release, not the agency's home page, plus the date you retrieved it. Government series get revised on published schedules, so later readers may see different numbers.

Keep reading