Using U.S. Census Bureau and BLS data for secondary research

What to take away

• Secondary market research in the United States starts with two federal producers: the U.S. Census Bureau for demographics and the Bureau of Labor Statistics for prices, spending and work.
• The American Community Survey gives tract-level detail on income, education, housing and commuting that no private panel matches at that sample size.
• The Current Population Survey supplies monthly labor force numbers and an annual demographic supplement that analysts use to size households and workers.
• The Consumer Price Index and the Consumer Expenditure Survey answer different questions: one tracks price change, the other tracks what households actually buy.
• The BLS developers API returns published series as JSON, so you can refresh a table without retyping numbers from a web page.
• Every figure you publish needs the survey, the geography, the period and the seasonal adjustment status attached to it.

Why Census and BLS data anchor secondary market research

Federal statistical data is the cheapest defensible evidence a small team can buy, because it is free and because the sampling is documented. A private panel of 500 respondents in one metro cannot tell you how many households in that metro earn above $100,000. The American Community Survey can.

The two agencies split the work. The U.S. Census Bureau counts people, households and housing. The Bureau of Labor Statistics measures prices, wages, employment and spending. Together they cover the demographic and economic context that frames a market before you spend money on primary work.

That framing matters most when a client asks whether a market is big enough to enter. You are not proving demand. You are establishing how many people, households or firms could plausibly buy, and what their incomes and costs look like. Census and BLS data do that job well.

Use them as the spine of a project grounded in consumer research ethics guidelines us, then layer trade association data and company filings on top. Federal series are consistent across states, which makes regional comparison possible without adjusting for methodology differences.

One caution. Federal data is published for statistical purposes, not for commercial targeting. It tells you about populations, not about named prospects. Treat it as context, not a lead list.

American Community Survey datasets for demographic context

The American Community Survey is the Census Bureau's continuous household survey, and it is the workhorse for local demographic profiles. It publishes one-year estimates for areas of 65,000 people or more and five-year estimates for areas down to census tract level.

Five-year estimates pool 60 months of responses. That buys precision in small places at the cost of currency: a five-year estimate centered on a recent period still reflects conditions from several years back. For a fast-changing metro such as Austin or Charlotte, that lag matters.

One-year estimates are more current but only available for larger geographies. If your client needs a current figure for a mid-sized county, you may have no one-year option at all.

The tables analysts use most often cover median household income, educational attainment, age distribution, housing tenure, rent burden, commuting mode and language spoken at home. Each is downloadable as a table, a map or an API call.

Data profiles and subject tables are the quickest entry point. They pre-select variables and present them with margins of error, which saves you from assembling a custom tabulation for a routine market profile.

Margins of error deserve attention. ACS estimates come from a sample, so a small tract's median income might carry a margin of error of several thousand dollars. Reporting the point estimate alone overstates what you know.

Compare ACS geography to your client's actual trade area before you build anything. Metropolitan statistical areas, counties, places and tracts do not line up with drive-time radii, and a mismatch will quietly distort every per-household calculation you make afterward.

Current Population Survey for labor force and household detail

The Current Population Survey is a monthly survey of about 60,000 households conducted for the Bureau of Labor Statistics. It produces the national unemployment rate, labor force participation and the counts behind the monthly employment situation report.

Its value for market research is the supplement structure. The Annual Social and Economic Supplement, fielded each spring, adds detailed income, poverty and health insurance questions. The October school enrollment supplement, the tobacco use supplement and others appear on rotating schedules.

Those supplements support specific research questions. If you are sizing a market for after-school programs, the enrollment supplement gives you school attendance by age and state. If you are studying benefit take-up, the ASEC income detail is more granular than anything in the monthly series.

BLS publishes an overview of its demographic data that explains how these surveys represent the population and where the sampling limits sit. Consult it before you quote a subgroup estimate, because small subgroups carry wide confidence intervals that a headline number hides.

Microdata is available through the Census Bureau and through the National Bureau of Economic Research, which means you can build custom tabulations rather than relying on published tables. That takes more work but answers questions the standard tables were never designed for.

CPS is a household survey, so it does not cover business establishments. For firm counts and payroll employment by industry, you need separate programs. Mixing the two without saying so is a common error in market sizing decks.

CPI and Consumer Expenditure Survey series for spending context

The Consumer Price Index measures the change in prices paid by urban consumers for a fixed basket of goods and services. It is the standard deflator for any multi-year spending comparison in the United States.

That deflation step is not optional. If you compare 2019 and 2024 revenue without adjusting for price change, you are partly measuring inflation. CPI lets you separate volume growth from price growth, and the index is available by region and by detailed item category.

The Consumer Expenditure Survey answers a different question: what do households actually buy? It collects spending by category alongside income and household characteristics, and the CE home page is the entry point for published tables and microdata.

CE data is published as cross-tabulations by income quintile, age, region, household size and race. That structure lets you estimate a category's share of wallet for a specific household type rather than applying a national average to everyone.

Two limits apply. The survey undercounts some populations and relies on recall, so totals run below national accounts aggregates. And the sample is smaller than CPS, which means state-level detail is thin outside the largest states.

Use CPI to adjust, CE to allocate. A spending estimate that combines both, with the vintage of each noted, holds up far better under scrutiny than a single vendor number. For a structured review of whether a figure is trustworthy, keep a secondary market research checklist beside the data.

Series	Produced by	Best used for	Geography available
American Community Survey	U.S. Census Bureau	Income, education, housing, commuting profiles	Nation to census tract
Current Population Survey	Bureau of Labor Statistics	Employment, unemployment, income supplements	Nation and state
Consumer Price Index	Bureau of Labor Statistics	Deflating spending and revenue over time	Nation, region, selected metros
Consumer Expenditure Survey	Bureau of Labor Statistics	Category spending by household type	Nation, region, some states

Pulling BLS series through the developers API

The BLS developers API exposes published series as structured data, which removes the copy-and-paste step from a recurring report. Registration is free, and the documentation covers series identifiers, request formats and limits.

Series identifiers are the hard part. Each one encodes the survey, the seasonal adjustment, the item or industry and the geography, and a single wrong character returns a different series rather than an error. Build your identifiers once and store them.

1. Register for an API key through the BLS developers page and note the daily request limits for registered versus unregistered users.
2. Look up the series identifier for each series you need, using the survey's own table pages to confirm the code matches the right geography and adjustment.
3. Send a GET request to the version 2 endpoint with your series IDs, the start and end years, and your key in the registration header.
4. Parse the JSON response, map the period codes to calendar months or years, and load the values into your working table.
5. Cache the response locally and set a refresh schedule that matches the release calendar for each survey.

Two operational details cause most failures. Monthly and annual series use different period codes in the same response object, so a parser that assumes months will misread annual values. And series get revised, so a cached figure from last month may no longer match the current release.

Schedule refreshes around the release calendar rather than a fixed weekly job. CPI comes out monthly, employment situation data comes out monthly, and CE tables are released on their own schedule. Pulling before a release wastes calls and can leave you reporting a stale number.

Geography and industry breakdowns for regional comparisons

State and metro comparisons are where federal data earns its keep, because every state is measured the same way. Large states such as California, Texas, Florida and New York all have full ACS and CPS coverage, so a multi-state profile needs no methodology patchwork.

Start with the largest geography that answers the question, then narrow. A national CPI figure supports a broad inflation adjustment. A regional CPI figure supports a coastal versus interior comparison. A metro-level ACS table supports a site selection memo.

Industry breakdowns come from establishment surveys rather than household ones. Those programs publish employment, wages and establishment counts by industry and area, which is what you need to size a business-to-business market.

Keep the household and establishment views separate in your write-up. A household survey can tell you how many people work in an industry; an establishment survey tells you how many firms employ them. Both numbers are useful and they will not match.

For regional work, the practical constraint is sample size. Small states and rural counties have wider margins of error, and a comparison that ignores them can turn noise into a finding. Rank states by estimate only after you have looked at the error bands.

If you are tracking how a category has shifted across regions over several years, hold the geography, the vintage and the deflator constant. Otherwise you are comparing unlike things, which is the same discipline any review of common market research strategy questions needs.

Methodology notes and definitions that prevent misuse

The BLS Handbook of Methods is the reference for how each series is constructed, and it is the document to cite when someone questions your numbers. It covers sampling, estimation, seasonal adjustment, revision policy and known limitations for each program.

The Census Bureau publishes comparable documentation for ACS and CPS, including response rates, question wording and the rules for comparing estimates across years.

Four definitions cause most of the trouble. Seasonally adjusted and not seasonally adjusted series are different numbers and are not interchangeable. Real and nominal dollars differ by the deflator. Household and family are distinct units in CPS. And metropolitan area boundaries change over time.

Vintage is the fifth. ACS five-year estimates are re-released each year with a new end date, and comparing overlapping vintages can double-count the same underlying responses. Always state which vintage you used.

Suppression is the sixth. Small-area estimates get suppressed or published with wide error bands to protect confidentiality, so a blank cell is not a zero.

When a vendor number and a federal number disagree, the federal methodology documentation usually explains why. Check the definition before you assume either one is wrong. A short review of consumer research panel cost canada before publication catches most of these mismatches.

Documenting secondary market research sources properly

Documentation is what separates a defensible analysis from a plausible one. For each federal figure you use, record six things: the survey or program, the exact table or series identifier, the geography, the period, the seasonal adjustment status and the date you retrieved it.

A retrieval date matters because federal tables get revised. A figure pulled in March may differ from the same table pulled in September, and a reader trying to reproduce your work needs to know which version you saw.

Store the raw download alongside the analysis. If someone challenges a number two years later, you can show the file rather than reconstructing the query.

Link to the source page, not to a search result or a third-party summary. Federal agencies keep stable URLs for their main program pages, and a reader who lands on the program page can find the exact table. The CPS home page is a good example of a stable program URL.

When you combine federal data with commercial estimates, label each layer. A market size built from ACS household counts and a vendor's penetration rate is a hybrid, and the vendor assumption is the part most likely to be wrong. Apply the same scrutiny you would apply to diy versus agency consumer research cost from any other source.

Keep a running source log for recurring reports. The next analyst who inherits the file will spend their first day rebuilding what you already knew.

Common questions

Can I use Census and BLS data for commercial market research? Yes. Both agencies publish data for public use, and the main constraints are statistical rather than legal. You cannot use the data to identify individuals, and you should not present estimates as counts of named businesses or households.

How current is American Community Survey data? One-year estimates arrive roughly a year after the collection year. Five-year estimates lag further because they pool 60 months of responses, so small-area figures are always several years behind the present.

Should I use seasonally adjusted or unadjusted CPI? Use seasonally adjusted series for month-to-month comparisons and unadjusted series for annual averages or for anything tied to a contract or index. Never mix the two in one table.

What is the difference between CPS and the establishment survey? CPS surveys households and counts employed people, including the self-employed. The establishment survey collects payroll records from employers and excludes agricultural workers and the self-employed. Their totals differ by design.

Do I need an API key to pull BLS series? You can make a limited number of requests without registering, but a free key raises the daily limit and is required for the version 2 endpoint features. Register through the BLS developers page.

How do I cite a federal data table in a client report? Name the program, the table or series identifier, the geography, the period and the retrieval date. Link to the program page so the reader can locate the table independently.