The jobs report is one document with two surveys inside it. Most misreadings come from treating it as one number. The headline payroll figure and the unemployment rate come from different surveys, count different things, and can move in opposite directions in the same month without either being wrong. This guide teaches you to read the U.S. Bureau of Labor Statistics (BLS) Employment Situation release the way its own technical note describes it: know which survey each number comes from, read the revisions, and check whether a change is large enough to mean anything at all.
This article comments on no specific month’s data, forecasts no release, and gives no investment advice. Figures quoted from the BLS technical note are the agency’s own worked examples, not current data.
## 1. Quick answer: two surveys, one headline, three revisions
The Employment Situation is published by BLS, usually on the first Friday of the month at 8:30 a.m. Eastern Time, covering the previous month. “Usually” matters: the BLS release schedule governs, and some releases fall on the second Friday. Check the schedule rather than assuming the date.
Three facts to hold before you read any coverage:
1. **The payroll number and the unemployment rate come from two different surveys.** The establishment survey counts jobs on payrolls. The household survey counts people and their labor force status. They are answers to different questions, published in one document.
2. **The first estimate is preliminary.** Each month’s payroll figure is revised twice in the two following releases, and the whole series is re-anchored once a year against near-complete employment counts. The revised number is the one to quote.
3. **One month is not a trend.** BLS publishes its own error bands. A monthly change smaller than the error band cannot be distinguished from no change at all. Section 6 shows the arithmetic.
The Federal Reserve reads this same release when it assesses labor market conditions; our guide to reading a Fed decision explains how that assessment appears in the statement. See [How to Read a Fed Rate Decision: Statement and Dot Plot](https://news.meiguoyes.com/how-to-read-fed-rate-decision-statement-dot-plot/).
## 2. The two surveys: jobs versus people
The technical note at the end of every Employment Situation release opens with the same sentence: the release “presents statistics from two major surveys,” the Current Population Survey (CPS, the household survey) and the Current Employment Statistics survey (CES, the establishment survey).
**The household survey (CPS)** is a sample survey of about 60,000 eligible households, conducted by the U.S. Census Bureau for BLS. It classifies each person 16 years and over as employed, unemployed, or not in the labor force. From it come the unemployment rate, the labor force participation rate, and the employment-population ratio. Its results appear in the release’s “A” tables, marked HOUSEHOLD DATA. Its reference period is generally the calendar week containing the 12th day of the month.
**The establishment survey (CES)** collects payroll records each month from about 119,000 businesses and government agencies, representing approximately 622,000 individual worksites — about 26 percent of all nonfarm payroll jobs. It produces employment, hours, and earnings of employees on nonfarm payrolls, by industry. Its results appear in the “B” tables, marked ESTABLISHMENT DATA. Its reference period is the pay period including the 12th, which may or may not match the calendar week.
The differences between the two surveys explain most apparent contradictions:
– **The household survey counts people; the establishment survey counts jobs.** A person with two jobs appears once in the household survey and twice in the establishment survey, once on each payroll.
– **Coverage differs.** The household survey includes agricultural workers, self-employed workers whose businesses are unincorporated, unpaid family workers, and private household workers. The establishment survey excludes all of these.
– **Unpaid absence is treated differently.** The household survey counts people on unpaid leave as employed. The establishment survey does not count workers who receive no pay for the reference pay period.
– **Age limits differ.** The household survey covers only people 16 and older. The establishment survey has no age limit.
BLS itself notes that the establishment survey’s month-to-month employment change has a smaller margin of error than the household survey’s, because of its much larger sample — while the household survey has the more expansive scope. Neither survey is “the real one.” Read each number against the survey that produced it.
## 3. Nonfarm payrolls: what the headline number is
“Nonfarm payroll employment” is the count of employees on nonfarm payrolls — people who worked or received pay for any part of the reference pay period, including people on paid leave. The headline is the *change* in that count from the previous month, not the total. A gain of any size means payrolls grew by that many jobs, net; it does not mean that many people found work, and it is not the total number of jobs in the economy. Keep the change and the level separate.
**Seasonal adjustment.** The headline figures are seasonally adjusted. BLS adjusts because employment follows regular annual patterns — holidays, weather, the school year — that can swamp the underlying movement. The technical note’s example: payroll employment in education falls by about 20 percent at the end of the spring term and rises again with the fall term. Seasonal adjustment tempers these regular patterns so that nonseasonal change is easier to spot. Two practical consequences: the adjusted and unadjusted figures for the same month will differ, and seasonal factors are recalculated every month, which is one reason earlier months get revised.
**Industry detail.** The establishment survey breaks payrolls down by industry, classified under the North American Industry Classification System. Reading only the total misses the composition: gains concentrated in one or two industries tell a different story from gains spread across many. The release narrative names the industries with notable changes; the B tables carry the full breakdown. Check whether the story you read mentions any industry at all — if it only quotes the total, it has read half the survey.
**Hours and earnings come from the same survey.** Average hourly earnings and the average workweek are establishment-survey figures, not household-survey figures. They describe payroll jobs, and earnings figures are averages across all employees on private payrolls — they are not the pay of any particular worker.
## 4. The unemployment rate: who counts as unemployed
The unemployment rate is the number of unemployed people as a percentage of the labor force, and both halves of that fraction come from the household survey. BLS classifies a person as unemployed only if all three conditions hold: no employment during the reference week, available for work, and specific active efforts to find employment during the four-week period ending with the reference week. People laid off and expecting recall need not be looking for work to count as unemployed. The estimate has nothing to do with unemployment insurance: BLS states plainly that the household survey’s unemployment data “in no way depend upon the eligibility for or receipt of unemployment insurance benefits.” A person can be counted as unemployed without receiving benefits, and receiving benefits does not by itself determine the classification.
**Read the participation rate with it.** The labor force is the employed plus the unemployed. The labor force participation rate is the labor force as a percentage of the population. The unemployment rate can fall for two very different reasons: more people found work, or fewer people are counted in the labor force at all. A falling rate alongside a falling participation rate is not the same news as a falling rate alongside a rising one. The employment-population ratio — the employed as a percentage of the population — is the third leg: it shows whether employment itself is growing relative to the population, regardless of who is counted as looking.
**U-6, defined and left there.** BLS publishes six alternative measures of labor underutilization, U-1 through U-6, in Table A-15 of the release. The official unemployment rate is U-3. U-6 is the broadest: total unemployed, plus all people marginally attached to the labor force, plus people employed part time for economic reasons, as a percentage of the civilian labor force plus the marginally attached. “Marginally attached” means wanting and available for a job, having looked sometime in the past 12 months, but not actively searching in the last four weeks. “Part time for economic reasons” means wanting and available for full-time work but having settled for a part-time schedule. U-6 answers a broader question than U-3; it does not correct it. When a story quotes U-6 without saying what it includes, it is quoting a different definition, not a truer one.
## 5. Revisions: the first number is a draft
The establishment survey revises its initial monthly estimates twice, in the immediately succeeding two months, to incorporate additional sample reports and recalculated seasonal adjustment factors. The reason is mechanical: estimates for the most recent two months rest on incomplete returns, and the tables label them preliminary. Only after two successive revisions, when nearly all sample reports have been received, is an estimate considered final.
Then comes the annual benchmark. Once a year, the sample-based estimates are adjusted to universe counts of payroll employment from the administrative records of the unemployment insurance program. The difference between the March sample-based estimate and the March universe count is the benchmark revision, which BLS describes as “a rough proxy for total survey error.” In the current technical note, the average absolute benchmark revision for total nonfarm employment over the prior ten years is 0.2 percent, ranging from −0.4 to 0.3 percent. Benchmarks also incorporate changes in industry classification when necessary.
A further annual pass revises five years of seasonally adjusted history in both surveys. The practical rule for readers: never quote a first estimate as settled, and when comparing two months, compare figures of the same vintage — a first estimate against a twice-revised figure is not a like-for-like comparison.
This is the same status discipline that applies to any official document: a first version is not the final version. Our guide to Federal Reserve proposals makes the parallel point about comment deadlines and effective dates. See [Federal Reserve Proposals: Comment Deadlines Are Not Effective Dates](https://news.meiguoyes.com/federal-reserve-proposals-comment-deadline-effective-date/).
## 6. Common misreadings
**Treating one month as a trend.** Monthly changes are noisy. BLS conducts its analyses at the 90-percent level of confidence, and the current technical note puts the confidence interval for the monthly change in total nonfarm employment at about plus or minus 122,000. BLS gives its own worked example: a reported increase of 50,000 means the 90-percent interval runs from −72,000 to +172,000, which includes values below zero — so “we could not say with confidence that nonfarm employment had, in fact, increased that month.” A reported rise of 250,000, by contrast, has an interval entirely above zero. BLS updates this figure periodically, so check the current technical note rather than memorizing the number. For the unemployment rate, the note puts the 90-percent interval for a monthly change at about plus or minus 0.3 percentage point when the rate is around 6 percent. A 0.1-point move is usually inside the noise.
**Reading “jobs added” as “people hired.”** The payroll change is a net change in jobs on payrolls. It is not a count of people who started work, not a count of hiring, and not the total number of jobs. Gross hiring and separations are far larger than the net change; the release does not report them.
**Reading a falling unemployment rate as “everyone found work.”** The rate is a ratio. It falls when the unemployed find jobs, and it also falls when people leave the labor force. Check participation and the employment-population ratio before drawing the conclusion.
**Ignoring the weather footnote.** Severe weather affects the two surveys differently. In the establishment survey, an employee who receives pay for any part of the pay period — even one hour — is counted; weather reduces payroll employment only for workers off without pay for the entire pay period, and more often shows up in average weekly hours. In the household survey, people who miss the whole reference week for weather reasons are counted as employed whether or not they were paid. When a release month had major storms, read the release’s own discussion before attributing a move to the economy.
**Quoting the headline without the revision.** Every release revises the prior two months. A story that quotes this month’s first estimate but ignores last month’s revision is using half the document. The revision paragraph sits in the establishment-survey narrative; read it before the headline.
## 7. How to check the release yourself
Start at the BLS news release page for the Employment Situation. The document has a stable shape:
1. **The summary.** The opening paragraphs state the payroll change and the unemployment rate, in that order, with the household and establishment narratives following.
2. **Summary Table A (household data) and Summary Table B (establishment data).** Table A carries the labor force, unemployment, and participation figures; Table B carries payroll employment, hours, and earnings by industry. Match each number you quote to its table.
3. **The detailed A and B tables.** These carry the demographic and industry breakdowns. The alternative underutilization measures are in Table A-15.
4. **The technical note.** It sits at the end of the release and holds the survey descriptions, the seasonal adjustment explanation, and the reliability discussion — including the confidence intervals quoted in Section 6. It is the part most coverage never opens, and the part that answers most “is this change real?” questions.
5. **The FAQ page.** BLS publishes a short Employment Situation FAQ answering why there are two employment measures, why revisions happen, and how weather affects the estimates. It is the fastest check when a claim about the report sounds wrong.
When a headline number and the release disagree, the release wins. Our five-step verification method applies here exactly as it does to any official announcement: find the original document, read the relevant section, and compare the quoted figure with the figure in the table. See [How to Verify a News Claim Against Official Sources](https://news.meiguoyes.com/how-to-verify-news-check-claims-official-sources/). The same compare-the-headline-with-the-filing discipline we use for earnings claims works for the jobs report: the headline is a summary of the document, not the document. See [How to Check a Company Earnings Claim in the SEC Filing](https://news.meiguoyes.com/check-company-earnings-claim-sec-filing-10-q-8-k/).
## 8. The 15-minute reading checklist
1. **Check the date and the reference month.** Confirm which month the release covers and that you are reading the current release, not last month’s.
2. **Read the payroll change and its error band.** Is the change larger than about 122,000? If not, do not call it a gain or a loss with confidence.
3. **Read the revision paragraph.** What happened to the prior two months? The revised picture may differ from the one headlines carried.
4. **Read the unemployment rate with participation.** Did the labor force grow, shrink, or hold steady? Did the employment-population ratio move the same way?
5. **Scan the industry detail.** Which industries moved, and is the movement broad or narrow?
6. **Check the technical note for anything unusual.** Weather, benchmark timing, and classification changes are disclosed there.
7. **Hold the conclusion.** If the change is inside the error band, if the two surveys point in different directions, or if the revision rewrote last month, say so — and wait for the next release before calling it a trend.
Sources and update time: facts in this article were read on the official pages below on November 17, 2026 (HST). BLS updates its technical-note figures periodically; check the current release and technical note for the latest values. No specific month’s data is quoted or commented on in this article.
– U.S. Bureau of Labor Statistics, Employment Situation news release (current release): https://www.bls.gov/news.release/empsit.nr0.htm
– U.S. Bureau of Labor Statistics, Employment Situation Technical Note (two surveys, seasonal adjustment, reliability, benchmark revision): https://www.bls.gov/news.release/empsit.tn.htm
– U.S. Bureau of Labor Statistics, Employment Situation Frequently Asked Questions (two measures, revisions, weather effects): https://www.bls.gov/news.release/empsit.faq.htm
– U.S. Bureau of Labor Statistics, Schedule of Releases for the Employment Situation: https://www.bls.gov/schedule/news_release/empSit.htm
– U.S. Bureau of Labor Statistics, Current Population Survey concepts and definitions (alternative measures U-1 to U-6): https://www.bls.gov/cps/definitions.htm
This article is a reading guide only. It is not a forecast, an economic assessment, or investment advice.

