The difference between job postings data and workforce headcount data is the difference between what a company wants and what it has. Postings record the roles an employer says in public it intends to fill. Headcount records the people it employs on a stated date. This post sets each beside the official measure it resembles, in the definitions of the US Bureau of Labor Statistics (BLS), shows what each can and cannot tell you, explains why the two so often disagree, and works one example through so that you can reconcile them for a company you follow.
What is the difference between job postings data and workforce headcount data?
Job postings data records the roles a company advertises: demand for people, stated in public and dated by the day. Workforce headcount data records how many people the company employs on a given date. Postings count positions waiting to be filled and move daily. Headcount counts filled positions, and it changes only as people join and leave.
At the source, both come from the employer: a posting is its own publication of a role, and a stated headcount is its own statement of its size. They disagree because they measure different things.
| Property | Job postings data | Workforce headcount data |
|---|---|---|
| One record | One advertised role, with its title, location and dates | One count of employees on one date |
| What it measures | Demand: positions the employer intends to fill | The people employed, after every arrival and departure |
| How it moves | Daily, as roles open and close | With each statement by the company |
| Detail | Function, seniority, location, work mode, advertised pay | One total, sometimes with a split the company chooses |
| Read alone, it leaves open | Whether a role was filled, and who left | Where the company is hiring, and what moved inside the total |
Postings are the core of hiring data. Headcount is a field of firmographic data, the description of a company's size and shape.
Job postings vs employment and headcount data in BLS terms: JOLTS job openings, hires and employment
In the terms of the Bureau of Labor Statistics, job postings data sits closest to JOLTS job openings: positions open on the last business day of a month. Workforce headcount data sits closest to employment: persons on a payroll in the pay period that includes the 12th. Hires and separations are the flows between the two.
The Job Openings and Labor Turnover Survey (JOLTS) is a monthly survey of a sample of approximately 21,000 nonfarm business and government establishments in the United States. Its definitions keep apart three things that a count of postings tends to blur. The payroll survey, Current Employment Statistics (CES), supplies the definition of employment.
| Official measure | What it counts | When | Nearest company-level data |
|---|---|---|---|
| Job openings (JOLTS) | Positions with work available, that could start within 30 days, for which the employer is actively recruiting from outside the establishment | The last business day of the month | Open postings at the end of a day |
| Hires (JOLTS) | All additions to the payroll | The whole month | A closed posting is the nearest trace of a hire |
| Total separations (JOLTS) | Quits, layoffs and discharges, and other separations such as retirements | The whole month | The part of a change in headcount that hires leave unexplained |
| Employment (CES) | Persons on establishment payrolls who received pay. It is a count of jobs: a person at two worksites is counted at each | The pay period that includes the 12th | Stated headcount on the date it refers to |
| Employed (household survey) | People classified as employed in the reference week, each counted once even with more than one job | Usually the week that includes the 12th | Headcounts summed across companies, less the people on two payrolls |
The scale of each measure shows why they are not read as one. For August 2026, BLS reported 7.1 million job openings on the last business day of the month, with 5.2 million hires and 5.1 million total separations during it, across US nonfarm establishments, seasonally adjusted and preliminary. Total nonfarm payroll employment for the same month, on the same basis, stood at 159.0 million, 133,000 above July. By our arithmetic on the two releases, hires and separations together came to about 10.3 million moves, some 77 for each net job added to payrolls. A count of postings follows the gross figure. A headcount shows the net.
The two official surveys are reconciled with each other as well. Hires minus separations, by our subtraction of the levels in that release's table (5,192,000 less 5,070,000), came to 122,000, beside a payroll change of 133,000. The JOLTS technical note says that the two figures should be comparable, that definitional differences and sampling and nonsampling errors historically drew them apart, and that BLS applies a monthly alignment method to limit the divergence. Expect to do the same work for a company.
The rates carry over directly. In the same note, BLS divides hires by employment for the hires rate, and job openings by the sum of employment and job openings for the job openings rate, so that the rate is a share of all positions, filled and open. Eurostat builds its job vacancy rate in the same form, over occupied posts plus vacancies. Build a company's posting intensity that way: open postings over headcount plus open postings.
Where a posting differs from a job opening
A posting is evidence of a job opening, and three things separate it from the official count of one.
- The conditions. A JOLTS job opening has work available, could start within 30 days and is being recruited for from outside the establishment. A posting shows the recruiting. The other two conditions stay with the employer.
- The channel. Active recruiting, in the BLS definition, includes networking, help wanted signs, employment agencies and job fairs as well as internet notices. JOLTS collects its figures directly from sampled establishments, and a BLS article of August 2026 states that its method does not rely on online job listings. The same article notes that a posting may be left online after the position has been filled.
- The unit. The UK Office for National Statistics, in its guide to measuring labour demand from online job adverts (last revised 5 November 2024), notes that an online advert may hold more than one position, may be speculative, or may advertise training that leads to a job later. In the source it studied, the average number of live UK adverts in August 2024, after deduplication, was 66% higher than the monthly volume of vacancies in its official estimates for the same period.
That comparison is of adverts deduplicated after collection. Postings read from each employer's own board begin with one record for each published role, as our post on reading job postings at the source sets out. The first two points hold for any posting: read it as an intention to hire, stated and dated by the employer.
Where headcount differs from employment
CES employment is a count of jobs at US establishments, taken every month on one definition. A company's headcount is its own figure: the scope, the date and the way of counting are the company's. Two points decide how to read it.
- What is counted. Item 101 of the SEC's Regulation S-K, the description of business, in the edition of 1 April 2025, has a registrant discuss its human capital resources, including the number of persons employed, to the extent material to an understanding of the business. Read the sentence around a stated number for what it covers: full-time staff or all staff, one entity or the whole group.
- The date. A count refers to one date and is published later. An annual report on Form 10-K is due 60 days after the fiscal year ends for a large accelerated filer, 75 days for an accelerated filer and 90 days for other registrants. A count given in an annual report can therefore be two to three months old when it appears, and it stays the latest one until the next report.
What each dataset can and cannot tell you
Postings answer questions of direction and timing. They show which functions, seniority levels and countries a company is recruiting in, what pay it advertises, and the day demand rose or stopped. Outcomes lie outside a posting: one that closes may have been filled from outside, filled by a transfer or withdrawn, and one posting can stand for several hires.
Headcount answers questions of size and net result. It scales every other measure and places a company in a size band. The change between two counts is the sum of everything that happened to the workforce in between: hires, resignations, layoffs, acquisitions and disposals. Composition, and timing inside the interval, lie outside it.
Each supplies the other's missing term. What remains is an estimate of the flows that neither records, and the example further down works it out.
Where job postings and headcount disagree, and why
Four patterns cover most disagreements. Each has more than one cause, and the last column says what to check before you choose between them.
| Postings | Headcount | What may be happening | What to check |
|---|---|---|---|
| Rising | Flat | Replacement hiring, where new hires offset leavers, or roles that stay open unfilled | New postings against closed postings, and how long the open ones have been open |
| Rising | Falling | A company changing shape: reducing one function while recruiting in another, or selling a business | The mix of postings by function, and announcements of restructuring |
| Falling | Rising | An acquisition that added staff, or hires agreed earlier starting work as new demand slows | Announcements of acquisitions, and the date the count refers to |
| Falling | Falling | A contraction that both measures confirm | Whether closed postings outnumber new ones in every function |
Timing explains much of the rest. End the posting window on the date the count refers to: a count for the end of December belongs beside postings up to the end of December, even when it reached you in February. A posting series has traps of its own, such as a change of job board that reads as a fall, and the guide to hiring data in equity research covers them.
A worked example: Acme Robotics, postings up and headcount flat
Acme Robotics is an invented company and the example is illustrative. It combines a year of postings with two stated counts.
Assumptions:
- Stated headcount is 2,000 at the end of one financial year and 2,060 at the end of the next.
- In the twelve months between, 600 postings opened and 520 closed, and open postings rose from 80 to 160.
- Each posting stands for one position.
- 70% of closed postings ended in a hire from outside. The rest were withdrawn or filled by transfer. This is the assumption to vary, because a posting carries no record of its outcome.
- Every outside hire came through a posting, and Acme made no acquisition or disposal in the year.
The working rests on one identity.
headcount at end = headcount at start + hires - separations + acquired - divested- Check the postings: 80 plus 600 minus 520 gives 160 open at the end, which matches.
- Net change in headcount: 2,060 minus 2,000 is 60, or 3.0% of 2,000.
- Posting intensity in the BLS form: 80 / (2,000 + 80) = 3.8% at the start, and 160 / (2,060 + 160) = 7.2% at the end.
- Implied hires: 70% of 520 is 364.
- Implied separations, from the identity: 2,000 plus 364 minus 2,060 is 304.
- Rates on the average headcount of 2,030: hires 364 / 2,030 = 17.9%, separations 304 / 2,030 = 15.0%.
- Of the 364 hires, 304 replaced leavers (83.5%) and 60 were growth (16.5%).
Read alone, the postings describe a company recruiting hard: 600 roles opened against a staff of 2,000. Read alone, the headcount describes a company almost standing still. Together they describe a company that replaced about 15% of its staff, grew by 3%, and ended the year with twice the open demand it began with.
The fill assumption carries the result, so vary it. At 50%, hires are 260 and separations 200, or 9.9% of average headcount. At 90%, hires are 468 and separations 408, or 20.1%. The reading holds across the range, since most hiring was replacement, and the level of turnover is known only within that band. As a test of plausibility, set the implied rate beside an official one: 15.0% a year is 1.25% a month, against a total separations rate of 3.2% for US nonfarm establishments in August 2026. A negative figure, or one several times the official rate, says an assumption has failed: an acquisition, hires made without a posting, or postings that each stood for many positions.
How to use job postings and headcount together
- Date both series by what they refer to. Take each stated count on the date it refers to, and end the posting window on that date.
- Scale postings by headcount. Open postings over headcount plus open postings gives an intensity that compares a company of 200 with one of 20,000.
- Separate the flows. New postings, closed postings and the net of the two answer different questions. Keep postings that were already published when a company was first observed apart from new ones.
- Reconcile with the identity. Solve for separations under a stated fill assumption, then vary the assumption.
- Look for moves that carry no posting. Acquisitions, disposals and layoffs change headcount directly, so read the company's announcements between the two dates.
- Treat the months after the last count as a hypothesis. Postings are the one measure of the two that moves between counts. Write down what the next count should show if your reading is right.
- For a back-test, use what could be known on each day. That is the first-seen date of a posting and the publication date of a count, which is what point-in-time data means in practice.
How Fokals delivers both
Fokals holds both sides on one company index. In the hiring dataset, Job Postings holds every role a company publishes, labelled by job function, seniority and ten role flags, with location, work mode and advertised pay on one annual US-dollar scale. Hiring Activity holds each company's open, new and closed postings for every day, written once after the day closes. Employee Headcount, part of the announcements and scale dataset, holds stated headcount over time: each count as stated, on the date it refers to.
All three carry the same stable company ID, so step 2 is a join on a key, and every observation is dated. Company News, delivered beside Employee Headcount, classifies company announcements and regulatory disclosures into 13 event types, which serves step 5. The data dictionary defines every field. The hiring dataset is refreshed daily, and all of it is delivered by REST API or as bulk files.
Frequently asked questions
Is a job posting the same as a job opening?
No. In the definition of the Bureau of Labor Statistics, a job opening is a position with work available, that could start within 30 days, for which the employer is actively recruiting from outside the establishment, counted on the last business day of the month. A posting is one way of recruiting. An opening can exist with no posting, a posting can stay online after the role is filled, and one posting can stand for several positions.
Does JOLTS count online job postings?
No. The Job Openings and Labor Turnover Survey collects its figures directly from a sample of approximately 21,000 nonfarm business and government establishments, which report the positions open on the last business day of the month. A BLS article of August 2026 states that the survey's method does not rely on online job listings. Posting an internet notice is one of several forms of active recruiting, and active recruiting is one of the three conditions of an open position.
What is the difference between job openings, hires and employment in BLS data?
Job openings are unfilled positions on the last business day of a month. Hires are all additions to payrolls during the month, and separations are all departures from them. Employment is the number of persons on establishment payrolls in the pay period that includes the 12th. For August 2026, BLS reported 7.1 million job openings, 5.2 million hires and 5.1 million total separations at US nonfarm establishments, beside payroll employment of 159.0 million, all seasonally adjusted and preliminary.
Why does headcount stay flat when a company is posting many jobs?
Often because hiring replaces people who leave. Headcount shows the net of hires and separations, while postings follow gross demand: across US nonfarm establishments in August 2026, BLS reported 5.2 million hires beside 5.1 million total separations. In the illustrative example above, a company of 2,000 hires 364 people in a year, loses 304 and grows by 60, or 3%, after opening 600 roles. Postings left unfilled, filled by transfer or withdrawn widen the gap.
Can you estimate a company's headcount from its job postings?
Postings alone give the direction of demand, and a stated count gives the level. A headcount depends on the hires that resulted from postings and on the departures of the same period, and a posting records neither. With a stated count as the base you can estimate where the next one will land: apply a fill rate to closed postings, subtract an assumed rate of separations, and treat the result as a hypothesis that the next stated count tests.
How often is company headcount data updated?
As often as the company states its headcount. A count given in an annual report arrives once a year: under SEC rules a Form 10-K is due 60 to 90 days after the fiscal year ends, depending on the filer, so the count can be two to three months old when published. Fokals Employee Headcount holds stated headcount over time, each count on the date it refers to, and Hiring Activity supplies the daily movement of postings between counts.
The queries and code on this page are examples to adapt. Test them in your own environment before you rely on them.
What this page says about the products it names was checked against their public documentation on 7 October 2026. Product and company names are trademarks of their owners. Fokals is not affiliated with them or endorsed by them.