What 67,324 job postings look like from the inside.

Longbow exists to prepare people for specific jobs, and doing that means watching a very large number of job postings very closely. This is the first of a series about what we see when we step back and look at all of them at once.

It is an introduction, with data. Where the postings come from, what they say, how we count them, what we will never publish, and three findings already clear enough to preview before they get their own issues.

Where the postings come from

When an employer hires, the job goes up on their own careers page first. Those pages run on hiring systems, the software behind the Apply button, and that is where most of our feed comes from: 58,539 of the 67,324 postings in this report, across 53 systems. The other 8,785 are job-board listings with no exact twin on an employer site.

It is a wide net rather than a deep one. 57% of the 20,017 employers in this report have exactly one active posting. The ten largest employers together account for 9.2% of postings, and one of them, a national pizza chain whose stores each post their own openings, accounts for 4.7% on its own. When a finding could be moved by a single employer like that, we say so or we set them aside.

Paylocity carries a fifth of the feed and is the shallowest system per employer: 5,349 employer names, fewer than three postings each. SmartRecruiters is the opposite, 253 employers posting 21 jobs each. The system an employer uses says a lot about their size, which matters later when we compare how those systems behave.

What a posting tells us

Every posting in the feed carries the same set of fields: title, employer, location, whether the job is on-site, hybrid or remote, the skills the posting names, the benefits it mentions, the pay range when one is given, and two dates, the day the employer posted it and the day we first saw it. Our own pipeline then reads the posting text and records what the employer actually wrote about experience, degrees, seniority and pay period.

Two-thirds of postings are on-site. Of the 62,820 postings that state a work arrangement (93% of the report), 69.6% are on-site, 17.2% hybrid and 13.2% remote. Remote work is concentrated in a few kinds of work.

Employers state a minimum experience in 36.8% of postings, and the median minimum is three years. Software engineering, engineering, HR, marketing and product postings typically ask for five; customer service, administrative and medical support postings ask for two; registered nursing asks for one. A degree is explicitly required in 20.6% of postings, from 33% of finance advisory postings down to 14% of software engineering postings.

Postings are specific about skills. 95.6% list at least one, and the median posting names 12. Communication is the single most-listed skill, named in 17.4% of all postings, ahead of any technical skill. SQL, the most-listed technical skill, appears in 8.6% and Python in 7.9%. Six in ten postings mention at least one benefit.

Three things the postings already told us

These are previews. Each becomes a full issue with its own charts, and each already meets the rules in the methodology block below.

The office is where entry-level work happens. Among the 24,786 postings that state a minimum experience, those asking for under two years are 77.9% on-site and 8.9% remote. Those asking for five to ten years are 53.4% on-site and 19.3% remote. The ladder flattens at the top: ten-plus-year postings are 57.7% on-site. Hybrid is more common in senior-labeled postings (31%) than entry-labeled postings (14%), but does not rise steadily across every seniority label. Entry-labeled postings are 13% remote, senior-labeled postings 20% remote. Issue 02: The Remote Work Ladder.

Pay is stated twice as often where a pay-range law is in force. Our pay field was incomplete for postings dated June 22 to late August while a fix was applied, so this preview uses only postings dated May 11 to June 21, 2026, when the field was complete. In that window, 64.5% of postings located in a state with a pay-range law stated a pay figure (n = 10,126). In states without one, 30.3% did (n = 13,390). That is a description, not a verdict: we have not yet tested what changed when each law took effect. Issue 03: The Pay Transparency Gap.

Employers list people skills more than tools. Communication, customer service and problem solving are the three most-listed skills in the feed. The first technical skill, SQL, comes fourth. Issue 05: What Employers Actually List.

How we count

The short version, so nobody has to take our word for it. The long version, with every definition, lives on one page and every report links to it: About Longbow Research.

• A dated snapshot. This issue counts the postings active in our feed on September 4, 2026. Their employer posting dates run from January 20 to September 4, 2026, and 99.8% are dated May 11 or later.
• One universe.Employer postings in the United States, exact duplicates removed, employer names resolved to one company record. Job-board listings with an exact twin on the employer’s site are removed as duplicates; the 8,785 without one are counted like any other posting.
• Honest denominators. Every percentage says what it is a share of. Remote shares are of postings with a stated arrangement (93%). Experience shares are of postings that state a minimum (37%).
• Machine-read fields, disclosed. Work arrangement, industry and listed skills come from the data vendor. Role family, seniority, years required, degree requirement and pay period are extracted from the posting text by our own pipeline.
• No group under 300 postings. Every group in this issue is in the thousands.
• Concentration, disclosed. One employer is 4.7% of the report. Where a single employer could move a finding, we split or we say so.
• Gaps, disclosed. Our structured pay field was incomplete for postings dated June 22 to late August 2026. Every pay figure in this issue is measured on the weeks it was complete, and we say so where it appears.
• What the data cannot prove. A posting is what an employer wrote, not what happened next. We will never tell you a relisted posting is a ghost job, because the feed cannot see whether anyone was hired.

What we never publish

Longbow Research is built from postings employers published for anyone to read. Nothing our users type, upload or report is in it. Not resumes, not the jobs people are pursuing, not interview outcomes, not conversations with Robin.

If user data ever enters a report, three rules already written into the product apply first. Nothing individual, ever. No company named below five contributing people. No pay figure below ten contributing people, and only in $5,000 bands. Full rules on About Longbow Research.

What’s coming

• Issue 02 · The Remote Work Ladder. Remote, hybrid and on-site by stated experience, seniority, role family and state. Published September 2026: The office is where careers start.
• Issue 03 · The Pay Transparency Gap.Pay disclosure by state, law versus no law, and what changed when Virginia’s law took effect. Published September 2026: Who tells you what a job pays, and who doesn’t.
• Issue 04 · The Benefits Gap. Dental cover, a retirement plan, paid time off and parental leave: how often postings name them, by state law, employer, pay and kind of work. Published September 2026: What comes with the job, and who tells you.
• Issue 05 · What Employers Actually List. The skills named in postings, by role family.

Each issue goes out as an edition of The Longbow Index on LinkedIn the day it publishes.