The office is where careers start.

We wanted to know who actually gets to work from home, so we took the 33,833 US job postings active in our feed this week and sorted them by what employers ask for. The clearest pattern is experience. When a posting asks for under two years of experience, the job is on-site about 80% of the time. When it asks for five to ten years, that drops to 54%, and both hybrid and remote roughly double.

Issue 01mentioned this in passing. Here we go through it properly: how the mix of on-site, hybrid and remote changes with the experience a posting asks for, why job titles give a different answer than years do, what happens when you look inside a single profession, which kinds of work tend to be remote and which tend to be hybrid, how often “remote” means the job is actually remote, and how the picture differs from state to state. One thing to flag first: before counting anything we took out eight employers whose postings are all remote and all identical across dozens of states. Section 06 explains who they are and how much they change the numbers.

How the mix changes with the experience asked for

About 37% of postings state a minimum number of years of experience. Among the 11,570 that state one and also say where the work happens, the arrangement shifts steadily as the requirement goes up. Postings asking for under two years are 79.9% on-site, 13.0% hybrid and 7.0% remote. Postings asking for two to five years are 65.3% on-site, 22.1% hybrid and 12.6% remote. At five to ten years the on-site share is down to 54.0%, hybrid is up to 28.3% and remote to 17.7%.

The pattern does not continue all the way up. Postings asking for ten or more years are 58.9% on-site, 25.7% hybrid and 15.5% remote, which is a step back toward the office compared with the five-to-ten band. Our reading is that the most experienced roles in the feed are mostly director, executive and practice-lead positions, and a fair share of those are expected to be present in person. So the most senior roles are not the most remote ones.

Job titles versus years of experience

Postings also carry a seniority word in the title, about 42% of them: entry, manager, lead, senior, director or executive. If you sort by that word instead of by years, you get a different picture. Postings labelled senior are the most remote group in the feed at 19.5%, and another 32.1% of them are hybrid, so more than half are at least partly away from the office. Postings labelled executive are 18.7% remote. But postings labelled manager, which are the largest labelled group at 6,028 postings with a stated arrangement, are 70.7% on-site and only 10.4% remote, barely above entry level.

The explanation is what “manager” usually means in a job feed. The biggest group of manager postings is general and location management, meaning store, branch and site managers: 1,187 postings, 86.7% of them on-site. Operations, HR, sales and property managers come next, and most of those roles involve running people or places that are in one location. A senior analyst or senior engineer does work that can travel. The person managing a building does not.

Entry-labelled postings are 7.0% remote and 22.1% hybrid, which is close to the under-two-years band. So the two ways of measuring agree at the bottom of the scale and disagree in the middle, where the title “manager” covers a lot of very different jobs.

Inside a single profession

There is a reasonable objection to all of this: maybe experience has nothing to do with it, and the pattern is just a mix effect. Junior postings are concentrated in retail, care and administration, which are on-site at every level, and senior postings are concentrated in software and analysis, which are remote at every level. To check, we looked inside single professions and split each into postings asking for under five years and postings asking for five or more. Our rule of not publishing any group under 300 postings leaves three professions where both halves are large enough.

In sales, postings asking for five or more years are 26.5% remote compared with 17.1% for under five, and 56.8% on-site compared with 69.9%. In engineering and architecture, remote goes from 6.5% to 11.2% and on-site from 75.1% to 65.9%. So in both of those the experience pattern holds inside the profession, not just across professions. Data and analytics is different: 18.9% remote under five years, 20.7% at five and over, and about 45% on-site in both bands. Our interpretation is that in data work the arrangement has largely been settled at the profession level, so experience makes little difference to it, while in sales and engineering it has not, and experience is one of the things that decides it.

Which kinds of work are remote, and which are hybrid

Remote work is concentrated in a few kinds of work. Software engineering postings are 30.0% remote and product management 27.6%. After those two the shares fall away: data and analytics 19.6%, creative and design 17.1%, customer service 16.1%, project management 15.2%, sales 14.6%, IT 14.4%. Fourteen of the 26 families large enough to publish are under 10% remote, and the hands-on professions are close to zero: medical support 2.9%, education 2.2%, skilled trades 2.2%, therapy 1.2%, registered nursing 0.6%.

Hybrid work has a different set of leaders. Creative, design and media postings are 43.0% hybrid, the highest of any family, with only 39.9% on-site. Product management is 37.6% hybrid, data and analytics 33.5%, finance advisory 31.9%, legal and compliance 30.9%. Software engineering, the most remote family, is only 25.8% hybrid. When software work leaves the office it tends to leave entirely, whereas creative, legal and finance work tends to leave for part of the week.

Remote-only, or remote allowed

The feed records two kinds of remote posting: ones where the job is remote, and ones where remote is allowed. Of the 3,247 remote postings in the analysis, 38.7% are remote-only and 61.3% are remote-allowed. Interestingly, the split moves in the opposite direction to the remote share itself. Among remote postings asking for under five years, 38.5% are remote-only. At five years and over, only 27.6% are. So the more experience a remote posting asks for, the more likely it is that remote is being offered as an option rather than being how the job is set up.

One way to read this: for experienced hires, employers often seem willing to bend on location to get the person they want, without the job itself being designed as a remote job. At entry level there are far fewer remote postings, but the ones that exist are more often remote by design, such as customer service roles that were set up to be done from home. We cannot see the employer’s reasoning in the posting, so treat this as an interpretation of the pattern rather than something the data proves.

The eight employers we set aside, and why

Before counting anything above, we removed eight employers. Each has 30 or more active postings, every one of them remote, and not one stating a minimum experience. Six of the eight state no pay either. Each posts the same roles across 19 to 46 states at the same time, mostly insurance and benefits sales, customer service, and travel-agent roles that the data vendor files under hospitality. Together they add up to 721 postings, which is 2.1% of everything in the feed but 18.2% of every remote posting in it. The largest of the eight accounts for more than a quarter of those postings on its own.

We are setting them aside openly rather than quietly, because the choice changes numbers a reader would notice. With them in, remote is 12.2% of postings; without them it is 10.2%. Customer service goes from 26.0% remote to 16.1%, banking and insurance from 19.0% to 7.7%, and food service and hospitality from 26.2% to 2.7%, which is the figure you would expect for work that happens in a kitchen or a hotel. The experience pattern in section 01 does not change, because none of these postings states a requirement, so they all sit in the “not stated” group. Within that group, remote drops from 11.5% to 8.4% once they are out.

Remote and hybrid by state

Remote postings are spread thinly everywhere. Among the 25 states with 300 or more postings, the remote share runs from 3.7% in Wisconsin to 7.9% in Minnesota, with California at 7.8% and Texas at 5.9%. Another 32.1% of remote postings do not name a state at all, which makes sense for a remote job. Hybrid is where the state differences show up. New York postings are 28.3% hybrid, Massachusetts 23.5%, Illinois 22.3% and California 21.0%. Texas is 12.2%, Florida 12.6%, Ohio 11.2%, Tennessee 10.3% and Wisconsin 10.0%.

Most of this follows from which professions are concentrated where. States whose postings lean toward finance, law, media and analysis have more hybrid postings; states whose postings lean toward retail, logistics, care and manufacturing have more on-site ones. Texas has more postings than New York in this feed and less than half the hybrid share.

What postings state, and which ones

One caveat worth stating plainly. The experience pattern is measured on postings that state a minimum number of years, and stating one is itself uneven: 32.0% of on-site postings do, compared with 48.1% of hybrid and 48.0% of remote postings. Postings that state nothing are 77.5% on-site. So the pattern describes the postings that spell out their requirements, and on-site postings are the ones most likely not to.

Pay disclosure follows the same order. On postings dated August 31 or later, when our structured pay field is complete, 64.0% of hybrid postings state a pay figure, 53.8% of remote postings and 47.7% of on-site postings. Hybrid postings are concentrated in office professions and in states with pay-range laws, which probably explains most of that. We are not publishing pay levels in this issue; that is Issue 03, once the pay field has a full clean window behind it.

How we counted

The short version. 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 15, 2026. Their employer posting dates run from January 20 to September 15, 2026, and 99.0% are dated July 1 or later.
• One universe, with one exclusion. 34,554 employer postings in the United States, exact duplicates removed, employer names resolved to one company record. Eight employers (721 postings) are set aside under the rule in section 06; every figure is measured on the remaining 33,833 unless it says otherwise.
• Clear denominators. Every share of on-site, hybrid or remote is a share of postings with a stated arrangement (31,751, or 93.8%). Experience bands are shares of postings that state a minimum (37.2%). Seniority labels are shares of postings that carry one (41.8%).
• Machine-read fields, disclosed. Work arrangement comes from the data vendor, including its two remote values. Role family, seniority and years required are extracted from the posting text by our own pipeline.
• No group under 300 postings. That floor is why the within-profession chart shows three professions and two bands, and why the intern and mid-level labels are not charted.
• A repaired feed. This is the first issue on the repaired active universe. Issue 01 carries a dated correction explaining why its count was higher.
• What the data cannot tell us. A posting says where an employer wants the work done, not why. We describe the pattern; we do not claim that experience causes remote work, or that remote work is handed out as a reward.

Coming next. Issue 03, The Pay Transparency Gap: pay disclosure by state, law versus no law, once the pay field has a full clean window. Issue 04, What Employers Actually List: the skills named in postings, by role family. Each goes out as an edition of The Longbow Index on LinkedIn the day it publishes.