Analysis of WFH Research's September 2026 SWAA update, distinguishing the July share of paid full workdays performed at home from the 12-wave distribution of full-time employees across on-site, hybrid and fully remote arrangements. Includes survey methodology, education-group annual averages and limits on interpretation.
About 26% of paid full workdays among the survey's covered US population were worked from home in July 2026, according to WFH Research's September update, the team behind the Survey of Working Arrangements and Attitudes (SWAA) — economists Jose Maria Barrero, Nicholas Bloom, Shelby Buckman and Steven J. Davis. That figure answers a narrow, specific question: for each day last week, did a respondent work a full shift of six hours or more, and if so, from where? It comes from what people say about their own workweek on an online panel survey, not from badge swipes, employer records or job postings — a distinction worth holding onto before treating it as a verdict on offices generally.
The team's September 2026 update is dated September 1, 2026, and lists August 2026 as the latest survey wave included. The document includes the August 2026 survey wave, but its headline work-from-home-days chart explicitly labels its reading July 2026. This article therefore identifies the 26% figure as the July reading rather than treating it as an August estimate. The underlying data can also be browsed directly on the project's public data portal.
Share of paid full workdays (6 or more hours) worked from home in the US, July 2026, per WFH Research's SWAA survey.
How the survey defines a work-from-home day
SWAA has run monthly online since May 2020 and has now collected more than 200,000 cumulative observations. Its target population is US residents aged 20 to 64 who earned at least $10,000 in the prior year — working-age adults with some existing connection to paid work, not every US adult. Respondents are recruited through market-research firms using panel aggregators such as Lucid, then the raw responses are re-weighted to match 2010-2019 Current Population Survey worker shares across age, sex, education and earnings cells, so the reported figures are weighted survey estimates rather than simple raw averages.
Quality control removes a meaningful share of respondents before any figure is calculated: the researchers exclude roughly 16% as "speeders" and report that roughly another 12% fail attention checks. The report gives a median response time of 7 to 12 minutes after excluding speeders. These checks can reduce some low-quality responses, but they do not turn a voluntary online panel into an administrative record or eliminate sampling bias.
How full-time employees actually split between remote, hybrid and on-site
The cleanest breakdown in the update covers full-time (5 or more days a week) wage-and-salary employees over the twelve SWAA waves from September 2025 through August 2026, with 39,121 respondents in that group. Over that year, 62% worked full-time on-site with no remote days, 26% worked a hybrid mix of in-office and remote days, and 12% were fully remote with no in-office days at all.
Full-time (5+ days/week) wage-and-salary employees only, N=39,121. Source: WFH Research, SWAA, September 2025–August 2026 waves.
Why the two headline percentages differ: the 26% figure measures the share of paid full workdays performed at home in a monthly survey reading. The 12% / 26% / 62% split classifies full-time wage-and-salary employees by their work arrangement during a reference week, pooled across twelve survey waves. A person who works from home two days and on-site three days belongs to the hybrid group but contributes only two home days to a five-day workweek. The two measures have different denominators, populations and time windows, so they should not be treated as interchangeable.
38% were hybrid or fully remote in the survey reference week (BestCoworking calculation)
WFH Research reports the 12% fully-remote and 26% hybrid shares separately; it doesn't publish them as a single combined figure. Adding them together — 12% plus 26% equals 38% — gives the share of full-time wage-and-salary employees whose reported reference-week arrangement was hybrid or fully remote, across the September 2025 to August 2026 survey waves. It does not mean that 38% of employees worked remotely at some point during that entire year; the classification is based on each respondent's reference week. This is BestCoworking's own arithmetic on the source's reported shares, not a number WFH Research itself stated.
Where the variation actually shows up
The headline split hides differences by industry, education and age. The following comparisons refer to the groups and periods specified by WFH Research, not to all US adults. The same September 2026 update includes several findings at the slide-title level that are safe to cite without attaching a more specific number than the source itself gave:
Concentrated by industry
WFH Research reports that working from home is most prevalent in the finance, tech and professional and business services sectors — the update doesn't attach a specific percentage to this claim, so it's presented here as a directional finding rather than a number.
Tied to education
About half of full-time wage-and-salary employees with a four-year college or graduate degree work from home at least one day a week, per the source's own reported figure. Lower-education groups work from home far less often, though the update doesn't give a comparable single percentage for those groups in the material reviewed here.
Split by age, not gradual
Workers in their 50s and 60s are fully on-site and fully remote more often than younger workers, according to the update — a bimodal pattern at the older end of the workforce rather than a smooth decline in remote work with age.
Employers want less remote work than employees do
Employer plans for work-from-home have hovered around 1.3 to 1.5 days per week and have tracked actual work-from-home levels fairly closely since mid-2022, the update reports. But employers still offer fewer fully remote jobs and more fully on-site jobs than employees say they want, and the gap between how much employees want to work from home and how much they actually do fluctuates near half a day.
The source also notes that work-from-home patterns broken out by education group have been stabilizing since roughly 2023, after several years of larger swings earlier in the pandemic recovery. None of these findings come with a specific percentage attached beyond what's quoted above, so none is presented here with an invented number bolted on.
How working from home varies by education
WFH Research also publishes annual averages of the share of full paid workdays performed at home by education group. The 2026 values below cover the months available when the September update was released, not the full calendar year. This series includes people who worked at least one full day in the reference week, including self-employed and contract workers, so its population differs from the full-time employee arrangement chart above.
| Education group | 2025 | 2026 to date* |
|---|---|---|
| High school degree or less | 20.8% | 19.1% |
| 1–3 years of college | 23.9% | 23.1% |
| Four-year college degree | 33.2% | 30.8% |
| Graduate degree | 34.5% | 32.7% |
*2026 figures average the months available at the time of the September 1, 2026 report. Source: WFH Research, SWAA September 2026 Updates, education-group annual averages table.
The table shows a higher share of full paid workdays at home among degree holders in both periods. It does not establish that education itself causes remote-work access; job type, industry and other differences may also matter.
What this survey can't tell you
Limits worth keeping in mind
- This is self-reported survey data about where someone worked, not badge, fob or building-access data — it measures a different thing than occupancy trackers that count physical entries into specific buildings, and the two shouldn't be read as confirming or contradicting each other.
- The sampled population excludes anyone without recent earnings and anyone who doesn't use a computer, smartphone or similar device to take an online survey.
- Before re-weighting, the sample under-represents less-educated workers, particularly those without a high school diploma, and may still skew toward more tech-savvy respondents within that group even after re-weighting.
- Nothing here says anything about demand for coworking space, private offices or meeting rooms specifically — the survey measures where paid work happens, not why, or whether workers who work from home are using or would consider a flexible workspace instead.
These limitations do not negate the value of a long-running monthly survey with a disclosed methodology, but they matter when interpreting its results. It's simply a self-reported measure of a specific working-age, previously-earning population, not a census of every US worker. Anyone weighing their own remote or hybrid setup, including whether a coworking space, private office or meeting room fits better than working from home some days, can browse current listings and contact operators directly on BestCoworking rather than treating any single national survey figure as a decision for them.
Key takeaways
Workday share and employee arrangement share measure different things. Survey estimates are not building occupancy or coworking demand. Interpret the July figure and September 2025–August 2026 employee split using their distinct populations and periods.Primary source: WFH Research, SWAA September 2026 Updates, dated September 1, 2026, including the July 2026 work-from-home-days chart, September 2025–August 2026 employee arrangement chart and education-group annual averages table. WFH Research data portal. The 38% combined share is BestCoworking arithmetic (12% + 26%); it refers to reference-week classifications, not the share of workers who ever worked from home over the full year.
PRE-PUBLISH CHECKS: Confirm slug availability; preview wr-bar styling and table on mobile; click-test US directory link; confirm OG image inheritance.