By Karel de Bruin. Updated ; first published .
Five years after offices emptied, the numbers have settled. Most knowledge workers are hybrid, not remote. Fully remote work costs some productivity on average, while two days at home costs none. And the offices people come back to are barely half full on their busiest days. Here is what the research says, and why it points to two or three managed office days a week.
In May 2026, about a quarter of all paid working days in the United States were worked from home1. Before the pandemic the figure was 5 to 7 percent2. Among full-time employees, 12 percent are now fully remote, 62 percent fully on site and 26 percent hybrid1. Look only at people whose jobs can be done remotely and the picture flips: Gallup finds 51 percent hybrid, 21 percent fully on site and the rest fully remote, with hybrid workers spending 2.3 days a week in the office3.
Europe moved less far but in the same direction. Across the EU, the share of workers doing any work from home rose from 14 percent in 2019 to 23 percent in 2025. Almost all of that growth is people who work from home "sometimes" rather than "usually"4. In Great Britain, 28 percent of working adults were hybrid in early 2025, rising to 41 percent among graduates5. Worldwide, graduate employees average about 1.2 days a week at home, down from 1.5 in 2022 and flat since 20232.
Employer plans have converged on the same point. Since mid-2022, the number of home days companies plan for has hovered around 1.3 to 1.5 a week, and has matched what employees actually do1. The trend has stopped trending. Hybrid is the settled state.
The most careful study to date is a randomised trial at Trip.com, published in Nature in 2024. 1,612 engineers, marketers and finance staff were assigned either to five days in the office or to two days a week at home for six months. Over the two years of performance reviews that followed, hybrid working had no effect on grades, no effect on promotions and no effect on the lines of code engineers wrote. What it did change was retention: resignations fell by a third, and job satisfaction rose. The managers involved had expected hybrid to cut productivity by 2.6 percent; after the trial they estimated it had raised it by 1 percent6.
Fully remote work looks different. At a large call centre, moving on-site workers home cut the calls they handled by about 4 percent and lowered call quality, with the worst effects among inexperienced staff7. Among 10,000 IT professionals at an Asian services firm, hours worked went up, including 18 percent more time outside normal working hours, while output slipped, so productivity fell by 8 to 19 percent. Meetings and coordination ate the hours that used to be uninterrupted work, and there was less coaching and fewer one-to-ones8.
The economists who run the largest ongoing survey of working arrangements sum it up plainly: studies of hybrid arrangements often find productivity gains or no discernible effect, while several studies find that fully remote work yields lower productivity than on-site work9. The 2015 finding this article once led with, that call-centre staff who could choose to work from home became 22 percent more productive, still stands as an early result. The 2024 trial is the better evidence, and it says two days at home is free.
Where fully remote work does the most damage is to people at the start of their careers. A study of software engineers at a Fortune 500 firm found that sitting near teammates increased the feedback engineers received on their code by 18 percent and improved its quality. The gains were concentrated among younger and less experienced staff. Experienced engineers, meanwhile, wrote less code when they sat near their teams, because they were spending the time mentoring10.
The same authors show that, in jobs that can be done remotely, unemployment among young graduates has risen relative to older workers since the pandemic10. Office days matter most for onboarding and mentoring. That is an argument for hybrid work that is designed around when teams overlap, not for five days a week and not for none.
Americans value the option to work from home two or three days a week at about 8 percent of pay911. Six in ten employees in remote-capable jobs want a hybrid arrangement, about a third would prefer fully remote, and fewer than one in ten want to be on site every day3. Asked how many office days they would choose, hybrid workers most often say three (32 percent) or two (25 percent)12.
Take the flexibility away and people leave. In a 2025 survey of 2,000 US workers, 40 percent said they would start looking for another job if remote flexibility were removed, and a further 22 percent would expect a pay rise to compensate12. Gartner found that strict return-to-office mandates lower intent to stay by 8 percent on average, and by 16 percent among high performers13. A mandate is a pay cut in disguise, and the best people are the first to notice.
The last two years have been a natural experiment. Amazon required five days a week from January 202514; Microsoft moved to three days for staff near an office from early 202615. Across the Fortune 100, 67 companies are hybrid and 29 are full time in the office, and the average number of days required by policy rose 10 percent in a year. Actual office attendance rose 2 to 3 percent16.
The clearest evidence on outcomes comes from a study of three million tech and finance employees at S&P 500 firms. After a return-to-office mandate, turnover rose by 14 percent, the time to fill a vacancy grew by 23 percent and the hiring rate fell by 17 percent. Departures were concentrated among senior, skilled and female employees. The same firms showed no improvement in return on assets or market value17. Mandates mostly cost talent.
The office that people return to is a shared resource, used unevenly. Sensor data from about 63,000 workspaces puts average utilisation at 43 percent in 2025 and 45 percent in early 2026, against 60 to 65 percent before the pandemic. Tuesdays and Wednesdays peak at just over 50 percent; Fridays run at 30 percent. On any given day a quarter of desks are not used at all, and two thirds are used for less than three hours18.
Occupiers see the same thing from the other side. CBRE's European survey reports 46 percent average weekly utilisation with a 71 percent peak day, and 55 percent of companies expect to reduce their office footprint, citing hybrid work first and cost second19. JLL's benchmark tracks global utilisation from 61 percent before the pandemic to 49 percent in 2024 and 56 percent in 202620. And 69 percent of CBRE's clients now say more than 40 percent of their people share desks21.
Put those together and the office of 2026 is smaller, shared, and full on two days and empty on one. Which is why the question has moved from "should we let people work from home?" to "how do we run a building that most people use two or three days a week?"
The conclusion this article reached in 2021 has held up, and it now rests on a randomised trial rather than on surveys. Two or three days in the office a week keeps performance and promotions where they were, cuts resignations by a third, protects the mentoring that junior staff depend on, and is worth about 8 percent of salary to the people doing it. What has changed is the office itself: with utilisation near half and a quarter of desks untouched on any day, seats have to be shared, and sharing has to be organised.
That is the job a desk booking system does. Colleagues see who is in on which day and book a desk near their team; managers see which days fill up and how much space they actually need; and nobody arrives to find their desk taken. Deskdragon adds that to Microsoft 365 or Google Workspace in an afternoon, with your own floor plans. See how desk booking works.
1 Barrero, Bloom and Davis (2026), Survey of Working Arrangements and Attitudes, June 2026 update, WFH Research, https://wfhresearch.com/wp-content/uploads/2026/06/WFHResearch_updates_June2026.pdf
2 Aksoy, Barrero, Bloom, Davis, Dolls and Zarate (2025), The Global Persistence of Work from Home, Global Survey of Working Arrangements, https://wfhresearch.com/wp-content/uploads/2025/06/Global-WFH-2025-PNAS_final_v1.pdf
3 Pendell (2025), Hybrid Work in Retreat? Barely., Gallup, 2 September 2025, https://www.gallup.com/workplace/694361/hybrid-work-retreat-barely.aspx
4 Eurostat, Employed persons working from home as a percentage of total employment (lfsa_ehomp), EU27, 2019 to 2025, https://ec.europa.eu/eurostat/databrowser/view/lfsa_ehomp/default/table?lang=en
5 Office for National Statistics (2025), Who has access to hybrid work in Great Britain?, 11 June 2025, https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/articles/whohasaccesstohybridworkingreatbritain/2025-06-11
6 Bloom, Han and Liang (2024), Hybrid working from home improves retention without damaging performance, Nature 630, 920 to 925, https://www.nature.com/articles/s41586-024-07500-2
7 Emanuel and Harrington (2024), Working Remotely? Selection, Treatment, and the Market for Remote Work, American Economic Journal: Applied Economics 16(4), https://www.aeaweb.org/articles?id=10.1257/app.20230376
8 Gibbs, Mengel and Siemroth (2023), Work from Home and Productivity: Evidence from Personnel and Analytics Data on IT Professionals, Journal of Political Economy Microeconomics 1(1), https://www.journals.uchicago.edu/doi/abs/10.1086/721803
9 Barrero, Bloom and Davis (2023), The Evolution of Work from Home, Journal of Economic Perspectives 37(4), https://www.aeaweb.org/articles?id=10.1257/jep.37.4.23
10 Emanuel, Harrington and Pallais (2023, revised 2026), The Power of Proximity to Coworkers, NBER Working Paper 31880, https://www.nber.org/papers/w31880
11 Barrero, Bloom and Davis (2021), Why Working from Home Will Stick, NBER Working Paper 28731, https://www.nber.org/system/files/working_papers/w28731/w28731.pdf
12 Owl Labs (2025), State of Hybrid Work 2025, survey of 2,000 US full-time workers, July 2025, https://owllabs.com/state-of-hybrid-work/2025
13 Gartner (2024), Gartner HR Research Finds High Performers, Women and Millennials Are Greatest Flight Risks Under Strict RTO Mandates, 30 January 2024, https://www.gartner.com/en/newsroom/press-releases/2024-01-30-gartner-hr-research-finds-high-performers-women-millennials-are-greatest-flight-risks
14 Jassy (2024), Strengthening our culture and teams, Amazon, https://www.aboutamazon.com/news/company-news/ceo-andy-jassy-latest-update-on-amazon-return-to-office-manager-team-ratio
15 Coleman (2025), Flexible work update, Microsoft, 9 September 2025, https://blogs.microsoft.com/blog/2025/09/09/flexible-work-update/
16 Flex Index (2025), 67% of the Fortune 100 are hybrid, 22 July 2025, https://flexindex.substack.com/p/67-of-the-fortune-100-are-hybrid
17 Ding, Jin, Ma, Xing and Yang (2024), Return-to-Office Mandates and Brain Drain, SSRN 5031481, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5031481
18 XY Sense (2026), Workplace Utilization Index, Q4 2025 and Q1 2026, https://xysense.com/workplace-utilization-index/
19 CBRE (2025), European Office Occupier Sentiment Survey 2025, 1 October 2025, https://www.cbre.com/insights/reports/european-office-occupier-sentiment-survey-2025
20 JLL (2026), Global Occupancy Planning Benchmark Report 2026, 19 May 2026, https://www.jll.com/en-us/insights/occupancy-benchmark-report
21 CBRE (2026), 2026 Global Workplace and Occupancy Insights, 20 January 2026, https://www.cbre.com/insights/articles/the-hybrid-reality-why-the-office-is-more-important-than-ever