A New York Fed survey finds 48.8% of workers had no in-person meetings on their most recent office day, and more than half of those with meetings held at least one online while everyone was in the same building. Mandates are filling desks, not meeting rooms.
Key Takeaways
Return-to-office policies rest on a simple promise: bring people into the same building and they will collaborate face to face. New research from the Federal Reserve Bank of New York suggests that promise is often not being kept. On a typical office day, millions of workers commute in, sit down, put on a headset, and spend the day on video calls and chat with colleagues who are somewhere else, or sometimes just down the hall.
The New York Fed staff report “Remote Work from the Office”, by economists Natalia Emanuel and Emma Harrington, is built on an original survey of 1,835 full-time, college-educated American workers conducted in September 2026, after Labor Day. The authors call that timing conservative, since many companies push hardest for attendance after the summer.
Even so, the findings are stark. On their most recent day in the office, 70.6% of respondents were missing at least one member of their team, and on average they were missing 34.3% of it. 48.8% had no in-person meetings at all, with teammates or anyone else, and more than a third had neither in-person nor hybrid meetings. 15.1% reported no casual face-to-face conversation with anyone that day. The authors summarize it plainly: roughly half of in-office days are spent entirely on Zoom calls or doing solo work.
Proprietary badge-swipe data from a Fortune 500 company tells the same story. Under a four-day mandate, 72.8% of employees who came in had teammates out of the office, and those present saw only 58.5% of their team. Counting both people working from home and people who commuted in only to sit apart from their teammates, 84.3% of employees had a remote-work experience on a given day. Time-use data points the same way: the typical worker spent 7.1 hours working in person with colleagues on an office day before the pandemic, and 6.3 hours by 2026.
The report identifies four causes, and only one of them is about attendance. The biggest is what the authors call organizational sprawl: teams whose members are fully remote or assigned to other offices. 37.0% of survey respondents had at least one teammate in that position. At the Fortune 500 company, around 60% of corporate employees had fully remote or distant teammates in 2026, up from 22% before the pandemic. A team in that situation can comply perfectly with its in-office policy and still never meet face to face.
Compliance is the second factor. Respondents’ companies required 3.74 office days a week on average, while respondents reported coming in 2.59. At the Fortune 500 company, 63.6% of employees complied with its four-day policy. Coordination is the third: 35.2% of respondents with an in-office mandate said it let workers choose their own days, which leaves teams to line up schedules on their own. The fourth is habit. 45% said their organization follows a “One Zoom, All Zoom” norm, moving an entire meeting online the moment one person dials in. And among people with any meetings that day, 52.4% had at least one that was needlessly remote, held on video even though every participant was in the same building. Screen-sharing is one of the reasons people give for doing it.
None of this has slowed the push to bring people back. CBRE’s third-quarter U.S. office market data, reported October 27, shows 89% of employers now require at least three in-office days a week, up from 78% a year earlier. Overall vacancy fell to 18.2%, its fifth straight quarterly decline, and net absorption has been positive for ten consecutive quarters. “The main benefit to being in the office is that other people are in the office,” said Stefan Weiss, head of CBRE’s U.S. office research. The Fed data suggests that for many teams, that benefit is exactly what is missing.
The meetings themselves carry their own friction. Owl Labs’ 2026 State of Hybrid Work, a survey of 2,000 full-time U.S. knowledge workers, puts in-office attendance at 65%, up from 63% a year earlier. Yet getting a meeting started takes an average of six minutes, 18% say it regularly takes more than ten, and 66% lose meeting time to technical difficulties. 61% say they feel disengaged in meetings. More office days layered on the same meeting habits produce more commuting, not more collaboration.
The Fed research does not argue against the office. It argues that the office only works when the people who need each other are in it at the same time and actually meet. Companies that keep adding mandated days without fixing sprawl, scheduling and meeting habits are paying for the commute and the real estate while their teams keep talking through a screen.

Report
Half of office days are spent on video calls and solo work, so commuting in does not fix a meeting problem. This report shows how chat platforms and back-to-back meetings fragment focus, and what to change before adding more office days.
Download
Report
With 60% of corporate employees at one large company now on teams spread across offices or fully remote, a single headquarters no longer brings everyone together. This report examines the resurgence in coworking demand and what it means for leasing strategy.
Download
Whitepaper
Screen-sharing is one reason colleagues in the same building still meet over video. This whitepaper looks at the hidden cost behind the material on that screen: the hours teams spend building and reformatting presentations.
Download
Fidelity’s five-day mandate took effect in September with one carve-out: phone-based customer service. The exception says a lot about what attendance rules actually measure.
Gartner finds 63% of procurement leaders expect AI to significantly improve performance within three years, yet only 25% have redesigned jobs or roles because of it. Middle-skilled work is under the most pressure.
A study of 236 sales reps at Stripe and Checkr finds managers gave higher-value work to 44% of top-quartile reps after AI adoption, versus 12% of the bottom quartile. The study points to AI literacy, not talent, as the deciding factor.