Remote Work From the Office: 48.8% of Office Days Have No In-Person Meeting. Mandates Reach 37% of the Cause.
A New York Fed paper splits the lost face-to-face contact into three causes. Each has a different owner, and only one responds to an attendance rule.
A New York Fed staff report published in October names something most hybrid employees already recognise: remote work from the office. Economists Natalia Emanuel and Emma Harrington surveyed 1,835 college-educated, full-time American workers in September 2026 and asked about their most recent office day. 48.8% had no in-person meeting with anyone. 70.6% were missing at least one teammate. 36.7% took at least one video call in which every participant was in the same building.
Those headline figures have been widely reported. The part that should change how a company sets policy is further down: the authors split the lost face-to-face contact by cause. Teams spread across locations account for 48% of it. Colleagues who stay home account for 37%. The lever most leadership teams reach for, one more mandated day, acts on the second cause and does nothing for the first.
What the paper measured
The authors recruited 2,422 respondents through the Prolific panel and kept 1,835 after quality checks. Questions about the most recent office day were limited to people whose last visit was a weekday within the past week, and team questions to teams of 15 or fewer. They paired the survey with personnel and badge-swipe records from one unnamed Fortune 500 company, time diaries from the American Time Use Survey, and job-posting and résumé data covering the Fortune 500.
The survey ran after Labor Day, when US attendance pushes are at their strongest, so the authors regard the numbers as conservative. On that most recent office day, respondents were missing 34.3% of their team on average. More than a third had neither an in-person nor a hybrid meeting. 15.1% had no casual face-to-face conversation with anyone at all.
The time diaries point the same way. Before the pandemic, a typical office day included 7.1 hours working in the physical company of colleagues. In 2026 the figure is 6.3 hours, which the authors calculate as a 59.1% increase in office time spent alone.
Remote work from the office is as common as working from home
The badge data lets the authors compare two groups on the same days: people working from home, and people in the office with at least one teammate somewhere else. Under a four-day mandate, 42.3% of employees were at home on a given Tuesday to Thursday. Add the people who commuted in to an incomplete team and 84.3% had a remote experience of some kind. The two groups are the same size.
The company ran two-, three- and four-day mandates for different parts of the organisation, and the paper reports both figures for each. From those you can work out what share of the people who did come in were missing a teammate. The last column below is our arithmetic; for the four-day rule the paper states the same result, 72.8%.
| Mandate | Working from home | Home, or in office without full team | Of those in the office, missing a teammate |
|---|---|---|---|
| Two days | 79.7% | 95.1% | 76% |
| Three days | 53.5% | 81.1% | 59% |
| Four days | 42.3% | 84.3% | 73% |
Going from three days to four put more people in the building: 57.7% on a given midweek day, up from 46.5%. It did not make the people already there more likely to find their whole team. The mandates covered different segments at different times, so this is not a clean experiment. It is still a caution against assuming that presence and contact rise together.
Three causes, three different owners
The paper names several contributors and puts a share on two of them. They are worth separating because no single person in a company controls all three.
| Cause | Share of lost contact | Evidence in the paper | Who can change it |
|---|---|---|---|
| Organisational sprawl: teammates assigned to other offices or fully remote | 48% | 37.0% of respondents have at least one distant teammate. At the Fortune 500 firm about 60% do, up from 22% before the pandemic. | Whoever approves hiring locations and team composition |
| Colleagues not attending | 37% | Policies ask for 3.74 days on average; workers report 2.59. Badge compliance with the four-day rule was 63.6%. | Whoever sets and enforces the attendance policy |
| Uncoordinated days and needless video | Not separately quantified | 35.2% of mandates let people pick their own days. 36.8% of meetings on office days were on video with everyone on site. | Team leads |
Sprawl is the cause that attendance policy cannot touch. If a team of six has two people in another city and one fully remote, every team meeting is a video call on every day of the week, with perfect compliance. The people who do commute sit at a desk wearing headphones. The authors expect mandates applied to such teams to draw more pushback, which is a polite way of describing a rule that imposes the commute and withholds the benefit.
Sprawl is also the result of decisions that looked sensible one at a time. In 2015, 0.67% of Fortune 500 job postings listed more than one location. In 2026, 4.17% do, with an average of 3.69 locations each. The paper gives the example of an engineering role open to candidates in Denver, Detroit or Dallas when the group sits mostly in Denver. Each location-flexible hire widens the talent pool and adds one more team that can never be whole in one room.
“A mandate buys presence. Only team design and coordination buy proximity.”
Why one more day disappoints: the arithmetic of a whole team
Take a team that can be co-located, so sprawl is not the issue. For you to have your whole team around you, every one of your teammates has to be in on the same day. If each of them shows up with probability p, independently, the chance that all n of them are present is p multiplied by itself n times.
With five teammates and the 63.6% badge compliance the paper reports, that chance is about 10%. At 80% attendance it is 33%. At 90% it is 59%, and at 95% it is 77%. With seven teammates, 90% attendance gives you a full team less than half the time.
Real attendance is not independent. People come in when they expect colleagues to be there, which is why the survey finds 29.4% of workers with a full team and not 10%. The shape of the curve is the lesson. Whole-team presence responds steeply to how reliably each person attends on a given day and to team size, and barely at all to the count of days in the policy. Two fixed days that 95% of the team honours will produce more full-team days than four floating days honoured at 64%.
The requirement is "everyone", not "most", because of a meeting convention. 45% of respondents said their organisation moves the entire meeting to video once a single participant needs to dial in, a practice the authors call One Zoom, All Zoom. It is a considerate norm for the remote attendee. It also means one absence converts the day for everybody else.
What to check before changing the policy
- Count the teams that can be whole in one building. Group people by reporting line and by assigned office. For any team that spans sites, no attendance rule produces in-person meetings. The honest options are to re-form teams around sites as roles open up, or to drop the commute requirement for that team and pay for periodic gatherings. The paper cites research linking occasional face-to-face socialising with lower attrition.
- Measure overlap, not badge swipes. A swipe count tells you how many people were in the building. The useful number is, per team, the share of office days on which every member was present. If you already hold badge data, this is one query.
- Fix the days before adding days. More than a third of mandates leave the choice of days to the individual, which hands each team a coordination problem with no owner. Team-level anchor days cost nothing and raise the per-day reliability that the arithmetic above depends on.
- Decide the single-dial-in rule on purpose. Keeping One Zoom, All Zoom is defensible on a sprawled team. On a co-located team with anchor days, a hybrid room for the one person who is out preserves the in-person meeting for the rest.
- Book the room. More than a third of meetings on office days were on video with every participant on site, usually because joining from a desk is the default in the calendar invite. Make a room the default for meetings where all invitees share an office.
Presence is countable, proximity is the point
The authors close on the distinction between presence and proximity, and it explains a good deal of the frustration on both sides of the return-to-office argument. Executives look at badge counts and see a policy working. Employees look at a day of video calls taken from a hot desk and see a commute with no purpose. Both are describing the same data correctly.
Badge data, calendar data and the org chart are enough to measure team overlap directly, and most companies with an office already hold all three. The next attendance policies worth copying will probably be written around that number, and they may well ask for fewer days than the ones in force now.
Frequently asked questions
Related reading
Junior Developer Hiring Is Down 20%, 25%, 50%, 67% or 73%. Each Number Measures Something Different.
Junior developer hiring statistics range from 20% to 73% declines. They use different populations, baselines and units. Here is what each one actually counts.
Deepfake Interview Fraud Is Up 1,300%. In-Person Hiring Won't Fix It.
Recruiters are going back to in-person final rounds to stop deepfake interview fraud. That fixes the wrong stage: the fraud that costs money completes at onboarding and payroll, not on the video call.
The AI Notetaker Lawsuits Aren't About Recording. They're About Who Gets Told.
A 2026 wave of lawsuits against AI meeting notetakers turns on two design choices, not on recording itself: whether the bot visibly joins the call, and whether transcripts train models afterwards.