Companies that went fully remote during the pandemic have spent years wrestling with a question that once seemed simple: how often, if ever, do employees really need to be in the same room? Some firms have mandated a full return. Others have kept workers home permanently. A growing group is trying something in between, but until now there has been little rigorous evidence about how much in-person contact is actually needed to capture the benefits of an office.
A new NBER working paper tests a version of that question at the low end of the spectrum. It asks whether bringing fully remote employees together for a single coordinated day each month can meaningfully change how they work, how they interact, and whether they stay with the firm.
A minimal dose of the office
The study was conducted by Cevat Giray Aksoy of the European Bank for Reconstruction and Development and King’s College London, Nicholas Bloom of Stanford, Steven J. Davis of the Hoover Institution, Victoria Marino of the EBRD, and Cem Ozguzel of the Paris School of Economics. Working with a large multinational business-process outsourcing firm in Turkey, they designed a nine-month randomized controlled trial involving 248 customer-service employees who handle inbound calls for a telecommunications client.
All of the employees were working from home. The firm had kept remote work as its default since 2020, but managers reported weaker team cohesion, less coaching, and stubborn retention problems. The researchers wanted to know whether occasional coordinated in-person contact could restore some of what had been lost, without giving up the flexibility of remote work.
Of the 661 employees invited to participate, 476 volunteered. After filtering for tenure and living within roughly 45 minutes of the firm’s Şanlıurfa office, 248 eligible employees remained. They were split into two groups of 124 based on the parity of their pre-existing company ID numbers. One group continued working entirely from home. The other was required to come to the office together one fixed day each month, from August 2024 through April 2025.
The intervention was deliberately stripped down. Pay, working hours, workload, technology, incentives, and formal training were all left unchanged. Treated employees rode a company shuttle to the office, worked their normal shift on their normal laptops, and shared meals and coffee breaks with colleagues. That was it. Attendance ran at about 95 percent.
Productivity climbed slowly, then stuck
The firm records detailed operational data, and the researchers’ main productivity measure was calls handled per hour. In the two months before the experiment began, both groups averaged 10.8 calls per hour, essentially identical.
For the first several months of the intervention, the two groups tracked closely. Then a gap began to open. By months six through nine, the treated group was pulling ahead, and in the five months after the intervention ended, when treated employees had returned to fully remote work, they were handling 12.4 calls per hour compared with 11.5 in the control group. That is a 7.8 percent difference.
Some of that gap reflects composition: different people were leaving the two groups at different rates. When the researchers restrict the comparison to changes within the same worker over time, they still find a within-employee gain of about 0.63 calls per hour, or roughly 5.8 percent of the pre-treatment average. The remaining share of the gap appears to come from the treatment group retaining more of its higher performers.
Call duration got shorter, on average by about 18 seconds by the end of the study, while time spent on administrative work, holds, and breaks did not change appreciably. Customer satisfaction ratings and internal audit scores also held steady, suggesting the faster pace did not come at the expense of service quality.
The gradual buildup matters for interpretation. If the productivity gains were driven simply by novelty, excitement about being picked, or a temporary morale bump, the authors would expect them to appear immediately and then fade. Instead, they emerged slowly and persisted after the office days stopped, a pattern the researchers interpret as more consistent with learning, feedback, and stronger workplace ties.
Who talked to whom
The team also wanted to see whether office days changed communication patterns after everyone went back home. They ran two tests.
First, in a survey at the end of the intervention, treated employees reported spending 36 more minutes per week communicating with their three most frequent colleagues than control employees did.
Second, the researchers randomized seating assignments during office visits. When two employees happened to be assigned as desk neighbors, they were 11 percentage points more likely to communicate with each other in the following week of remote work, compared with a baseline communication rate of about 9 percent. Pairs who happened to interact over a morning coffee break, lunch, or an afternoon break showed even larger increases in follow-up contact, though those interactions were not randomly assigned.
Survey responses also indicated that treated employees were more likely to say they received regular feedback from their manager, that team communication worked well, and that they felt they fit with the company’s culture. Measures of broader well-being, such as life satisfaction and work-life balance, did not show statistically significant treatment effects.
Fewer people walking out the door
Retention was the other headline outcome. Over the course of the experiment, cumulative attrition reached 21.0 percent in the control group but only 13.7 percent in the treatment group, a reduction of roughly a third. The gap remained visible after the last office day, suggesting the effect was not just a delay in separations.
A closer look at who was leaving reveals a notable pattern. In the fully remote control group, the employees who quit were actually somewhat more productive than those who stayed. In the treatment group, the opposite held: the stayers were substantially more productive than the quitters. The authors interpret this as evidence that monthly in-person contact helped managers identify and retain their stronger performers, a group otherwise prone to leaving.
The economics for the firm
The researchers offer a back-of-the-envelope cost-benefit calculation from the company’s perspective. Direct costs, including transport, meals, refreshments, and office space, ran about $13,000 over nine months for the 124 treated employees. Reduced attrition saved roughly $6,400 by avoiding an estimated nine separations, each costing about $714 to replace. The productivity gains, valued at employer labor costs, added roughly $60,300.
Total benefits came to about $66,700, or roughly five times the program’s costs.
Caveats worth keeping in mind
The study was conducted in one specific setting: a highly standardized inbound customer-service operation where calls are routed centrally, tasks are uniform, and output is easy to measure. Whether similar effects would appear in more open-ended knowledge work, or in firms with different cultures and different pre-remote histories, is not something this trial can answer.
The eligible sample also excluded people who declined to volunteer, most of whom cited family or care responsibilities. And while the design pins down the effect of coordinated in-person contact, it cannot separate how much of the improvement came from time with the boss, time with peers over lunch, or something else entirely. The authors are careful to note that they identify a combined effect, and that their evidence on the underlying channels is suggestive rather than definitive.
