Remote work productivity: what recent studies say about home, hybrid and office work

Few workplace debates have produced as many confident headlines as the fight over where people should work. Chief executives cite falling output to justify return to office mandates. Employees point to saved commuting hours and quieter days. Both sides usually quote a study. The problem is that they are often quoting different studies about very different jobs.

I have spent the past few years reading this research while covering startups that went fully distributed, then partly walked it back. The pattern I keep seeing is that the evidence is more consistent than the arguments built on top of it. This article walks through what the better known studies found, why their results seem to conflict, and what that means for managers and workers making decisions now.

What does remote work productivity actually mean?

Before comparing results, it helps to pin down the terms, because studies use them differently.

• Fully remote work means an employee works from home or another location of their choosing nearly all the time, with rare or no office visits.
• Hybrid work means a fixed or flexible split between home and office, most commonly two or three days at home each week.
• Office work means the employee is expected on site for the full working week.

Productivity is the harder term. Economists usually mean output per hour worked. Many workplace surveys instead measure how productive people feel, which is a different thing. Company studies often track narrower metrics such as calls answered, lines of code merged, tickets closed or performance review scores. A study that measures calls per shift in a call center cannot tell you much about a product design team, and a survey of how people feel cannot tell you much about output per hour.

Keep that distinction in mind. Much of the public confusion about remote work productivity comes from treating these measures as interchangeable.

What have the major studies found?

The most useful studies are those where researchers could compare similar workers doing similar tasks, ideally with random assignment. Here are several that are frequently cited, summarised in plain terms. If you plan to rely on any of them for a business decision, read the original paper, since methods and caveats matter.

The Ctrip call center experiment

In one of the earliest randomized trials, economist Nicholas Bloom and coauthors studied call center staff at Ctrip, a large Chinese travel company, in the early 2010s. Volunteers were randomly assigned to work from home or stay in the office. The home group performed about 13 percent better, mostly because they took fewer breaks and sick days and worked in a quieter setting. Attrition also fell. However, many home workers later asked to return to the office, citing loneliness, and promotion rates for home workers were lower.

The Trip.com hybrid trial

A decade later, Bloom and colleagues ran a larger randomized trial at the same company, now called Trip.com, published in the journal Nature in 2024. This time the subjects included engineers, marketers and finance staff, and the treatment was hybrid: two days a week at home. The researchers found no measurable difference in performance reviews or promotions over two years, while quit rates fell by roughly a third. For this kind of white collar work, hybrid looked roughly neutral on output and positive on retention.

Indian IT and data entry studies

Other research points the opposite way. A study by Michael Gibbs, Friederike Mengel and Christoph Siemroth of a large Indian IT services firm during the pandemic found that employees logged more hours at home, but output did not rise to match, so productivity per hour fell. The authors linked part of the decline to more time spent in meetings and coordination.

A separate field experiment in India led by David Atkin, Antoinette Schoar and Sumit Shinde assigned data entry workers to home or office. Office workers were more productive, by roughly 18 percent in the published findings, and the study also found that workers who preferred working from home tended to be less productive in either setting.

Proximity, mentoring and collaboration

Some of the most interesting findings are not about raw output. Research by Natalia Emanuel, Emma Harrington and Amanda Pallais on software engineers at a large company found that engineers sitting near their teammates received more feedback on their code, which especially helped junior staff. Senior engineers, however, produced less of their own work when they spent time mentoring. A Microsoft study of its own employees, published in Nature Human Behaviour in 2021, found that firm wide remote work made collaboration more siloed, with fewer connections across groups and more reliance on scheduled communication.

Why do the studies seem to disagree?

Read together, these results are less contradictory than they appear. Several factors explain most of the gap.

1. Type of task. Independent, measurable work such as answering calls can thrive at home if the environment is quiet. Work that depends on fast coordination or tacit knowledge tends to suffer more.
2. Fully remote vs hybrid. Many negative results involve fully remote setups. Hybrid trials more often show neutral effects, because people still get regular face time.
3. Selection. Who chooses remote work matters. If less productive workers prefer home, comparing remote and office staff will exaggerate the cost of remote work unless the study uses random assignment.
4. Home conditions. A quiet room and reliable internet are not universal. Studies in places with crowded housing or unstable power can show larger losses.
5. Pandemic distortion. Research from 2020 and 2021 captured remote work under school closures, illness and stress. Those results may not describe remote work in normal conditions.
6. What gets measured. Short term output is easier to count than learning, innovation or loyalty. A setup can look neutral on monthly output while costing or gaining a great deal over years.

Is hybrid work the compromise most companies need?

For many office based jobs, the current evidence leans toward yes, with conditions. The Trip.com trial is the strongest single piece of support: performance held steady, and lower attrition is a real saving, since replacing an employee typically costs a meaningful share of their annual salary once recruiting and training are counted.

Hybrid works best when the office days are coordinated. If each person picks different days, people commute in only to sit on video calls with colleagues who stayed home. Teams that agree on shared anchor days get the mentoring and spontaneous conversation that justify the trip, while keeping focused work for home days.

Hybrid is not a universal answer. Customer support, data processing and some engineering roles can run well fully remote with good tooling and clear metrics. Roles that rely on hands on training, lab equipment or tight physical collaboration may need people on site most of the week.

Who gains and who loses from remote work?

Averages hide the people at the edges, and this is where the research is most consistent.

Junior employees

New graduates and early career staff appear to lose the most from fully remote setups. They benefit from overhearing how experienced colleagues handle problems and from quick, informal feedback. The software engineering research suggests this effect is real and measurable, not just nostalgia for the office.

Experienced specialists

Senior staff with established networks and clear tasks often gain focus at home. The tradeoff is that their absence can reduce the mentoring juniors receive, which is a cost to the organisation rather than to the individual.

Caregivers and long distance commuters

People with caring duties or long commutes often value flexibility highly. Lower quit rates in hybrid trials suggest that flexibility functions partly as a benefit that keeps skilled people from leaving.

Cross team innovation

The Microsoft findings point to a quieter risk: when everyone is remote, people talk mostly to their own team. Over time, that can slow the flow of ideas between groups, which matters for product companies and startups that depend on new combinations of knowledge.

What should managers and workers do with this evidence?

The research supports practical steps more than ideological positions. Based on these studies and on what I have seen at companies of different sizes, here is a sensible approach.

• Match the policy to the job. Group roles by how much they depend on coordination and hands on learning, and set different expectations for each group rather than one rule for everyone.
• Measure output, not presence. Badge swipes and online status lights say little about productivity. Agree on a few clear outcome measures for each team.
• Protect juniors. If you allow remote work, build structured mentoring, regular code or work reviews and planned in person time for new hires.
• Coordinate office days. Shared anchor days make commuting worthwhile.
• Run your own trial. Many companies can test a policy on a few teams for six months and compare results before rolling it out. A rough internal experiment beats borrowing conclusions from a different industry.
• Count retention. Include quit rates and hiring costs in any assessment, since these can outweigh small changes in output.

For workers, the lesson is similar. If you are early in your career, time with experienced colleagues is an investment, even if home is more comfortable. If you are experienced and productive remotely, keep visible records of your output, because the Ctrip findings on promotion suggest that being out of sight can still carry a career cost.

The bottom line on remote work productivity

The honest summary is that remote work productivity is not a single number. Fully remote work helps some tasks and hurts others. Hybrid work, when coordinated, appears to keep output roughly level while improving retention. The largest risks sit in areas that are slow to show up in quarterly metrics: training new people, building networks across teams and generating new ideas. Companies that design around those risks, rather than arguing about averages, are likely to get the best of both settings.