Discussions about working from home always begin with impressions. Some worry that working from home feels comfortable and therefore might lead to people becoming more lax. Others believe that because it is quiet, concentration will improve. Because everyone’s experience is different, these debates rarely reach an easy conclusion. What companies truly want to know is not how it feels, but what the results are: whether mixing in remote work lowers performance, and whether people stay longer.

These two questions are directly tied to organizational costs. When one person leaves, it is not simply that a seat becomes empty. The sequence and context of the work that person knew, and even the unspoken agreements a team shared, leave with them. Filling that gap takes time and money. In the end, the work-from-home debate is not only about productivity, but also about retention.

What pushed this debate forward was not a slogan but an experiment. The Trip.com study published in Nature drew attention because it tested a two-days-a-week remote schedule inside a real company and then verified the outcomes through a randomized controlled experiment.

The researchers ran the experiment for six months across 2021 and 2022, involving 1,612 employees. The setup was highly realistic. It included typical knowledge-work roles that require both office collaboration and individual focus, such as software engineering, marketing, finance, and accounting.

Most important was how people were divided. If those who wanted remote work were simply allowed to choose it, the results would become blurred. People who already had high autonomy might have preferred remote work more, and people who already performed well might have used remote work more effectively. In that case, it becomes difficult to separate whether remote work caused the effect or whether people’s characteristics produced the outcome.

So the researchers split employees into groups using a rule that was unrelated to personal preference. One group was allowed to work from home two days a week, while the other group maintained the existing five-days-a-week office schedule. Because of this design, it became much clearer where the changes came from, whether positive or negative.

The biggest change appeared in turnover. The resignation metric in the hybrid group was about one third lower than in the comparison group. In the paper, it is presented as an average of 7.20 for the comparison group and 4.80 for the hybrid group. This change did not appear equally for everyone; it showed up more strongly in certain groups.

Reports that turnover reductions were significant among non-managers, women, and employees with long commutes show who gained the most value from the policy. What the company reduced was not merely dissatisfaction, but the cost of vacancies. When fewer people leave, the time and money spent on hiring and training go down. An even larger benefit comes from the fact that a team’s pace does not keep getting interrupted. When stability improves not only day to day but quarter to quarter, the same headcount can go farther.

So what about performance. Contrary to common fears, hybrid work did not clearly reduce performance.

The researchers reported that over the subsequent two years, hybrid work did not operate as a disadvantage in performance evaluation ratings, and they present conclusions suggesting that no clear difference was found in promotions either. For some groups where output is easier to quantify due to job characteristics, they separately checked work output indicators and concluded that it was difficult to find signals that hybrid work was disadvantageous there as well. The meaning of this section is simple. At least in this company’s knowledge-work environment, two days of working from home did not provide evidence supporting the fear of productivity loss.

Changes in satisfaction add not emotion but reasons to this conclusion. According to the study, job satisfaction in the hybrid group rose from an average of 7.84 in the comparison group to 8.19. The reasons employees cited were not flashy. Commute time and costs fell, there was flexibility to handle occasional personal needs during the day, and any shortfall could be made up in the evenings or on weekends.

In other words, hybrid work is less a system that makes people work less and more a system that changes how work is scheduled. When a rhythm emerges in which some days are for deep focus, some for coordination, and some for recovery, people can endure for longer.

Another interesting scene appears here as well. Managers’ beliefs changed. The study reports that 395 managers re-evaluated the impact hybrid work would have on productivity before and after the experiment, and that on average, what they had viewed as minus 2.6 percent before the experiment was revised to plus 1.0 percent after the experiment.

More important than the numbers themselves is the direction. When hybrid work wobbles, the first thing that wobbles is not the work, but trust. When distrust grows, people try to prove their presence rather than their results. Meetings increase, reporting becomes overcrowded, and the time spent on real work shrinks. Even a small shift in belief can make an organization quieter, and a quieter organization finds it easier to produce performance again.

The message of this experiment is not a simple verdict that remote work is good or bad. It is that the structural advantages of a two-days-a-week rhythm become visible. Fully remote work requires replacing almost everything the office used to do with different methods. From communication rules to onboarding and training new hires, everything must be rebuilt.

Hybrid work, by contrast, leaves the office inside the rhythm. Work that benefits from being together can be gathered on office days, and work that requires deep focus can be placed on remote days. Instead of collaboration and concentration fighting each other, it becomes possible to divide roles by day of the week. This structure reduces the large friction cost of commuting while preventing the complete loss of the relationship and learning functions that offices used to provide.

However, applying this conclusion to every organization without adjustment can lead to misunderstandings. This experiment is a result from a particular company, a particular period, and a particular job mix. Roles where customer response is broken into minute-by-minute pieces and roles centered on creative problem solving have different rhythms.

Organizations with high interdependence across teams and organizations with a high share of individual tasks are also different. The key is not to copy the number two days a week, but to understand the conditions that made that number work.

Hybrid work is not permission, but operations. If meetings increase on remote days, the benefit of focus disappears. If there is no feedback and mentoring on office days, the learning speed of new hires can fall. If evaluation standards are vague, visibility competition grows, and the advantages of remote work are swallowed by anxiety. Hybrid work is not a policy that changes space, but a task of rearranging how work runs. If the arrangement is done well, two days a week of working from home can become a design that helps people stay without undermining performance.

Bloom, Nicholas; Han, Ruobing; Liang, James. (2024). Hybrid working from home improves retention without damaging performance. Nature