To increase worker performance, employers often invest in a number of things, from rewards
and incentives to education and training. These traditional approaches develop employees’
skills and enrich their work experience. But we discovered a surprisingly simple way to increase
productivity, one that was low-cost and had immediate impact: better office seating
arrangements.
Research we conducted suggests that who an employee sits next to affects how they perform
— and grouping the right types of coworkers together can improve productivity and work
quality.
We analyzed two years’ worth of data on more than 2,000 employees of a large technology
company with several locations in the U.S. and Europe. (The company is a client of Cornerstone
OnDemand, which one of us, Jason, works for.) We created unique identifiers for each worker
after merging five different data sources:
- A master file that contained employee data, such as hire and termination dates, job position, compensation, and direct managers
- Two engagement surveys conducted across the organization
- Monthly reports on each employee’s location and assigned cubicle over time
- Building maps and floor plans to calculate the distance between each cubicle on a given floor
- Worker performance data broken down into three metrics:
- Productivity. We measured the average length of time it took a worker to complete a task. For any given worker in this company, tasks were fairly similar and occured regularly.
- Effectiveness. We measured the average daily rate at which a worker needed to refer a task to a different worker to solve. This occurred when the employee couldn’t resolve the task on their own and it was assigned to another.
- Quality. We measured the client’s satisfaction with the task on a five-point scale
For every performance measure, we looked at “spillover,” a measure of the impact that office
neighbors had on an employee’s performance. Assume a worker has three coworkers: one sits
next to her, one sits 25 feet away, and another sits 50 feet away. We looked at the
performance of the three coworkers along with their distance from the worker, and through
various data modeling techniques we measured the average spillover of their performance on
the worker.
We saw that neighbors have a significant impact on an employee’s performance, and it can be
either positive or negative. In terms of magnitude, we found that approximately 10% of a
worker’s performance spills over to her neighbors. Replacing an average performer with one
who is twice as productive results in his or her neighboring workers increasing their own
productivity by about 10%, on average.
One unique feature of our data is that workers were randomly assigned to teams and to desks,
and they were periodically moved around in a quasi-random way due to the demand and
supply of workers. A centralized human resource department sent workers to different
locations based on the essentially random flow of new workers arriving at a particular time.
This suggests that the effects of neighbor location on performance may be causal.
We categorized workers into three types: productive workers, who completed tasks quickly but
lacked quality; quality workers, who produced superior work but did so slowly; and generalists,
who were average across both dimensions. In our sample, 25% of people were productive
workers, 25% were quality workers, and the remaining 50% were generalists.
Pair People with Opposite Strengths
In our sample, where groups of workers were clustered together, we found that the best
seating arrangements had productive and quality employees sitting beside each other, because
each helped the other improve. There was a spillover effect on both workers’ areas of
weakness: A quality worker tried to match the speed of a productive worker, while the
productive worker tried to improve their work quality. When productive workers were seated
next to quality workers (and generalists were grouped together), we found a 13% gain in
productivity (speed of work) and a 17% gain in effectiveness (fewer unresolved tasks) in that
group.
On the other hand, seating two productive workers together did not significantly increase their
productivity — nor did placing quality workers together increase their work quality. And since
generalists were average in both categories, they were less affected by spillover effects from
either side.
An interesting finding, for this particular technology firm, is that workers who were strong on
one dimension (task quality or task speed) tended not to be sensitive to spillover on that
dimension, while workers who were weak on that dimension were sensitive to spillover. This is
why putting a fast worker next to a slow worker tends to speed up the slow worker instead of
slowing down the fast worker.
Separating Toxic Workers
Toxic workers in our sample were employees who ended up being terminated for reasons
related to toxic behavior, which included misconduct, workplace violence, drug or alcohol
abuse, sexual harassment, falsification of documents, fraud, and other violations of company
policy. Toxic workers negatively influenced their neighbors’ performance.
If toxic employees were near each other, it increased the probability that one of them would be
terminated by 27%. But in contrast to productivity and quality spillover, any type of worker
seemed susceptible to toxic spillover. If a toxic worker sat next to a nontoxic worker, the toxic
worker’s influence won out, and the nontoxic worker had an increased chance of becoming
toxic. This suggests that companies should pay close attention to employee engagement
surveys to understand how employees feel about their work environment. Surveys can root out
toxicity by providing an early warning for managers and HR to intervene.
What Drove These Spillover Effects?
We found that these effects occurred almost immediately but vanished within two months.
(Performance was measured daily or weekly, depending on the metric of interest, and then
averaged by month.) This suggests that, instead of employees learning from one another, which
would likely take some time, the effects were driven by a combination of inspiration and/or
peer pressure from sitting near high-performing workers. Of course, we could not distinguish
which factors truly drove the effect.
Our study leads us to believe that better spatial management of workers can enhance
individual and team performance. But managers need to first look at employees’ performance
and see where they would want spillover to occur. We estimate that a strategic seating chart
could bring in $1 million in annual profit from greater productivity for an organization of 2,000
workers.
To be sure, different organizations will have different tasks and different kinds of spillover; the
optimal seating arrangement for the firm we studied may not be the best for all firms. But it
illustrates how office design can influence performance. Once an organization identifies which
spillovers exist, management can plan the space of the organization to produce better
outcomes. In this way, physical space, which companies can manage relatively inexpensively,
can be an important business resource.