The Open Space That Emptied the Office Instead of Filling It: The Story of a Failure

Two corporate headquarters agreed to something unusual: before tearing down their walls, they let researchers measure exactly how people worked. For three weeks, employees across sales, human resources, finance, and technology wore sociometric badges and microphones — devices that logged, second by second, who talked to whom, for how long, and how loudly. Then the companies carried out the renovation everyone had approved: private offices and cubicles came down, open, transparency-enhancing floor plans went up. Months later, the same employees put the badges back on for another three weeks. The comparison that followed, published by Ethan Bernstein of Harvard Business School and Stephen Turban of Harvard University in the Philosophical Transactions of the Royal Society B, is one of the few instances where a workplace redesign’s central promise was tested with instruments instead of opinions. The open office had been built to bring people together. The badges recorded the opposite.

What the data actually showed

Face-to-face interaction fell by roughly 70 percent in both companies after the move to open architecture — in the first case study, a 72 percent drop, alongside a 56 percent rise in email volume and a 67 percent increase in instant messaging. The second study tracked something more granular: 1,830 specific pairs of colleagues, comparing how much each pair interacted before and after the redesign. Of those pairs, 643 reduced their face-to-face contact once the walls came down. Only 141 increased it. The mechanism wasn’t ambiguous — it showed up at the level of individual working relationships, not just aggregate office noise.

The researchers’ explanation runs against the design brief that gets pitched to facility managers every time an open-plan renovation is proposed. Removing spatial boundaries was supposed to increase visibility, and visibility was supposed to increase spontaneous collaboration. Instead, constant visibility triggered what the authors describe as a natural human response to social withdrawal. When every conversation can be overheard and every desk can be seen, people don’t talk more — they put on headphones, avoid eye contact, and move the conversation somewhere it can’t be observed: email, instant messaging, a channel with a recorded, controllable, asynchronous footprint. The badges could measure the retreat because it was physical. The productivity cost was harder to see, but the companies involved reported it declining anyway, consistent with the idea that information carried face-to-face — tone, nuance, the follow-up question — doesn’t survive the move to a text thread.

The failure had already been measured elsewhere

Bernstein and Turban’s study is the most direct evidence because it used instrumentation rather than self-report, but it confirms a pattern that shows up wherever anyone has actually measured open-plan offices instead of assuming they work. A University of Sydney analysis of 42,764 workplace surveys collected across 303 office buildings in the United States, Finland, Australia, and Canada — data gathered by UC Berkeley’s Center for the Built Environment — found open-plan layouts carrying the highest dissatisfaction rates of any office type measured, driven overwhelmingly by noise and lost privacy: roughly half of workers in partitioned open-plan seating and about half in unpartitioned layouts reported active dissatisfaction with sound privacy specifically. A separate 2008 field study led by researcher Annu Haapakangas found open-office employees losing an average of 21.5 minutes per day to conversational distraction alone, with speech — not equipment hum, not traffic noise — identified by 48 percent of respondents as the single most disruptive sound in the room. The U.S. General Services Administration’s own survey of 3,700 federal employees found 60 percent saying they could get more done in a quieter workspace. None of these are anecdotes about people disliking change. They are independent datasets converging on the same failure mode: a design intended to increase interaction instead increased the incentive to disappear from it.

What this means for how space actually gets designed

The lesson isn’t that open plan is inherently wrong — later acoustic research complicates that reading. A body of work measuring 21 real open-plan offices found the share of highly disturbed employees ranging from 6 percent to 71 percent depending on the building, which means outcomes are set by specific, measurable conditions — sound-masking, layout density, distance between desks — and not fixed by the open-plan concept itself. That is the more useful and more uncomfortable finding for anyone managing space today: removing walls is not a strategy, it’s a starting condition, and whether it succeeds or fails depends on variables most renovation budgets never measure before or after the fact.

Bernstein and Turban’s badges are the reminder worth keeping. The two companies in that study didn’t fail because open plan is a bad idea in the abstract. They failed because they redesigned the office around a hypothesis about human behavior — that visibility breeds collaboration — and never tested it against what people actually do when they’re watched. The office that was supposed to fill up with conversation emptied out into inboxes instead, and the only reason anyone could prove it is that, for once, someone was measuring before they started tearing down walls.

SOURCES

Ethan S. Bernstein and Stephen Turban, “The Impact of the ‘Open’ Workspace on Human Collaboration,” Philosophical Transactions of the Royal Society B (Harvard Business School / Harvard University); University of Sydney analysis of Center for the Built Environment (UC Berkeley) workplace survey data (Jungsoo Kim et al., Journal of Environmental Psychology); Annu Haapakangas et al., field study on noise and distraction in open-plan offices; U.S. General Services Administration federal workplace survey.

Workplace Insider workplaceinsider.eu