Open Offices: Not a Behavior Fix #

Key Result: Face-to-face interaction decreased by roughly 70% after employees moved to an open-plan office, while email and instant messaging increased (peer-reviewed field study). BS-0076

Background #

Few workplace interventions have been adopted as widely, or justified as confidently, as the open office. The logic sounds airtight: walls block conversation, so removing walls should produce more of it. Tear out the private offices, seat everyone within sight of everyone else, and spontaneous collaboration will bloom. Organizations bought this story at enormous scale because it promised a behavioral outcome - more knowledge sharing - from a purely architectural input: a new floor plan.

The strongest evidence tests exactly that promise, and it is the reason this case carries a “validated” confidence rating. Bernstein and Turban (2018) measured interaction objectively, before and after an open-plan adoption, rather than relying on surveys or managerial impressions. The finding inverted the sales pitch. Face-to-face interaction did not rise when the walls came down; it collapsed by roughly 70%, while email and instant-messaging volume went up. BS-0076 The environment change went beyond failing to produce the intended behavior. It produced the opposite one.

What actually drove the reversal #

The open office fails for a reason the blueprint never models: employees were not waiting for permission to talk. They were optimizing for something else entirely, focused work and privacy, and the redesign attacked both. When an environment conflicts with the behavior people are actually trying to perform, people do not surrender their goal. They adapt around the environment.

The target behavior’s fit profile exposed two constraints, while Capability Fit remained high:

• Disposition. Many knowledge workers characteristically prioritize focused output, autonomy, and control over interruption. A professional self-concept may reflect those priorities, but it does not define them.
• Capability. Knowledge workers generally possess the communication skills required for face-to-face collaboration; missing ability is not the bottleneck.
• Context. Constant interruption degrades deep work, while total visibility increases social risk. Every interaction now has an audience, so the cheapest way to communicate becomes the channel nobody can overhear: email and instant messaging.

That last point explains the signature pattern in the data. Communication did not disappear after the move to open plan; it migrated. Digital channels offered what the new floor plan had removed, privacy and control, so the behavior flowed there. Headphones, avoidance routines, and messaging apps are all versions of the same adaptation: people rebuilding, at their own expense, the walls the organization tore down.

The corrective is not a better slogan about collaboration. It is matching the environment to the behavior, the discipline of context engineering: private focus rooms, noise control, and privacy or hybrid options that let the space support focused work instead of fighting it (company-reported best practices).

Behavior Fit Assessment #

These ratings are analyst examples derived from the study’s mechanism, not direct fit measurements. For the target behavior, spontaneously collaborate more face-to-face, Dispositional Fit is low because many knowledge workers reliably prioritize focused individual output, autonomy, and control over interruption rather than visible sociability. A professional self-concept can express those characteristic priorities, but it is a secondary signal. Capability Fit is high because knowledge workers generally possess the communication skills required for face-to-face collaboration. Context Fit is low because constant interruptions, permanent visibility, low privacy, and workplace norms raise the cost of interaction and push communication into digital channels. Degraded focus is a consequence of that external environment, not evidence that people lack collaboration ability. A behavior with this profile does not appear just because the furniture changed.

Results #

• Face-to-face interaction decreased by ~70% after the move to an open office, while email and IM use increased (peer-reviewed, single-site before/after study). BS-0076

• Practical mitigations such as private focus rooms, noise reduction, and privacy or hybrid options realign the environment with focused-work behavior (company-reported best practices).

Limitations #

The ~70% decrease comes from a single-site before/after study (Bernstein & Turban 2018); the magnitude may vary with office design, team structure, and culture. Open-plan implementations also differ widely (hot-desking versus assigned seats, noise levels, density), so not every open office produces the same behavioral effects. Effects further depend on job type and the degree of openness, and some collaboration gains may occur in specific contexts.

Lessons #

1. Environments configure; they do not command. Changing a layout changes the constraints on behavior, not what people are trying to accomplish. Design must match the behavior people are optimizing for, not the outcome leadership wishes for.
2. Behavior adapts to protect its goal. Remove privacy and people manufacture substitutes - headphones, messaging, avoidance, remote work. Any intervention that ignores this adaptive response will measure the intended behavior falling and the workaround rising.
3. Audit before you architect. A behavioral audit asking “what are people actually optimizing for in this space?” would have surfaced the focus-and-privacy conflict before the walls came down, when it was still cheap to fix.

Sources #

• The impact of the ‘open’ workspace on human collaboration (Bernstein & Turban, 2018)
• Evidence Ledger: BS-0076