The Open-Plan Experiment

AFor most of the period in which employers were dismantling internal walls, the case for doing so rested on a promise that nobody had troubled to check. Removing partitions, it was said, would multiply the unplanned encounters from which good ideas are supposed to come, and the firms that moved fastest reported exactly that: forms returned by staff in the first months after a refurbishment showed people describing their new floor as friendlier and more open. What such forms record, however, is how an environment feels rather than what happens in it, and the two can diverge sharply. Nobody was counting conversations. A second reason for the promise's long life is that it was never the only reason for the change. Floor space is among the largest fixed costs an employer carries, and a plan without corridors or private rooms accommodates roughly a third more people on the same footprint, so a decision taken for the accountant could be presented as a decision taken for the culture. Between 1998 and 2014, when conversions were at their peak, almost no employer commissioned a direct record of who actually spoke to whom, and the promise passed into received wisdom by default.

BThat gap was closed, unexpectedly, by a technology developed for something else. Camilla Bergström, of the Halden Institute of Work Studies, fitted 312 employees at two large firms with badges that logged proximity and speech without recording content, and ran them for eight weeks before a move to an open floor and eight weeks afterwards. Face-to-face exchange fell by 41 per cent; electronic messaging between the same pairs of people rose by 63 per cent, and the volume of email sent to colleagues seated within ten metres nearly doubled. Bergström argues that the result is not a paradox but a straightforward consequence of exposure: a conversation held in the open can be overheard by everyone, so the cost of starting one rises, and people shift the exchange to a channel over which they keep some control. Her data have since been reproduced at four further sites. What they have not done is settle the question of why, because a link between removing walls and talking less is compatible with more than one story about the mechanism.

CDevraj Menon, an organisational psychologist, maintains that the story Bergström tells attributes to the plan of the room what belongs to something the room merely made visible. Employees withdraw, he contends, not because a space lacks walls but because it offers nowhere else to go; where an employer installs bookable rooms for concentrated work and a norm that using them is unremarkable, the withdrawal does not occur. Menon followed two software companies for eighteen months after conversion and found that face-to-face contact, having dropped in the first quarter by an amount close to Bergström's figure, returned to within 6 per cent of its previous level once such rooms were available. He is franker than most critics about the limits of this. Both firms had volunteered for the study, both operated in an industry with unusually strong habits of scheduled discussion, and the recovery was measured only in teams whose managers had themselves used the rooms. Bergström objects that a remedy which works when three conditions are met is not a refutation of a problem that arises when they are not.

DA finding on which the two agree has attracted far less attention than their disagreement, though it may matter more to anyone actually choosing a floor plan. The effect of an open layout is not uniform across the workforce. Staff whose output requires long uninterrupted stretches, such as writing, auditing or analysis, lose measurably: error rates in proofreading tasks rise by about a fifth when the same people are moved from enclosed rooms to an open floor, and the loss persists after the novelty of the new setting has worn off. Nor is the split a matter of taste, since the same individuals show the pattern whenever they are moved between settings, which rules out the idea that people simply gravitate towards whatever suits them. Staff whose work is chiefly coordination, by contrast, gain: supervisors in Bergström's sample settled queries in less than half the time once they could see the people they were settling them for. An arrangement that suits the second group therefore imposes a cost on the first, which is why a single floor plan applied to a whole building will always be a compromise rather than an optimum.

EThe argument has moved, then, from whether open plans work to what has to accompany them, and that is a more useful place for it to be. Neither party now defends the original promise that removing walls will by itself increase the exchange of ideas; the evidence assembled since 2016 points the other way, and the disagreement concerns how much of the damage the walls themselves are responsible for. Menon's position implies that the layout is cheap to fix and expensive to get wrong, since the remedy is a small number of rooms and a rule about using them. Bergström's implies that employers who provide neither will find that their savings on floor space have been recovered from somewhere less easy to see. What has changed is the standard of proof: an employer who converts a floor is now expected to say what will be measured afterwards, and very few can. Both would accept that the least defensible option is the one still most common. It is to convert, to celebrate the openness, and then to measure nothing at all.