A coffee break can look like a small leak in the working day: phones unanswered, screens idle, minutes paid but apparently unproductive. The Swedish habit of fika begins from a different premise. A shared pause can be part of how work gets done rather than time stolen from it.

A well-known MIT workplace case offers a striking parallel. At a Bank of America call centre, teams whose breaks were coordinated so colleagues could talk became more productive. Later accounts reported lower stress, turnover falling from about 40 per cent to 12 per cent, and gains valued at roughly $15 million when the approach spread more widely.

This is one organisational case study, not settled consensus. It was not a study of fika or Sweden, its public accounts do not provide every statistical detail, and an American call centre cannot prove that communal breaks improve every workplace.

Fika is social time, not simply coffee

Fika is often translated as a coffee break, but the social element carries much of its meaning. Friends, relatives and colleagues pause together, commonly with coffee or tea and something small to eat. Our guide to Swedish coffee and fika culture describes it as a deliberate slowing down whose value lies partly in not doing it alone.

That does not make every Swedish workplace identical. Fika can be scheduled or spontaneous, brief or leisurely, and not every employee drinks coffee. Some teams may talk about work; others make a point of not doing so. The practice is a living social convention, not a standardised productivity programme.

The Bank of America comparison is therefore a parallel, not an origin story or a scientific validation of Sweden. What connects the two is more basic: people stop at the same time, and the apparent interruption creates room for informal contact that staggered individual breaks do not.

The MIT work began by measuring conversation patterns

In 2009, researchers associated with MIT’s Human Dynamics Laboratory studied a bank call centre where 80 employees worked across four teams. An early MIT technical report describes staff and managers wearing sociometric badges for a month while researchers obtained daily measures such as number of calls, average handling time, speaking time and system use.

The badges were designed to detect patterns of face-to-face interaction, including who was near whom and features of their exchanges, rather than interpret the semantic content of private conversations. The team then compared communication networks with surveys and operational records.

The initial finding was observational. Employees who talked with more colleagues tended to handle calls faster and report less stress, while customer approval remained stable. That correlation did not prove that conversation caused better performance. Faster or more experienced workers might simply have had more room to socialise, or a well-managed team might have produced both stronger relationships and better numbers.

It did, however, identify a practical difference between teams. As Alex “Sandy” Pentland later explained in Harvard Business Review, average handling time offered a clear performance measure, while the communication data showed that the strongest teams were not behaving like collections of isolated high performers.

Coordinated breaks changed what the workplace allowed

Managers had staggered coffee breaks so that phone coverage remained steady and socialising did not distract too many people at once. The researchers proposed aligning breaks within teams. That simple scheduling change allowed colleagues to talk, decompress and exchange tips at the same time.

An MIT News account published in 2014 says the bank tracked the intervention for three months. It reports a 15 to 20 per cent increase in productivity, a 19 per cent fall in stress and turnover dropping from 40 per cent to 12 per cent after the change.

The scale and definition of those numbers require care. The public article does not give the denominator behind the turnover rates, confidence intervals, the exact follow-up window for each measure or a full account of concurrent management changes. Nor does it present the work as a peer-reviewed randomised trial.

The $15 million figure comes from a later MIT Initiative on the Digital Economy brief by Pentland. It says lagging teams caught up and that profits rose by $15 million when the bank implemented the advice across its call centres. Other retellings call the same figure annual productivity gains or savings. Without an underlying financial analysis in the public sources, the safest description is a company-reported estimate linked to the wider rollout, not an independently audited experimental effect.

The useful part of a break may be invisible to a stopwatch

Call-centre work can look completely scripted. Yet difficult customers constantly produce exceptions. Informal conversation can carry tacit knowledge that is awkward to capture in a manual: how one person calmed an angry caller, found a quicker route through a system or interpreted an unclear rule.

Pentland’s explanation was that face-to-face talk helps this knowledge move through a group. A formal meeting has an agenda and hierarchy. A shared break permits unfinished stories, questions that feel too small for an email and details that only become useful when another person says, “I had that problem too.”

Recovery may matter alongside information. A pause can interrupt sustained emotional labour, while humour and ordinary conversation can change how demanding the next call feels. The Bank of America case did not isolate knowledge transfer from rest, social support, stress relief or other mechanisms. The bundle may be the point, but it makes a single causal explanation difficult.

Fika offers a comparable social container. Coffee and a cinnamon bun are visible, but the less visible resource is predictable permission to step out of task mode together.

Later studies offer narrower supporting clues

A 2018 study followed 71 call-centre employees for two weeks, combining twice-daily surveys with objective sales records. Relaxing, social and cognitive microbreaks were associated with more positive affect, which in turn predicted better sales performance. But the microbreak effect varied with work engagement: the indirect association appeared among employees with lower general engagement, not among those with higher engagement.

That is useful because it complicates the idea of a universal benefit. Different workers may have different needs, and “break” covers activities as varied as chatting, stretching, scrolling and sitting quietly. The study was also an experience-sampling analysis, not a random assignment to communal fika.

A more literal fika experiment appeared in 2024. Four American emergency medicine residency programmes used a randomised crossover design during teaching conferences. Among 64 residents with paired data, self-reported sleepiness averaged 4.6 on fika days and 5.5 on control days. The 15-minute fika-style intervention included a pause outside the lecture area, coffee and other drinks, pastries and social conversation.

That study measured sleepiness during medical teaching, not productivity at work. Its control conditions varied, and the intervention bundled time away, refreshments and contact, so it cannot say which element mattered. It is supportive evidence for a structured shared pause in a very different setting, not a replication of the bank case.

A shared break stops working when it becomes another demand

The MIT story can easily be flattened into a management slogan: make people socialise and watch performance rise. That misses both the evidence and the humanity. A break that becomes a compulsory networking session, a surveillance exercise or another target on a dashboard may lose the ease that made it useful.

The badges also raise a question that early celebratory accounts treated lightly. Even when employers receive only aggregated patterns rather than recordings of what anyone said, workplace sensing changes the balance of power. Consent, data access, retention and the possibility of individual inference matter independently of whether the resulting recommendation sounds benign.

Nor is communal time equally restorative for everyone. Some workers need quiet after sustained interaction. Remote teams cannot reproduce a shared room simply by adding another video call. Front-line coverage, disability, language, hierarchy and whether a team feels psychologically safe all shape what a group break becomes.

The clearest parallel with fika is modest. Time away from direct production is not necessarily empty. When colleagues receive genuine, protected space to meet as people, they may return with knowledge, trust and capacity that a minute-by-minute efficiency model cannot see. The Bank of America case makes that possibility unusually visible; it does not turn a Swedish cultural practice into a guaranteed corporate return.