In 1964 Volvo opened a new car plant at Torslanda, in Gothenburg. Five years later, according to the Uppsala business researcher Thomas Sandberg, the plant was so hard to staff that some departments had no Swedish employees at all. Wildcat strikes broke out between 1969 and 1971.
“If you are a supervisor experiencing the situation where people don’t turn up in the morning, you grow more willing to change,” Volvo’s president, Pehr Gyllenhammar, told New Scientist in 1974. “You wonder every day if you have 15 or 20 per cent of your workforce to replace.”
He had already remade the plans for a new plant at Kalmar, on Sweden’s Baltic coast. It opened early in 1974 with no conveyor belt, and each team assembled one part of the car. Each car rode on its own battery-powered platform that could tip it on its side, so nobody had to work underneath. A team’s work on each car took 15 to 40 minutes. A worker was a “door expert” or an “electrical system worker,” the magazine reported, rather than a mere “auto assembler.”
Competitors, Gyllenhammar said a decade later, saw the plant as an “exotic little venture.” Inside Volvo, he said, it “had to be rejected by all those who defend stability, including my own colleagues in top management.”
Five things a job can give
Two researchers were asking the same question with questionnaires. In a 1974 report from Yale University for the US Office of Naval Research, J. Richard Hackman and Greg Oldham argued that people are motivated by the work itself when it feels meaningful, when they feel responsible for how it turns out, and when they know how well they are doing. All three, they wrote, “must be present.”
Five features of a job, they proposed, create those states. Skill variety: the work calls on several different skills. Task identity: you do a whole, identifiable piece, from beginning to end. Task significance: it affects other people. Those three make work feel meaningful. Autonomy, freedom over schedule and methods, makes people feel responsible. Feedback from doing the work itself tells them where they stand.
Work often motivates when it’s varied, whole, meaningful, yours to run and shows results.
Cooking dinner for friends has all five. Chopping onions for diners you never see, at a pace someone else sets, has almost none.
Hackman and Oldham turned this into a motivating potential score: average the first three features, then multiply by autonomy and by feedback. Each is rated 1 to 7, so scores run from 1 to 343. The multiplication was the bold claim: a varied job with little autonomy should still score near the bottom.
It was not meant for everyone. People who strongly want growth should respond well, they wrote; others may find such a job “anxiety-arousing.” The full model appeared in their 1980 book, Work Redesign.
Ten years to pay off
The first results at Kalmar were a letdown. Workers liked the teams, the chance to vary their pace and the working environment, the Linköping professor Christian Berggren wrote later. But each car took as many assembly hours as at Torslanda.
By Thomas Sandberg’s account, the design held autonomy back: a central computer kept cars in strict order, and managers delegated little. By 1984, after a rationalization programme, Kalmar needed 25% fewer assembly hours per car than Torslanda. Work intensity had risen too.
What the studies found
In 1987 Yitzhak Fried and Gerald Ferris reviewed nearly 200 studies. Job characteristics were related to both psychological and behavioral outcomes, but the model’s specific chain, from each feature to its matching state, fit less well.
A 2007 meta-analysis by Stephen Humphrey, Jennifer Nahrgang and Frederick Morgeson pooled 259 studies and 219,625 people. Motivational characteristics of the job explained 34% of the variance in job satisfaction and 25% in subjectively rated performance, but only 2% in turnover perceptions. Social characteristics added another 17% for satisfaction and 40% for organizational commitment. “The two best predictors of job satisfaction were autonomy and social support,” the authors wrote.
The formula fared worse. A 2001 review by Billy Boonzaier and colleagues found no evidence for the multiplication and recommended simply adding the scores. It reported strong links to satisfaction and motivation, and weak ones to work outcomes. Richer jobs go with liking the work more reliably than with doing more of it.
The payroll queue
Imagine a payroll specialist who fixes errors a system flags. She corrects a field, closes the ticket, and never learns whether anyone was paid right.
Her manager gives her every correction for two departments instead, from the employee’s first email to confirmation that the money arrived, and lets her set the order. The work now has a start, an end and a person attached. If she wants more stretch, her satisfaction may well rise. Whether she clears more tickets is a separate question; at Kalmar, the hours took a decade to come down. Job enrichment covers the practice.
When the assistant takes a share
An AI tool can redesign a job without anyone planning it. Manuel Hoffmann, Frank Nagle and colleagues, from Harvard Business School, Microsoft and GitHub, studied about 50,000 maintainers of popular open-source projects after GitHub’s Copilot coding assistant became generally available in 2022. Those on one side of a ranking cutoff got it free; those just past it did not.
Eligibility raised coding’s share of their activity by 5.4 percentage points and cut project management’s by 10. Their exposure to new programming languages rose by about 22%. In Hackman and Oldham’s terms, that looks like more autonomy and variety. But they also worked in repositories with 17 fewer peers, against a baseline of 22, which is the social side Humphrey’s team found matters so much.
The study counted public activity, not satisfaction or code quality, and it is a working paper GitHub helped fund. Whether those developers were more motivated, the study did not ask.