A hundred, give or take. In October 2012, two agile coaches at Spotify, Henrik Kniberg and Anders Ivarsson, set that as the most people one group of colleagues should hold. Citing what they called the “Dunbar number,” they wrote that “most people cannot maintain a social relationship with more than 100 people or so.” Past that size, they warned, you start to see restrictive rules, bureaucracy, politics and extra layers of management.
Spotify was six years old. It had over 15 million active users, over 4 million of them paying, and more than 30 teams across three cities. In three years its tech staff had grown from 30 people to 250. Their 14-page article explained how it meant to stay quick anyway.
The basic unit was the squad, “designed to feel like a mini startup.” It sat together, had every skill it needed to release to production, and picked its own way of working. It had a product owner but no formally appointed leader. Squads in related areas formed a tribe, kept under that hundred or so. People with similar skills inside a tribe formed a chapter, and the chapter lead was their line manager. Anyone could join a guild, a looser club for sharing knowledge.
The paper was frank about the price of all that autonomy: a loss of economies of scale. The tester in one squad might be stuck on a problem another squad’s tester solved last week. Chapters and guilds were the glue. The paper even raised the obvious objection itself, “isn’t this just a matrix org?”, and answered: “Yes. Well, sort of.”
Then the drawings spread. In 2014 Kniberg published animated videos about Spotify’s engineering culture. People started calling it the Spotify model. In a later podcast interview Kniberg said surprisingly many companies simply took it and applied it. Some, he said, just renamed their teams squads and changed nothing else.
The warning had been on page one. “This article is only a snapshot of our current way of working,” the authors wrote. “By the time you read this, things have already changed.” On the last page they added that the structure had been introduced gradually over the past year, that people were “still getting used to it,” and that it gave the company something to “grow into.” The 2014 post with the first video said its content “isn’t all true for all squads all the time.”
In 2015 Kniberg wrote that the material “wasn’t actually intended to be a generic framework or ‘model’ at all.” He found the copying “rather scary at first,” though he had yet to see a company end up worse off. Even the number drifted in the retelling: years later, on that podcast, he put Dunbar’s figure at somewhere around 150.
People who worked there were blunter. Jeremiah Lee interviewed for a product job at Spotify in 2017, and before his first interview the recruiter warned him not to expect an Agile utopia. He joined after the company had tripled to 3,000 people in 18 months. In a 2020 essay he wrote that the squad model was, in his experience, only ever aspirational and never fully implemented, and that managing engineers through chapters introduced more problems than it solved. He quoted a 2017 conference talk by Joakim Sundén, an agile coach at Spotify from 2011 to 2017:
“Even at the time we wrote it, we weren’t doing it. It was part ambition, part approximation. People have really struggled to copy something that didn’t really exist.”