Agile at Scale: How „The Spotify Model“ came to life and how it can be applied in a remote-work environment

Joakim Sunden was one of the Agile Coaches at Spotify working together with the CTO to develop a new approach to Agile at scale, aka “The Spotify Model” with Tribes, Squads, Chapters and Guilds. He is also a well-known figure in the agile community, having organized and spoken at many international conferences and community events over the years. In 2016 he held a keynote at Agile Tour Vienna, since then we partner with Joakim and also offer a training course about SAFe and the „The Spotify Model“. For this post, we asked Joakim how „the Spotify Model“ came to life and how it evolved since then.


How would you explain the Agile at Scale (Spotify Model) to someone who has never heard about it?

There’s no short answer to that so you’ll have to bear with me for some background story. Back in 2011 when I joined Spotify we had fewer than ten teams, or squads as we could come to call them. But we already experienced some growing pains and we knew that we would grow a lot as we had big plans and a lot of VC funding. We wanted to desperately avoid the increased bureaucracy and slow speed many of us had experienced with other big companies so we tried to come up with a new way of organizing ourselves to stay agile even at scale.

At it’s core it’s all about trusting the people you hire to do a great job and support them the best you can without getting in their way. This is why we called the core of our org an “autonomous squad”, a 5-10 people small and empowered product development team who are free to make a lot of more choices than any other team I had ever worked with in the past. Not just how to build things, but also to a large extent what to build. Within bounds of course.

One example of such bounds is the product direction, or goal, of your tribe. Each squad was part of a mission, or product area, called a “tribe”, together with maybe four to ten other squads. The Tribe Trio leadership would set a direction in dialogue with the squad and upper management, within which the squads would have freedom to explore different experiments to push the metrics the right way. Leadership is all about supporting you, making sure you have clarity of direction and everything you need to get there, in terms of skills, resources, decision power, etc.


How can you apply some of Spotify’s organizational structures and practices to your SAFe ARTs?


That’s also true for your closest manager who leads the functional area, or Chapter as our name was, you’re part of together with 5-10 coworkers with similar skill sets. Kind of a matrix model where the manager is a servant leader responsible for your development and success rather than for making decisions and relaying information, and so on.

On a surface level this is what most people understand as “the Spotify Model”; Squads, Chapters, and Tribes. Because that’s the visible most apparent structural parts of it. In practice it’s much more about mindset, culture, leadership, all those things that’s really different and makes it tick, but harder to observe and understand. And that’s what we drill down into in the course. That and the finer mechanics of some of what we learned along the way and from living with this model and evolving it  for several years.

So the first white paper on the Spotify model came out in 2012. What has happened since then?

Wow, a lot really… One of the most concrete things that we abandoned rather quickly after the paper came out was the idea that a Chapter Lead, that is, an engineering manager, should work 50% as a manager and 50% as an individual contributor in a squad. That simply didn’t work, it was too much. Which is not to say that it’s not the right thing for other companies to be doing, but it didn’t work for us with very high expectations on the people development part of the role.

Great experience. Completely realistic view. Down to earth presenter.

– Attendee feedback, Agile at Scale Workshop with Joakim Sunden in 2019

This is by the way a reason I don’t want to talk a lot about changes to Spotify’s way of working, since people seem to perceive it as some kind of evolution and “what Spotify’s doing now must be so much better than the old white paper, so let’s take a short cut and go straight to where they are now”. How Spotify evolved is not necessarily the way your organization needs to evolve. What was working well at some point in one context may not be working well in another context. In an organizations of Spotify’s size there’s now also multiple different solutions to similar problems in different parts of the company. Some held on to the Chapter Lead model while others tried other approaches to solve for other problems, reach other goals.

So rather than talking about the evolution of a specific model we spend a lot of time in the course talking about what we learned from different things, the thinking behind it, how things evolved. All with the goal of course participants being better equipped to solve their own issues themselves when they get back to work. Using Spotify as an inspiration, for different types of solutions, but also the approach to change and problem solving that we applied. And not just Spotify, we drew inspiration from plenty of sources ourselves, and so should you of course.

Can the Spotify model also be applied in a remote work environment?

Absolutely! Spotify was early on a distributed company with product development in Stockholm, Gothenburg, New York and San Francisco. Later we added Boston, where I lived and worked for a few years, and more recently London. So a remote work environment was there pretty much from the start for me.

Working in a remote environment it’s even more important to let go of control and trust your people and teams to work autonomously to solve problems. It’s even more important to have good practices in place for collaborating and communicating, to create clarity of direction and communicate progress when working remotely.

When I left Spotify and started consulting with companies I was shocked to learn how bad most of them were at remote work practices and how many companies just lack the tools to be able to collaborate efficiently in a remote environment. Even though the course is very little about remote work practices per se, indirectly it’s a lot about supporting that way of working. Since we’re also running it remote, participants will be able to learn some really good tools for remote workshops, should they not already be familiar with them.