At Spotify, their teams are called squads. These small cross-functional teams, similar to a Scrum team, are often grouped in what they call tribes of up to 100 people. A tribe usually focuses on one aspect of Spotify's business, for example, a tribe that focuses on native mobile apps. A tribe's collection of squads will often co-locate in the same area.
There are many tribes at Spotify, spread across many different locations around the globe. This distributed nature presents a challenge for an organization that needs to move coherently across its music player product range.
Creating alignment is key when we have a high degree of autonomy in our teams so that we can move quickly, unencumbered by a chain of command.
After several evolutions of alignment-by-objective frameworks, including using OKRs, Spotify developed an argument framework called DIBBs. The following is a hypothetical example of a DIBB:
The Spotify DIBB is data-driven, for example, by interpreting analytics taken from the website and mobile apps we can generate possible insights. The beliefs about these insights are then used to create a series of bets. Each bet is then distributed to the appropriate tribe to execute.
DIBBs are cascaded through the organization, as the following diagram shows:
Each squad is autonomous; it will make decisions regarding how to implement the particular bet it has been asked to carry out. Sometimes it will need to work in close collaboration with other squads; sometimes those squads might be part of another tribe.
Bets are similar in concept to bets that we make on a sports game or race. However, these bets differ because with a framework such as DIBBs we can adjust our bets as we get new information.
Therefore, feedback should happen on multiple levels, with each squad measuring the success of its own work. The tribe will also gauge the success that its squads are having, and in turn, they will feed this up the line to the group and so on. Adjustments to bets can be made at any level and cascaded up or down the line.