Discover more from Augmented collective intelligence
The organization as a network of teams
Revisiting organizational design from the perspective of network science
In my previous article I proposed a model for conscious leadership development from the understanding that the level of consciousness of people in an organization sets the limits for their capacity to collaborate and give rise to a collective intelligence. In this new article, I will cover another key element for human organizations to leverage their collective intelligence: the connectivity between people in the organization so as to create an effective system of communication, interaction and decision-making. This will take us back to the everlasting challenges of organizational design which I will try to illuminate from the perspective of network science, an academic field that has developed over the last decades to study complex systems.
The relevance of organizational design
I believe organizational design should be a key discipline in management. While all the talk about digital transformation typically revolves around technology, talent, culture… -all of them very relevant topics-, not so much attention has been paid to organizational structures. However, the org chart -with all its roles and reporting lines- is typically a very telling expression of the culture of a company, while leadership styles are often very conditioned by the structure that vertebrates an organization. Even the most basic processes of internal communication and collaboration are greatly influenced by the design of the organization. All these ideas are just to make the point that the design of an organization is of paramount importance in the emergence of its collective intelligence.
Being this the case, what is the problem that we want to solve when designing an organization? Fundamentally, it is the challenge of human collaboration at scale. We can illustrate this problem with Dunbar’s number. Robin Dunbar is a British anthropologist who leads the Social and Evolutionary Neuroscience Research Group at the University of Oxford. He is best known for formulating Dunbar's number, a measurement of the "cognitive limit to the number of individuals with whom any one person can maintain stable relationships". According to his research, that number is 150 people. This number is not an occurrence but the result of decades of research on group behavior in primates and humans. And following Dunbar's theory, this limit is determined by the size of our neocortex, so it is not something we can easily change. Therefore, we could say that the challenge we face is how we can design an organization in which more than 150 people can collaborate effectively.
The common response to the organizational challenge
If we look around us and analyze the response of existing organizations to this problem, we can easily see that the vast majority of corporations, businesses, public administrations, NGOs, etc are designed following two principles: hierarchy and functional division.
On one hand, hierarchy allows any organization to grow by creating additional levels of hierarchy and lines of reporting between them.
On the other hand, functional division allows any organization to manage its growth by dividing its teams according to their functional specialization.
This model of the hierarchical and functional organization dates back to the industrial revolution, i.e. it is more than 200 years old. And although it has worked very well for decades, it has failed to do so in the context of accelerated change in which we live today. At present, the most important capacity for any organization to succeed is its capacity to continuously learn and adapt to change. This requires a fluid communication within the organization that you can very hardly achieve under a hierarchical structure. At the same time, most of the challenges organizations face today are multidisciplinary in nature, making structures based on functional division badly equipped to meet these challenges. We can thus see why the traditional (hierarchical and functional) organization is falling apart.
The emergence of new forms of organization
For this reason, over the last 15 years different organizational models have emerged in an attempt to reinvent organizations: from teal organizations (as described by Frederic Laloux in his book Reinventing Organizations), to holacracy, sociocracy, DAOs, etc.
Following this trend, many organizations have launched different initiatives seeking to dismantle their hierarchical and functional structures to replace them with flatter, more liquid models. In the case of BBVA, for example, during the last decade an agile transformation program has deeply transformed the organizational blueprint of the bank resulting in a less hierarchical, more liquid and transparent organization. As a result, the bank has significantly improved its time to market, product quality and productivity. Other companies in different sectors have followed similar approaches, e.g. Spotify and ING pursuing agile transformations, Zappos and Medium embracing holacracy, Loomio implementing sociocracy, etc.
These new models have different approaches to people roles, organizational structures, decision making processes, etc, but they all have one thing in common: understanding the organization as a network of teams, be they scrums (agile), circles (holacracy and sociocracy), self-managed teams (teal organizations), etc.
A new perspective from network science
Although traditionally the way a company organizes its activities has been described using linear organizational charts, in reality the behavior of an organization is better described as a network of interactions and collaboration. Beyond the official chain of command, the informal network -capturing the complex patterns of communicative interaction between interdependent teams- is a much better representation of the real behavior of an organization. Accurate maps of these organizational networks can expose the lack of interactions between key areas, or identify individuals who play an important role in bringing different units together. From this perspective, network science appears as a relevant discipline to understand the behavior and development of human organizations.
Network science is an academic field that studies complex systems through their representation in network models, trying to uncover the patterns and mechanisms that characterize these systems whether they are biological, social, financial or technological. The field draws on diverse theories and methods including graph theory from mathematics, statistical mechanics from physics, and data mining from computer science. The impact of network science in our society is very broad, including developments in diverse fields like power grid planning, pharmaceutical design, fighting terrorism, controlling epidemics, etc.
The first idea we can borrow from network science is that networks have properties encoded in their structure that limit or enhance their behavior. In our analysis of human organizations, this means that the productivity (the capacity to produce outputs in the short term) and learnability (the capacity to learn and produce outputs in the long term) of an organization are both encoded in its network structure. This is true both at the individual and collective level:
An employee’s productivity is linked to his or her location in the informal organizational network. An employee who is highly connected to other people in his or her department and also connected to other people in other departments will have more levers to achieve his or her goals and to solve problems as they arise than another employee who is quite isolated from the rest of his or her team.
An organization’s performance is dependent upon the connectedness of its network graph. This fact is well understood from the perspective of Metcalfe’s law which states that while the cost of network-based services increase linearly with the number of nodes, the benefits are driven by the number of links created between nodes. In our organizational context, we could say that while the cost of an organization grows linearly with the number of employees, the performance of the organization grows with the number of connections between employees, i.e. it depends on the connectedness of its network.
Among the many properties a network can exhibit, one that is particularly interesting for our analysis is the so-called small-world phenomenon. Network science has shown that many real world networks have small-world properties, meaning that the average number of steps along the shortest paths for all possible pair of network nodes is relatively small. This property is often referred to as the “six degrees of separation” theory: the idea that all people on Earth are six or fewer social connections away from each other. Bringing this concept to an organization, we could say that an organizational network shows the small-world property when any two people within the organization are connected through a relatively short path for the size of the organizational network. We can thus think that a way to solve Dunbar’s number problem in an organization of thousands of employees is by developing an organizational network whose structure shows small-world behavior. This can be understood as a way to bring all employees close to each other despite the big size of the company.
The question is then how can we create a small-world organizational network. And the answer lies in a phenomenon called clustering. In network science, a cluster refers to a group of nodes that are more densely connected to each other than to nodes outside of the group. Network clustering is often quantified using a measure called the clustering coefficient. The clustering coefficient of a node is the proportion of its neighbors that are also connected to each other, and the global clustering coefficient of a network is the average clustering coefficient across all nodes in the network. High values of the global clustering coefficient indicate that nodes tend to cluster together, while low values indicate a more random or dispersed network structure. This phenomenon is relevant because networks with a high clustering coefficient tend to show small-world properties. Clusters can facilitate the spread of information within a network because the dense connections within a cluster allow information to be transmitted through multiple paths, increasing the ease of reaching all nodes within the cluster. For the same reason, clusters also enhance the resilience of the network to disruptions.
In organizational terms, a cluster is what we typically consider a team. A team is a group of people who work together on a recurrent basis through continuous interactions among all members of the team to achieve a common goal. In a team, all members are connected to each other and interact between them through multiple channels of communication, such as face-to-face meetings, emails or instant messaging. This creates a dense network of communication within the team which can thus be considered a cluster within the larger network of the organization.
But clusters can also create barriers to communication between different clusters thus limiting the overall effectiveness of the network. For this reason, for an organization to show small-world behavior we need not only to have a high clustering coefficient (by structuring the organization around teams), but also to connect those clusters (the teams) through bridge nodes. A bridge node is a node whose neighbor nodes are sparsely connected to each other and are likely to be part of different clusters if the node is removed from the network. Bridge nodes are essential for various phenomena in complex networks, including the transmission of information across the network and the generation of new links between nodes. Therefore, by bridging the different clusters (teams) in the network we seek to create a network of teams or what we can call a team of teams, as the connectedness between the different teams makes them also behave as a giant team, allowing information to spread quickly and efficiently.
Creating a team of teams: the case of Sngular
An interesting example of how a company can design its organizational structure as a team of teams is Sngular (for disclosure, I am a board director at the company). Sngular defines itself as “a global ecosystem of talent and technology”. With more than 20 years of experience, Sngular develops technology and innovation projects for leading companies in sectors like banking, retail, industry, telco and health, globally. It also offers services focused on the development of talent such as its talent agency or its radical learning projects. Sngular currently employs 1.375 people in 8 countries and defines its purpose as “to be the best ecosystem for good people to develop all their talent and make a positive impact on society through technology”.
The company has been growing at a fast pace, so much that in the last two years (2020-2022) it has more than doubled its sales (from 45 to 92 M€), EBITDA (from 6 to 13 M€) and headcount (from 630 to 1.375 people). More than doubling your employee base in 24 months can create many challenges, one of them being having to rethink your organizational structure to accommodate a much larger and complex organization. This was precisely the case for Sngular, as the company decided in 2022 to redesign its organizational structure to be able to scale it to thousands of employees while preserving its well-known culture of openness, adaptiveness and agility.
Pursuing this goal, the bet of the company was to create a team of teams organization, i.e. an organization based on a network of autonomous and interconnected teams. In this model, a team is a group of between 10 and 150 people that integrates different professionals working together to develop solutions that meet the needs of a certain type of customer. This means that teams are not communities of practice (they are actually multidisciplinary) and are not defined by a specific technology or industry but according to the need of a type of customer (e.g. a CTO who needs to build a new core platform, a CHRO who wants to redesign the employee experience, an Operations Director who seeks to automate processes, etc). Teams are the basic organizational unit around which the entire organization is built; they are self-contained and self-organized and have autonomy to make decisions on their goals, projects, talent management, methods and tools, etc.
Building on their autonomy, Teams are at the same time interconnected to create a highly connected network. The connectedness of the network is based on the following principles:
A shared vision of Sngular's purpose, values and strategy, accompanied by some more operational rules on internal and external communication, people management, compliance, etc.
Full transparency on the activity developed by all teams (regarding their goals, projects, customers, products, etc) to the rest of the network.
The creation of communities of practice that connect people from different teams who share an area of expertise (e.g. python development, design research, microservices...) with the aim of promoting knowledge management, common methods and tools, reusability of developments, etc.
The mobility of people between teams so that any person can move from one team to another as part of his or her professional development.
The search for opportunities for collaboration between different teams in order to develop solutions of greater scope or complexity than those that a single team is capable of creating.
Figure 1. Visualization of Sngular’s teams (red dots) and their people (green dots)
At the beginning of 2022 Sngular already had a few teams working (Sngular Studios, Sngular Design, Sngular Data & AI…), but in order to fully implement the team of teams model, the company had to transform the rest of its organization by creating new teams out of a big operations area. A key challenge was to identify the person or persons to lead the new teams to be launched, as the leadership role of a team does not only require good technical skills but also business development and people leadership capabilities. The creation of new teams (like Sngular Cloud Adoption, Sngular Apps, Sngular Payments, etc) was also a great opportunity to make progress in key priorities for the company like talent development (which is the first responsibility of any team) and commercial proactivity (leveraging the more specialized offerings of the teams).
Finally, although the team of teams structure is the backbone of Sngular’s organization, it is complemented by the role of country managers (who have a transversal view of the presence of different teams working with local clients in a given geography), industry experts (who have deep knowledge about business models, market dynamics, current trends, etc in a given industry) and corporate functions (that set global policies and give support to the teams in areas like finance, people, comms, etc).
In conclusion, network science tells us that networks with a high clustering coefficient tend to show small-world properties. Accordingly, organizations based on a network of autonomous and interconnected teams can more effectively overcome the challenge of collaboration at scale, facilitating the spread of information across the organization and making it more resilient to disruptions. Sngular’s team of teams organization is a good example of this model.
Thanks for reading Augmented collective intelligence! Subscribe for free to receive new posts and support my work.