“Software is eating the world,” said Marc Andreessen, co-founder and GP of the famous venture capital firm Andreessen Horowitz. As digital devices proliferate and new technologies change lives, software has become an indispensable factor for all businesses. Even those working in real estate or healthcare must use some system or application. Therefore, developing software effectively and efficiently is key to the survival and success of any business.
However, despite the emergence of many concepts and project management frameworks, such as Agile, Scrum, Extreme Programming (XP), KanBan, and Team Topologies, only a few companies know how to develop software efficiently. I've encountered many error patterns that ultimately resulted in schedule delays, software quality deterioration, and customer satisfaction. To avoid ineffective management and achieve better performance, you should understand the best practices, concepts, and principles of software project/product management.
One of the most impactful practices is the way teams are structured. Many studies (such as Conway's Law) prove that if the team structure is built in the wrong way, team performance will be greatly restricted. In this article, I will explain several principles that enable building effective structures in organizations, based on my experience and in-depth research.
1. System design first
The first principle of building a high-performance R&D team is to design the system you want before designing the team structure . American computer scientist and computer programmer Melvin Edward Conway proposed Conway's Law (Conway's Law) in 1967, which describes the structure and communication methods of an organization that will constrain the system design of its production . If an organization sets up a design team, business team, front-end team and back-end team, then the output of the system will also be business documents, design components, front-end components and back-end components, which requires a lot of team communication to complete the final integration and delivered to users. Team structure based on Conway’s Law is one of the most common anti-patterns and is still used by many organizations to this day.
(Example of Conway's Law)
The following team structure does not require frequent communication and allows the team to independently provide value to users (although it cannot be completely independent). It is also known as flow-aligned team, functional team or cross-functional team.
2. Consider cognitive load
"High-Performance Team Model: An Organizational Structure to Support Rapid Software Delivery" published by Matthew Skelton and Manuel Pais in 2019 describes four basic teams in software development organizations: Stream-aligned team , enabling team ( Enabling team) , Complex subsystem team (Complicated subsystem team) , and Platform team (Platform team) . Behind these four basic teams is an important psychological theory: cognitive load—the amount of working memory resources used .
The team structure explained by the previous principle is called a flow-aligned team in The High Performance Team Model. The flow alignment team is cross-functional and the main force in independently delivering value to users. However, the flow alignment team is too busy to keep up with new gadgets, such as new technologies, management methods, or leadership, nor to deal with infrastructure. If you put too much responsibility on flow-aligned teams, they won't be able to handle the tasks and team performance will suffer. Therefore, in order to reduce their cognitive load, the remaining three teams (enabling team, complex subsystem team and platform team) are coming to support .
(Team topology)
- The enabling team helps the flow-aligned team learn new things needed to deliver higher value to users , such as technology, architecture, UI/UX, management methods, etc. The enabling team may consist of internal/external mentors, coaches, or consultants who can teach and implement new things in the stream-aligned team.
- Complex subsystem teams are responsible for specific components that require special skills or can be isolated from normal delivery , such as machine learning models, special graphics/animations, or authentication APIs. The components worked on by the Complex Subsystems team will be determined through discussion, as there will be no clear boundaries between them and the Flow Alignment team.
- The platform team is responsible for building and providing a platform on which the flow alignment team can smoothly deploy and release to deliver value to users. In the context of team topology, a platform is a comprehensive concept that includes documentation, guidance, DevOps, and infrastructure designed to help flow-aligned teams deploy smoothly.
3. Keep your team lean
The third principle is to keep the team "small but refined." Many studies and well-known companies have shown that "small and refined" has higher efficiency and better performance than "big and complete" - this is because communication in small teams is more effective and everyone in the team can understand at any time situation and increase engagement . Amazon has implemented the "two pizza" principle (that is, no more than 9 people in building teams and attending meetings) since its inception and has achieved real success.
Jeff Southerland, the author of "Agile Revolution" also explained in the "Scrum Guide": "A Scrum team is small enough to maintain flexibility and large enough to complete important work within an iteration, usually 10 people or less. In general "We find that smaller teams communicate better and work more efficiently."
4. Servant Leadership
Daniel H. Pink proposed "Drive 3.0" in his book "Drive". He argues that human motivations have changed since the industrial age, and examines three elements of motivation—autonomy, expertise, and purpose—that continue to motivate us.
- Autonomy : Humans want to decide what to do.
- Mastery : Humans want to grow.
- Purpose : Human beings seek meaning.
The "Servant Leadership Theory" proposed by Robert K. Greenleaf is the embodiment of Drive 3.0. This is a leadership philosophy whose main practice is to serve the team and keep team members highly engaged, motivated and sustainably growing at work . In the traditional leadership style, the leader (boss) orders team members (subordinates) what to do and how to do it in a command and control manner. Traditional leadership styles fail to motivate individuals and increase their engagement and performance.
Servant leaders, on the other hand , set a goal and vision (purpose), delegate the means to achieve the goal and vision to individuals (autonomy), and provide opportunities to learn new things (expertise) .
5. Continuous improvement
As of January 2024, Toyota is the world's largest automobile company, and its practices and philosophies have been influencing all business industries, such as the Seven Wastes, 5S (Seiri Sorting, Seiton Straightening, Seiso Sweeping, Seiketsu Cleaning, Shitsuke Literacy) and 3M ( Muri overload, Muda waste, Mura uneven). Among them, the most famous concept is Kaizen (improvement), which is the continuous improvement of business processes. Regardless of business type or situation (even if the company has the highest market share in the industry), there is 100% room for improvement. No business is perfect, so it must continue to grow.
从长远发展来看，持续改进会产生复利效应。该术语最初来自金融学中的「复利」，意为利息长期呈指数级增长并产生被动收入。同样，如果团队和业务持续改进，他们的业务流程也将大幅改善，销售收入会成倍增加；其弊端在于大多数人由于短期看不到显著效果，会停止持续改进，但长期来看，持续改进所带来的效益是巨大的，因此需要坚持不懈和耐心。
Scrum is the most popular agile framework, originated from Toyota, and is a practice of continuous improvement. In Scrum, continuous improvement occurs during a Sprint Retrospective, where the Scrum team discusses what they can do better in future work. This is one of the biggest reasons why Scrum solves team performance problems and has become the most widely used agile framework.
6. Stability
The final principle of building a high-performing R&D team is stability, that is, the team structure/membership and work processes should remain stable . Instability can disrupt teams and impact productivity and performance.
The first is the stability of the team structure and members, which means they should not change or move around frequently. For example, if an organization frequently moves developers between teams, developers will need to re-familiarize themselves with work styles, code bases, and documentation when assigned to new teams, thereby reducing productivity.
The second is the stability of workflow flow. If the work process is not standardized and unified, team members will be confused about how to work and the quality of output will be inconsistent. Toyota defines standardization as one of the most critical practices in business because Ohno emphasized that work should flow smoothly, without obstructions and waste. As a result of the Toyota Production System (TPS), Kanban (Kanban), which is used to visualize work processes and task status, came into being.
Summarize
In software R&D organizations, building an effective team structure is crucial to improving team efficiency and performance. This article summarizes six principles for building high-performance R&D teams, which are:
Principle 1: System design should be completed earlier than team construction.
Principle 2: Fully consider cognitive load and form support teams (enabling team, complex subsystem team, and platform team) to assist the core team (flow alignment team) in their work.
Principle 3: Keep the team small and beautiful. As the team gets larger, communication becomes more complex, reducing productivity.
Principle 4: Practice servant leadership and help teams improve engagement and performance around Drive 3.0.
Principle 5: Insist on continuous improvement no matter what. Even if you are far ahead in the industry, there is definitely room for improvement.
Principle 6: Maintain the stability of team structure and work processes. This is important to avoid confusion and chasing costs and improve team performance.
(The original author is Eiki, and the content has been translated and compiled by LigaAI)
To learn more about technical information, R&D management practices and other sharing, please pay attention to LigaAI.
Microsoft's China AI team collectively packed up and went to the United States, involving hundreds of people. How much revenue can an unknown open source project bring? Huawei officially announced that Yu Chengdong's position was adjusted. Huazhong University of Science and Technology's open source mirror station officially opened external network access. Fraudsters used TeamViewer to transfer 3.98 million! What should remote desktop vendors do? The first front-end visualization library and founder of Baidu's well-known open source project ECharts - a former employee of a well-known open source company that "went to the sea" broke the news: After being challenged by his subordinates, the technical leader became furious and rude, and fired the pregnant female employee. OpenAI considered allowing AI to generate pornographic content. Microsoft reported to The Rust Foundation donated 1 million US dollars. Please tell me, what is the role of time.sleep(6) here?