What successful software engineering teams have figured out about culture

The idea that culture is important to an organization’s success isn’t exactly groundbreaking. Legendary CEO and investor Ben Horowitz just published a whole book about how culture created success for everyone from Ghengis Khan’s army to Shaka Senghor’s prison gang. But new research is increasingly showing that culture can be measured objectively and tied directly to positive outcomes for software engineering teams.

Here’s what successful software engineering teams have figured out about culture.

Culture matters

In Accelerate: Building and Scaling High Performing Technology Organizations, the authors write that high-performing teams enjoy a host of good outcomes, including higher revenues at their organization, greater speed in deploying software, lower mean time to recovery, and so on.

One thing that differentiates high-performing teams is they have what’s called a “generative culture.” Here’s what that means.

The Westrum Typology

Sociologist Ron Westrum spent decades studying how organizations can avoid catastrophic errors. Now called “human factors engineering,” this field of study originated in complex, technical, high-stakes fields like aviation and healthcare. In 1988 Westrum introduced the world to his insights in a paper titled, “Organizational and interorganizational thought” which he presented to the World Bank conference on Systems Safety.

In A typology of organisational cultures, Westrum was able measure something as amorphous as “culture,” and connect particular aspects of culture to patient safety outcomes for healthcare organizations.

What Westrum found was:

1. Cultures can be organized into three types: pathological, bureaucratic, and generative.

2. The most salient aspect of organizational culture is how new information moves through it.

3. Cultures that were able to incorporate and respond appropriately to new information measurably outperformed organizations that couldn't.

What do pathological, bureaucratic, and generative cultures do differently with new information? Let’s dig in.

Pathological, bureaucratic, and generative organizations

Westrum’s central insight is that you can predict organizational performance by categorizing organizations based on how information flows through them.

Keep in mind that cultures can vary within the same company. So your marketing department might have a bureaucratic culture while your engineering team has a pathological one.

Here’s a short explanation of how information moves within each culture type.

Pathological

Rampant power-seeking in pathological cultures hinders effective information flows. Because information is power, leaders in pathological cultures tend to withhold it. Pathological organizations also tend to be fear-based. Leaders rule with an iron fist. Workers are afraid to share knowledge with their superiors because they risk losing status or being blamed. In these organizations, leaders choose what’s best for their careers versus what’s best for the organization as a whole.

Bureaucratic

In a bureaucratic organization, the mission is secondary to the status quo. Worker hierarchies rarely change or get challenged. Leaders in these organizations care more about the rules than the results. If you can succeed within existing, established frameworks, great. If you can’t, oh well. Cooperation between departments is limited. Innovation is only tolerated as long as it isn’t disruptive.

Generative

A generative culture is one in which workers and leaders prioritize the mission over personal power or established rules. Performance trumps politics. Hierarchy is deemphasized. Trust is high. Information is rewarded, even when the news is bad. Leaders encourage innovation. When things go wrong, leaders meet failure with curiosity instead of looking for someone to blame. As a result, workers feel confident taking risks and sharing information.

Westrum in DevOps

In Accelerate, the authors measured the extent to which technology organizations had pathological, bureaucratic, or generative cultures through a survey.

The authors found that generative organizations had superior software delivery performance. Specifically, they had shorter lead times, released more frequently, and were able to restore service faster when outages occurred. Their workers also had higher levels of job satisfaction.

Google found similar results when they analyzed more than 250 attributes of almost 200 of their active teams to find what makes some teams more successful than others. They found the characteristics of the individual members of the teams mattered less than the team’s dynamics. Who is on a team is less important than how well a team works together.

Going forward

Culture can measurably impact a software engineering organization’s success. Whether you want shorter lead times, more frequent releases, faster service restoration, or higher levels of job satisfaction a change from a more pathological or bureaucratic culture to a more generative one is worth the investment. Next, read about the different types of team structure in Software Engineering.