Five cultural changes you need for DevOps to work

by Lalatendu Das, Ling Lau, Chris Smith

Rapid innovation, time to market, customer-centricity, acting like a start-up—sound familiar? They’re all buzz phrases that excite the CEO and instill fear in those tasked with running, transforming, and safeguarding their organizations.

The CEO has good reason to bang on about agile and its benefits. Agile methodologies really can accelerate product development, drive new value, and spur organizational change. That’s why most companies have at least started on some kind of agile transformation. But the next step—actually getting a product to market at the same pace at which it was developed—is proving painful. Most companies simply are not culturally prepared for that kind of sweeping change.

We’ve seen many organizations such as banks and telcos pour millions into agile development and deliver a finished product after every sprint—and then wait (and wait . . . and wait) for it to pass through traditional sign-offs and further rounds of testing. When the product finally does go live, it’s via a big, old-fashioned release.

So much for acting like a start-up. In the same way you can’t be a little bit pregnant, you can’t be half agile. The entire organization needs to be agile for a business to realize its true value, and that requires a change in culture.

Sources of resistance

So how do companies unlock the value of the agile transformations they have started? DevOps. By uniting software development and tech operations and automating their processes from end to end, companies can clear a seamless path from idea to production. But because the impact of DevOps cuts across every control function, it challenges how teams interact and fundamentally changes many people’s roles.

DevOps requires controls to be embedded in the processes of product creation and release rather than manually tested after the development team has finished. The newer way of working is alien—and rather horrifying—to the people in traditional control functions. Their training, KPIs, and incentives are all aimed, although indirectly, at reducing risk in ways that throttle the pace and quantity of change.

That deep-seated cultural aversion to risk goes double for heavily regulated industries like banking. Banks have to contend with a blizzard of regulations and stringent demands for compliance, and any breach can deliver a material blow to the organization. So it’s no wonder that big and small banks take great comfort from control functions that have kept them safe.

It takes a massive shift—a cultural shift—to overcome that caution. Every control function needs to reimagine its work so that the company can attain agility without jeopardizing its standing among regulators, law enforcement, customers, and investors. This isn’t to say that companies should ignore risk in the pursuit of speed; risk management is crucial, but it must complement DevOps, not throttle it.

Building a DevOps culture

Having implemented DevOps at dozens of companies, we have identified key cultural characteristics that enable a successful transformation:

  1. Push change from the top. Start it from the bottom. Change, especially cultural change, doesn’t happen without top-down sponsorship. But it doesn’t really take hold until it’s executed at the smallest unit possible. Implementing DevOps at the team level, for example, enables teams to demonstrate what is possible, locate obstacles, and break through them while the issues are still small enough to handle. Indeed, successful transformations are usually a continuous improvement journey rather than a “big bang” execution.
  2. Reimagine trust. Traditionally, organizations have established trust through audit-based control frameworks designed to improve quality, assurance, security, compliance, and risk mitigation via checklists and audits of activity. DevOps doesn’t work that way. It requires control functions to trust that product teams can and will be responsible stewards of organization-wide principles and requirements. Clearly trust needs to be earned, but this usually happens quickly when teams collaborate and demonstrate success through small pilots before scaling initiatives. This trust can and will lead to product teams becoming empowered to execute change that is right and safe for the organization.
  3. Design for autonomy and empowerment. DevOps requires engineering teams to own control responsibilities formerly owned by other functions. Engineering teams empowered to push change through to production must embed controls in their processes to give the organization confidence that testing, risk management, and escalation protocols are place. Control must be designed into the process right from the start. Automation can greatly help. But it isn’t just about digitizing existing, routine, time-consuming tasks. It’s about reimagining how controls are implemented to ensure they happen by default within the process without the external interference that usually causes bottlenecks.
  4. Crave improvement through testing. The hunger to improve—the process, the quality, the speed, the impact of every single person—must pervade every corner of the organization. That requires changing mind-sets from “Let’s make it perfect” to “Good enough, let’s see how it works and continue to iterate.” Supporting this cultural change requires putting in place flexible systems and ways of working to identify issues and opportunities, rapidly make adjustments, and test again.
  5. Measure and reward the result, not process compliance. Cultures change when people are measured and rewarded for the right things. Everything, from performance contracts at the C-level to weekly objectives for sys-admins, needs to be aligned with strategic business outcomes and the behaviors required to achieve them.

These characteristics materially improve the odds of success of any DevOps and agile transformation. Successful large-scale change starts small and sets the organization on the path to achieving the vision.

Lalatendu Das is a digital VP based in McKinsey’s Bangalore office. Ling Lau is a digital VP based in McKinsey’s New Jersey office and Chris Smith is a digital VP in our London office.