Every company should think about how its own IT can be more like the speedy high tech companies of Silicon Valley. Back in 2007, Ben Fried was a managing director at Morgan Stanley in New York City where he had a reputation for polite impatience. He focused on doing things quickly in Wall Street’s often fast-paced information technology arena.
But in 2008, when he started to work for Google Inc. as its CIO, he was not considered fast moving. “Google still has the muscle memory of being a small startup. It’s a truly impatient place.”
To uncover what makes the Valley’s high-tech companies so agile and swift, the Accenture Institute for High Performance interviewed 34 Silicon Valley and Internet industry CIOs and another 22 experts including consultants, executive coaches and academics. We concluded that any company that needs IT to improve services and processes at faster speeds should study speedy develop¬ment practices within Silicon Valley’s Internet companies.
We also identified a combination of technologies, management practices and cultural features that enable Valley technologists to work fast. These practices are used by their engineering and IT organizations.
The Valley’s Internet industry has a reputation for building IT systems faster than virtually any other region of the world.
For example, LinkedIn Inc. can run 1,000 experiments simultaneously on the company’s site and build, test and launch improvements three times day. At Facebook Inc., software engineers make thousands of bug fixes, improvements and new features each week. Google tests and releases hundreds of millions of lines of code daily, according to the Google test engineers who wrote the 2012 book “How Google Tests Code.”
Of course, all-digital Internet companies don’t have to retool a factory or run a global supply chain. They can afford more risks. If a social network malfunctions, for example, it won’t wipe out savings accounts. But as all businesses are becoming digital, IT now embodies the core of many industries. Technology functions as the primary tool for creating new business models, when companies enter markets once thought the exclusive preserve of another industry. How quickly companies can go to market, or make decisions, depends on the flexibility of their systems.
Following are the lessons the Accenture Institute for High Performance learned about Silicon Valley’s business culture and practices. Many of them can be adopted by IT organizations in other industries and regions, but only if they are determined to follow through on the major changes required.
Use cloud and open technologies to enable rapid improvements.
IT organizations at larger Valley companies limit their use of conventional licensed software to financial and other administrative or back-office applications where standard packages best meet their requirements for now. As they don’t see them as vital for growth or competition, or as temporary solutions until better software services emerge, they avoid investing energy or money in customization. “Why buy it if we’re not going to keep it plain vanilla?” asks Fried.
But for systems that need to keep improving, such as applications used by their customers or engineers and the infrastructure on which they run, these companies rely on cloud services and platforms and open source technologies. Cloud infrastructure and services are fast, scalable and accessible, meaning whether it’s the cloud service provider or their own software engineers any changes can be made rapidly and will instantly reach all users.
Open source software, such as the Apache Hadoop distributed computing software library or the Cassandra scalable database, is change-friendly so that users can add and share improvements. So, Valley companies frequently team with other companies, collaborating as a community of technologists to create more new technology faster than they could themselves. For example, LinkedIn contributes to the Voldemort distributed storage system and more than 10 more open source projects. “We contribute, they contribute and the code moves forward,” says David Henke, senior vice president of operations at LinkedIn.
Use iterative development disciplines and tools to create systems quickly.
“Move fast and break things” and “done is better than perfect” would be empty slogans if the software engineers at Facebook and other Internet companies couldn’t back it up. Their developers are versed in agile software development techniques that quickly deliver software that users can utilize, and then rapidly improve it in small, incremental steps.
Much of what enables companies such as Google and Facebook to be so agile are the cloud-based development and testing tools provided to their developers. At Google, testing operates like an automated factory, according to a description by Patrick Copeland, a senior engineering director there. Once new software code is checked in, appropriate tests are selected automatically, run, and reported in milliseconds. These tools provide the means to rapidly build and stress-test code to find and predict bugs. Large numbers of “A/B tests” can be run quickly try out alternatives on a sample of users similar to how optometrists test two different lenses on their patients.
Developers at these companies also practice good engineering hygiene. Teams take responsibility for testing and fixing their own code. The code they create is reviewed before it is released. At LinkedIn, says Henke, a central QA team provides the machinery to automate testing for fast iteration. When developers “break something, they have to revert the change or fix the bug as soon as possible. We care deeply about MTTR, mean time to repair.”
Silicon Valley CIOs tell us the Valley has a distinctive culture all its own.
It’s both the heartland of an extremely competitive industry driven by rapid advances in technology, and a close-knit business ecosystem of entrepreneurs, technology professionals and venture capitalists. Technologists there celebrate innovation and entrepreneurship and tolerate failure as the inevitable byproduct of both. Together, it makes for a culture that’s impatient, inventive and open.
That culture is like fuel that pushes IT professionals to embrace agile development, open software, community development and rapid experimentation. It also drives technology professionals to embed themselves in the Valley’s ecosystem by joining groups, gathering with peers, and participating in open source projects. In this environment, new ideas and technologies spread quickly.
CIOs in other regions and industries may be unable to replicate Silicon Valley’s distinctive ecosystem or culture, but their IT organizations can start to borrow Valley company practices. They can begin by taking advantage of the technology and services they have created. They can consider how they can benefit from the rapid iterations and speedy progress of cloud services and open community-developed software. They can lease cloud-based software development and testing platforms services known as platform-as-a-service. Like any technology or service, these need to be scoped for security, maturity and capability.
Also, IT executives can start using agile development techniques. They can identify development projects that are worthwhile candidates for agile development techniques. While traditional methods work well when software requirements are, or must be known, in advance, an agile development approach has the edge when users aren’t sure what they need.
CIOs can also accelerate innovation by collaborating with high-tech Valley companies. For example, software developers from Ford joined their peers at Facebook for a joint hackathon to devise social features for cars, such as listening to a friend’s music stream or receiving directions to Facebook events.
Before companies become fast, they have to learn how to accelerate. There are new skills to learn. But perhaps more challenging is the need to unlearn old ways. For example, traditional development practices, such as change control boards to oversee code changes, are inconsistent with rapid, iterative and agile practices. Companies that use both agile and traditional techniques must figure out how they will co-exist. While CEOs may not completely agree with LinkedIn Chairman Reid Hoffman’s maxim, “if you are not embarrassed by your first release, you’ve launched too late!”, they need to support CIOs who, in their quest to create faster, more agile IT organizations, attempt to follow its spirit.
Jeanne G. Harris the global managing director of information technology research at the Accenture Institute for High Performance. Based in Chicago, her email is firstname.lastname@example.org. Allan E. Alter, a Boston-based research fellow at the Institute, can be reached at email@example.com. Christian J. Kelly is a managing director for Accenture’s Strategy Consulting practice. He is based in Seattle and can be reached at firstname.lastname@example.org.
To download a copy of “How to run IT at the speed of Silicon Valley” you must login to WSJ.com
Read all insights from Accenture’s research study: “Silicon Valley’s Lessons for CIOs and Innovators”