Table 1: Innovative business initiatives, managed under Mode 2, use different techniques than systems-of-record, managed under Mode 1.
We see many organizations adopt bimodal operation in innovation labs where occurrences of agile development, cloud computing and DevOps or lean startup techniques are already underway. Shifting to a bimodal approach could have big cultural impacts, hence it’s important to socialize people in the tenets shown in Table 1. You might also need to alter the way people are evaluated since values under bimodal operations change.
New concepts should be framed in terms people already understand. For example, the sense-and-respond framework that sprung from the 1980s marketing literature is now widely accepted. Mode 2 is compatible with this thinking, offering a sense-and-respond model, where initiatives evolve as the future unfolds (versus executing on a plan that is developed well in advance).
Mode 2 operations also require a new category of partners experienced in building startups and leveraging software methods consistent with the ability to pivot software (unlike Mode 1 partners that align rigid processes with low-skilled, low-cost scaled staff, processes that work well when requirements don’t experience rapid change or require heavy interactions with end users in local geographies).
Expect teams to overlap
Most IT organizations are characterized by Mode 1, with teams that, by necessity, focus on risk aversion, security and compliance for traditional, industrialized applications. While Mode 1 teams rarely match the innovative, faster-paced velocity of agile Mode 2 teams, they are still under pressure to do things faster and to introduce innovative productivity gains. Hence, we see many organizations start to move Mode 2 techniques onto teams that are innovating systems of record.
For example, as a transformative initiative becomes mainstream, it starts to require the attributes of a Mode 1 system of record. Hence, think of bimodal not as separate work streams and styles, but rather as those that are complementary, where the best practices of each mode ease into a project’s natural evolution.
We’ve certainly seen this in organizations such as Salesforce, Twitter and LinkedIn, where projects initially characterized as Mode 2 migrate to the necessities of enterprise software management. Even digital pioneers such as Amazon and Google, which are typically considered Mode 2 companies, are adopting bimodal operations as they grow and mature. This is why digital pioneers have led the charge on reactive architectures leveraging microservices.
“...think of bimodal not as separate work streams and styles, but rather as those that are complementary, where the best practices of each mode ease into a project’s natural evolution.”
Use a pilot project for transition to agile
The iterative approach of Mode 2 is a chief underlying tenet of dual-mode environments, which is why many organizations introduce bimodal with agile. Of course, adoption of agile also requires a carefully planned introduction. No matter how much pressure the organization is experiencing to adopt agility quickly, it is unwise to take a big bang approach to speed and agility.
Start with a project that is well suited to agile development and characterized by low risk and high probability of success. A pilot lets you learn about agile while you gain experience before rolling out to the larger organization. The ideal pilot should be implementable over three to six months by one team. Make sure the project team gets buy-in from stakeholders willing to approach the project in ways that depart from the norm. It’s also a good idea to select a project that is important but not on the organization’s critical path. Projects that are time-sensitive and have some degree of uncertainty with regard to outcomes—and require heavy involvement from business leaders-—are good candidates.
Though agile techniques are all about speed, it’s important to realize agile is a highly collaborative process that requires a significant time commitment from the business. Without this commitment, the integrity of the project’s schedule will be at risk. Select a methodology where training and development are readily accessible, such as Scrum, a popular agile framework in software product development.
The ideal team will comprise both agile advocates and those that are more on the skeptical side. Make sure you get at least one team member who has some agile experience, teamed with neophytes who are highly motivated to learn. Keep the team size to fewer than 10 members (using the Jeff Bezos model that two pizzas should satisfy the appetite of any agile team).
Use DevOps as a bimodal bridge
DevOps is often aligned with Mode 2 due to its emphasis on agile and lean (and because it is still relatively new). DevOps implementations vary, since there are no prescribed patterns and practices given the DevOps focus on culture, organization, processes and tools. Beyond enabling speed, DevOps enables far greater quality. Since DevOps forces automation of Mode 1 manual behaviors, organizations find that releases are often much higher quality using DevOps. This makes DevOps a good candidate for acting as a bimodal bridge since it requires both Mode 1 and Mode 2 behaviors. Gartner, for example, predicts that by 2020, 80 percent of the practices identified with DevOps Mode 2 will be adopted by traditional Mode 1 groups to advance the interests of the entire organization.
Every IT organization is on a path to continual improvement, hence it’s important that Mode 1 and Mode 2 learnings are shared. Management should take steps to avoid Mode 2 teams becoming known as elite or special. Both modes have critical roles to perform, and each ultimately needs the other to succeed.
Build reactive architectures
Instead of fully integrating each piece of complex technology architecture into the whole, reactive architectures use common patterns to facilitate the appropriate connections between each part. This new thinking has positioned reactive architectures as the new model for equipping enterprise systems with the type of speed required to compete in today’s rapidly changing digital economy.
At Walmart, for example, a reactive architecture decouples the systems of engagement (like user experience, targeting and search) from the systems of record (such as content management, catalog management and inventory management). This shift has boosted mobile orders alone by a whopping
Reactive architectures are especially useful when organizations need to support millions of customers or millions of devices and interaction points. The demands for each of these devices are frequently changing, and a reactive architecture helps sustain a manageable technology infrastructure. Netflix, for example, uses a reactive architecture to scale across devices and to quickly respond to the changing tastes of its enormous customer base.