In 1913, it took 12 hours to build a Model T car from start to finish. That is, until Henry Ford, inspired by the production methods he saw in local canneries and mills, introduced the world’s first moving assembly line. Car after car soon rolled off the production line within 2 hours, painted any colour you wanted, as long as it was black – the textbook mass-produced commodity fully formed.
Yet before that, Ford had spent years perfecting production automation. He was a master of Agile thinking ahead of its time. Production en masse was the result of years of iterative experimentation, problem decomposition, removal of process bottlenecks, and the upskilling of workers. It was only when the outcome was certain and there were no more experiments left to be run, that Ford introduced a standardised production process.
In software development we often see the same pattern. Problem solving starts with Agile experimental processes, but over time off-the-shelf SaaS and PaaS solutions emerge, business processes become homogenous, and plan-based approaches can be easily applied. This is one of the concepts at the heart of what Gartner refers to as ‘bimodal IT’.
An important distinction between Ford’s production line and technology initiatives, is that problems once solved are rarely ever stable. Software or hardware that seem concrete and reliable interacts with complex, human systems that change constantly and require a more flexible approach than a binary shift between change-driven and plan-driven approaches.
Gartner defines bimodal1 as the practice of managing two separate, coherent modes of IT delivery, one focused on stability and the other on agility.
In this definition, the first mode (the traditional or plan-based approach) is sequential, emphasising predictability and accuracy when problems are well understood. Whereas the second, more Agile mode is more exploratory, aimed at solving new problems.
The challenge is, that rarely in business are things this black and white. Establishing processes that flex to accommodate increasing levels of uncertainty are imperative for adding value, regardless of the problem faced.
Just like Ford, firstly software companies, then more mainstream businesses undergoing digital transformation have applied Agile ways of working in the initial stages of using technology to solve a problem.
The less understood, predictable or stable the problem domain is, the more valuable experimentation can be in quickly determining a range of potential solutions (otherwise known as set-based design).
With the risk that committing too much to one promising outcome may leave money on the table, businesses need to be free to rapidly iterate and change direction based on feedback from the system – which is what an Agile approach provides.
But there are scenarios which will value standardisation
While rapid experimentation is essential in situations with many possible outcomes, it offers little incremental value when the outcomes have a high degree of certainty.
Over time, competing organisations begin to leverage similar customer insights, products mature and start to converge, and eventually any solution ceases to be novel and becomes a commodity.
But don’t assume the transition between Agile and a more linear approach is static. Lean principles still need to be applied to an otherwise plan-based approaches to reduce the lead time of initiatives and enable rapid response to change, otherwise processes that value the efficiency and effectiveness of individual components above the ability of the system to deliver, will swiftly descend into inertia.
Understanding risk can help you decide between applying mode 1 vs mode 2
Beyond any immediate need for speed or experimentation, deciding how and when to apply mode 1 vs mode 2 can be considered as an investment strategy.
When the upside potential is greater than the downside risk, a more Agile approach can be warranted. On the other hand, when a higher than normal chance of a negative return is predicted, then a more reliable methodical approach is generally followed particularly when it relates to core or mission critical IT systems.
Is there an alternative to a pure bimodal IT strategy?
Given the difficulties in defining the tipping point between mode 1 vs mode 2, there is another option that can be considered; a multi-modal approach inspired by Dave Snowden’s work on Apex Predator Theory, which allows you to select from a range of approaches depending on where each product is in the innovation cycle:
- Emergent products (Early Adopter stage) require experimental approaches with multiple parallel strands
- Growth products (Early Mainstream stage) require iterative, customer-centric approaches with linear experimentation along a successful path
- Stable products (Late Mainstream stage) require cost effective approaches that value system flow, but don’t require rapid iteration
- Declining products (Laggard stage) require waste eliminating approaches that maximise profitability, coupled with an appetite for disruption
Allowing for a multi-modal approach allows the business-as-usual to flow, and makes for an organisational ecosystem where rich insights, process efficiencies and collaboration can thrive.
These benefits are particularly appealing for larger businesses for whom digital transformation is as much a defense strategy as a way of achieving a competitive advantage.