What is a digital twin? I set out a few months ago to answer that question for myself and came up with a less than perfectly clear answer. In fact, the answer to that question almost felt like the answer you might get to the question “How long is a piece of rope?”. It certainly can have a specific answer, but it’s entirely dependent on the situation. A piece of rope has a specific user, a specific job, requirements for load, durability, safety, etc. A digital twin, likewise, should have a specific user, targeted outcomes, a budget, and other parameters that allow us to truly answer the question of what a digital twin is or is not.
In this blog, I’m going to attempt to provide at least some high-level definition around what various forms of a digital twin might be. But more importantly, I will explain how to think about them as a product unto themselves. As product development consultants, we are always looking to apply tried and tested product development methods to achieve product/market fit. It is my assertion that digital twins absolutely fall into this space.
When you hear the words “digital twin,” what comes to mind? Is it a complex mashup of buzzwords like Machine Learning, Artificial Intelligence, Internet of Things, Mixed Reality…Metaverse? Is it some magical recreation of a physical object in cyberspace (Tron anyone)? A common, but less-than-specific definition of a digital twin is: a digital representation of any aspect of physical objects or processes. A piece of rope is as long as you need it to be.
Here are some examples of digital twins in several contexts
Healthcare – Digital twins of organs, a genome, or a single cell can be generated, allowing researchers to experiment virtually with innovative treatment and surgical options.
Manufacturing – Digital twins of a manufacturing line can be used to analyze the process and important performance indicators in order to identify new ways to optimize production, reduce variances, and help with root-cause analysis.
Defense – Digital twins of power and propulsion systems for large, expensive surface ships that allow the Navy to discover ways to maintain more uptime, lower maintenance costs, and extend the life of the asset. These insights can be used to optimize a mission based on conditions or inform the acquisition of newer components or assets that improve reliability.
Transportation – Digital twins of a trucking fleet that enable fleet managers to predict and optimize fuel consumption by analyzing things from driver behavior to tire pressure to weight distribution.
Each of these examples does represent a fairly complex, and potentially expensive set of investments in time and technology. At the end of the day, these various digital twins are a collection of software, hardware, and data purpose-built to solve a specific business or mission problem. To me, that sounds exactly like many other products and as such we should apply the best practices of product development to building them. There is no way I will attempt to describe all these practices here; but I will attempt to lay out a course for approaching the building of a digital twin.
At a very basic level, any product manager should begin the journey by learning about both the problem they are solving and the solution they are creating: problem discovery and solution discovery. Let’s take the example of the transportation digital twin. We want to build a digital twin of a truck. A product manager will need to do the work to understand what problem or problems they are solving as well as how and what to build as a solution. We can apply some standard practices here to help us along the way.
Problem Discovery: For problem discovery we can look at a handful of tools that most product managers would be familiar with.
Solution Discovery: For solution discovery, we can leverage the concept of a Minimum Viable Product. We progress through a series of steps that initially help us learn about what and how to build the twin. We then transition to a series of releases that continue to add value to the user and customer which make economic sense for the business.
The reality of product development is that we are never done with discovery, and as long as the product exists we are never done with delivery of new enhancements/features. The same is true with digital twins. As new data is available, new models and model enhancements are discovered, and new computational capabilities are available, we will need to continue to evolve the digital twin product. Pulling again from best practices, digital twins should follow a Dual-Track Agile approach where we empower cross-functional teams to take on new business/mission challenges by executing problem/solution discovery and iteratively delivering MVPs to deliver value to the user.
Just like any other product development effort, you don’t have to follow these practices. You could, of course, just go all-in. You could collect as much data as possible, as quickly as possible, from as many things as possible. Then you could throw brilliant data scientists and engineers at it and hope they manage to create something useful for a real user that is technically viable, and that generates enough value to justify the time and money you spent. But maybe you just constructed a rope to zip line from the moon to Earth when all your customer really needed was a length of dental floss to get out a stubborn piece of chicken.
Ready to take your product development to the next level with Digital Twins? Contact us for a personalized consultation and discover how Digital Twins can empower your innovation.