Approximately 70 percent of executives responding to a recent McKinsey Global Industry 4.0 Expert Survey said their organization considers Industry 4.0 as the top priority. Given that a 2015 McKinsey Global Institute report predicted that the Internet of Things—which is just one out of a wide range of Industry 4.0 technologies—could generate $1-3.7 trillion in value for factories by 2025, the business case for prioritizing Industry 4.0 is clear. But the 2018 survey found that progress remains unexpectedly slow.
The reason: Pilot purgatory. As they endeavor to capture trillions of dollars in value, companies worldwide are testing their priority Industry 4.0 use cases by launching an average of about eight pilots. Companies in China and India are pushing even harder, at an average of ten pilots (Exhibit 1).
But only 30 percent of the pilots end up reaching scale across the entire organization. Put differently, companies are failing to capture value from 70 percent of their pilots (Exhibit 2). Moreover, the pilots are too slow. Some 85 percent of the companies surveyed spend more than one year in pilot mode, while 28 percent spend more than two years. For companies to begin to capture the predicted trillions of dollars in value, they must move more pilots to scale faster.
What is keeping companies in pilot purgatory?
Companies usually head into Industry 4.0 pilots with the intention of expanding them company-wide. Too often, however, the pilots are essentially one-offs that fail to anticipate critical requirements in expanding from self-contained pilots to enterprise scale. Industry 4.0 (and IIoT) therefore remains in purgatory.
Through our work with clients on the manufacturing front lines, and in a host of realistic learning environments, we have identified three steps that help expand a successful pilot across an enterprise.
- Start with the appropriate executive sponsorship and operating model, defining the roles and accountabilities for the business-unit, IT, and operations teams, along with the digital and analytics experts—all of whom need to work together and acquire the right talent.
- Choose use cases carefully. Start by identifying the ones that create the most value and are easiest to implement; these can serve as lighthouses to demonstrate technological capabilities and potential impact. Embedding these cases into business routines then ensures that analytic insights don’t end up only in the back office but are used every day by the managers, supervisors, and operators who make the decisions that will enable the plant to run better.
- Ensure the foundation IT/OT convergence can scale.
As important as the first two steps are, they’re also conceptually simple, if sometimes a challenge to achieve in practice. It’s the third step in IT/OT convergence that can turn into a high barrier. The complexities of IT/OT technology are genuinely difficult to tackle, especially because on most factory floors, the links between IT and OT tend to be immature in unique ways. We refer to this as the last-mile IT/OT problem.
Understanding the IT/OT ‘last mile’
Within the IT/OT last mile, five issues combine to prevent manufacturers from scaling Industry 4.0 and IIoT.
- Complex, heterogeneous environment: A typical medium-size plant has over 200 individual pieces of equipment, purchased at different times from a variety of suppliers. Each supplier has its own automation capabilities, hardware and software platforms, and communications protocols. This context makes it very difficult to collect, integrate, and contextualize data. Moreover, some equipment manufacturers offer data-analytical insights as a value-added service, creating further potential wrinkles in data accessibility. And the industrial-automation stack of hardware and software is itself highly complex, comprising everything from manufacturing-execution systems (MES), maintenance-software packages, production-scheduling software, and distributed-control systems, through to enterprise resource planning and product-lifecycle management.
- Old unconnected machines: It is not uncommon to find aging, uninstrumented equipment on the factory floor, as replacement cycles for heavy assets often last decades. Difficult or extreme operating environments, such as the intricacies of a paint shop, are further barriers for data capture. Finding ways to instrument these assets is therefore an imperative for driving IoT at scale. Also, equipment that has already been instrumented frequently relies on legacy software systems, such as late-90s desktop-based machine-control systems, with data stored offline in spreadsheets. At one site we examined, for example, 90 percent of data from the MES was purged every 30 days (except for parameters required for reporting to major customers).
- Security concerns: Some manufacturers are reluctant to move data to the cloud, preferring an IoT solution that offers a mix of sensor-based edge analytics for critical process control and on-premises solutions for predictive analytics.
- Limited last-mile OT capabilities at scale: Fundamentally, delivering Industry 4.0 at scale requires the ability to extract, interpret, and harmonize data from disparate systems that were not designed to work together. When building towards an IIoT platform, whether internally or with a third party, organizations must ensure that they can find capable OT service providers—ones that can support plants in locations across a wide-ranging footprint while harmonizing the collection (and connectivity) of data captured in the plant from PLCs, sensors, and historians. Without this step, the best analytic models and user interfaces will not have the data required to deliver the value expected from them.
- Poor collaboration between IT and OT: Often the problem is due to the historic lack of connection between IT activities and OT activities, typically by onsite manufacturing process engineers. OT’s goals usually focus on current performance, business-as-usual predictability, and avoiding disturbances to a system that works; IT favors security and trusted technology providers, often those that have a large installed base already. Different levels of focus on user management versus machine management yield very different problem-solving approaches. Consequently, getting the IT and OT staffs working together from the start is a must.
Being aware of last-mile IT/OT issues will help build a technical foundation that supports the full-scale roll-out of Industry 4.0 pilots. But careful planning that addresses the people issues of culture, process, and prioritization is equally important. Indeed, in our work with manufacturers, the specific success factors we’ve seen for sustained value capture point more to the human side of the equation: for example, leaders who are fully engaged, vendors who act as partners, and talent who have (and develop) the right skills.
Tackling the IT/OT convergence challenge
Manufacturers should reflect on a few essential questions that help address the last-mile IT/OT challenges:
- Do we have a complete understanding of the potential IIoT value chain, and the gaps we need to fill?
- How mature are our plants with respect t0 last-mile IT/OT, and how can we better understand their IT/OT challenges?
- How should we balance investment in additional data collection versus in getting more value from the 99 percent of our current data that is likely sitting unused?
- How can we use previous industrial-automation investments, in systems such as MES and DCS, to help achieve speed and scale in building our IoT architecture?
- To maximize our returns, how much of our last-mile IT/OT investment should be at plant level (sensors, connectivity, edge devices), and how much in broader IoT architecture (platform, cloud, application layers)?
- What vendor ecosystem would best serve our needs?
Addressing these questions in the IT/OT context takes isn’t easy, and we’ll discuss them in more detail in our next post. We’ll also provide guidance on designing pilots that not only create tangible results, but also provide a foundation for a repeatable, timely rollout process that delivers value in months rather than years.
The authors wish to thank Ani Bhalekar, Karel Eloot, and Bodo Koerber for their contributions to this post.