Insiders and outsiders will pitch you all day long on the value and virtues of big data and the industrial internet of things. People demand to know why you aren’t doing more with and for the IIoT. It’s enough to make you think you work for the IIoT, rather than the other way around.
Here’s a piece of advice: Make the IIoT work for you. Every time someone pushes an IIoT or big-data solution on you, ask three simple questions:
- What business results will this deliver? (Note: “Visibility” is not a business result.)
- Will this deliver relevant data or just more data?
- What do actual users need, want, and say?
The broader picture is this:
“Technology doesn’t achieve results; people do.” It’s old logic. In the right hands, technology is relevant, even critical. The same technology in lesser hands achieves nothing, or worse.
The best evidence of the IIoT’s infancy is that many conversations about it still focus on the technologies themselves. “Terabytes of data…in the cloud…machine learning…artificial intelligence.” It’s a brave new world that belongs to the innovators, the scientists, the technologists. But are they users? And if they’re not, what do they understand about users? Do they know enough to parse the relevant data from all the other data – the so-called junk data?
The accepted wisdom of the IIoT is that more information is better – that infinite information means perfect visibility and total situational awareness. Performance, safety, and efficiency are bound to improve, the logic goes, because more data means better insights, clearer understanding, cleaner interventions, and, eventually, perfect equilibrium.
The sheer scale of the information makes people shift in their seats, but what are the real implications? The CEO of a major global industrial company opened his speech at a recent industry conference with the following statement: “Next year there will be eight billion connected devices.” A hush fell on the crowd. It’s a big idea. But why is it important? Is it important? Will these devices tell us something new, something relevant?
Big-data scientists on Wall Street learned decades ago that correlations are dangerous things. They can be random, nonsensical, and consistent until they’re inconsistent. Betting on them has been risky. Wall Street learned that data without context – data in a vacuum, without precedents or without an understanding of the factors underlying and contributing to it – can be catastrophically misleading.
The IIoT will be different only when the data ties back to step-change improvements in performance – lower costs, higher output, and better safety – based on insights gleaned from knowledge that wasn’t available before the IIoT.
For a time, it felt like the IIoT heralded the industrial age of Aquarius: Variability would be tamed by common understandings of optimization and risk management. The algorithms, machines and computers would take over; waste and inefficiency would vanish; accidents would cease; and performance would settle around optimized production models. Disparities in performance across enterprises would narrow, too, and we would finally lay to rest the age-old debate about machine failure – the data would distinguish among operator, process, and design-induced breakdowns.
Here’s the problem: There are corners of the IIoT world where emerging applications make a difference. The same applications in other corners of the IIoT world make no difference at all. Understanding why couldn’t be more important because the IIoT really is different in at least one important way. IIoT data is playing a new role. Data isn’t just revealing how machines perform and fail; it’s revealing how they’re being operated. At massive scale, what used to be measures of mechanical integrity are becoming barometers of competence and culture. Knowing what went wrong has long been within reach. Increasingly, the IIoT can show us how it went wrong, and that is new.
IIoT data will challenge the world’s large, distributed industrial organizations because it will expose complex, ingrained, modified, partially adopted, and explicitly ignored practices throughout the enterprise. Early in the IIoT age it was easy to imagine that data was pure and uncompromised, that it offered clearer truths, surer knowledge, and undisputed facts. It had something of an idealistic, democratic, “the numbers don’t lie” appeal, until users discovered how data can have an agenda, too.