We’ve reached a point in industrial IoT where the hardware is mature. Software can be customized to the use case. Our AI tools have become mainstream. Organizations may be paying to maintain a large sensor population when a much smaller, strategically selected population could deliver more operational benefit. How would you know?
Data volume and data management cost are not necessarily proportional. A relatively small number of badly behaved devices could generate a disproportionate amount of work (…the potential 20% that brings 80% of problems).
The 80/20 of commercial value
What if 20% of IoT applications account for 80% of the actual economic benefit? A water utility might obtain most of its savings from leak detection and remote meter reading. A logistics operator might obtain most of its value from asset utilization and theft prevention. A manufacturer might obtain most savings from avoiding the downtime of a handful of expensive equipment failures.
Find me the top 20%
In these circumstances, adding another layer of instrumentation may be less valuable than improving the accuracy, reliability or integration of the existing system. The fact that you are collecting more data, also means you have to store more of it, parse more of it and cross reference more data points.
You could pare it down to the key data points that deliver the most operation value? Maybe we’re being redundant, but “less might be more.” How much less is a different question.
We don’t assume Pareto distributions because it is an attractive mental model. Neither do we discount the cascade effect of different events creating downstream havoc. A temperature sensor in a warehouse might be unimportant until a refrigeration failure threatens a shipment. The unit value is determined by the consequences of the event it detects, not just its normal contribution to daily operations.
The question to research…
“In industrial IoT deployments, what proportion of connected devices, data streams and applications account for the majority of measurable operational and financial value, and what proportion account for the majority of ongoing management costs?”
Mr. Holmes
That gives us a researchable question rather than a fashionable management slogan.
And one further thought…
The next competitive advantage in IoT may belong to the company that knows what NOT to connect. This idea suggests selling better designed, appropriately instrumented systems, rather than simply more units of hardware. You can bring in AI machine learning and bespoke software elements of next level systems to deliver “tailored” solutions. Sounds like a mature product.
The result might be context specific or the distribution might change significantly between industries and deployments? Either way, it would be interesting to know if benefits and costs are evenly distributed.
