Is now the time to invest in the Internet of Things, or should we wait a bit longer?

The hype around the Internet of Things (IoT) and its various potential uses has been around for a few years now. Each year we hear IoT companies say “this is the year of the breakthrough”, but it seems there’s a lot more “talking” and a lot less “doing”. So… is IoT mature enough to bring real, measurable value to those who actually use it?

IoT is nothing new, some might argue. And they would be right. Although the term itself was introduced in 1999 by Kevin Ashton of Procter & Gamble, later MIT’s Auto-ID Center, the concept of connecting machines to the web was even older. It was in 1982 that a Coke machine was connected to the Internet at Carnegie Mellon University (CMU), allowing it to report its inventory and drink temperatures to operators. One CMU alumnus told me lately they were remotely monitoring things quite a few years back. Being a new Pittsburgher, I must say this is something to be really fond of… and wonder why didn’t it catch on and scale up quicker…

I mean, the potential of such solutions is obvious – huge process automation and simplification, elimination of potential errors, better supply-demand management, failure prediction… applications for IoT and the likes are numerous. So what was holding up widespread adoption?

In this case listing challenges is pretty straightforward: sensor hardware, computing capabilities and the Internet itself. Sensors were big and pricy and it was difficult to design a piece of machinery with all those added elements, because it suddenly got a lot more expensive while not giving much of an added value. If they were installed, however, we did not have small computers that could utilize the data they could have produced. Fitting a ZX Spectrum or Commodore 64 (two popular home computers at that time) wasn’t really an option for items such as alarm clocks, refrigerators, packaging machines… or basically anything. On the other hand, do you imagine using a modern NASA supercomputer for calculating your travel route to the nearest grocery store? I’m guessing the price tag would be at least 5-digit. Yes, it was that expensive. Then there was the Internet. In 1982 TCP/IP was actually standardized, but it wasn’t until late 1980s and early 1990s that commercial Internet access became a more common sight. It still was based on telephone lines and modems (does anyone remember 1kbps-or-less modems these days?), so even using the web itself could cost a lot of money.

There we are. Three big reasons why the idea didn’t catch on in the 80s.

But times have changed quite drastically, haven’t they? Sensors are sometimes smaller than the pin of a needle, computers now easily fit into the size of an arm watch and broadband Internet is literally everywhere. So is this the time to actually adopt the ideas from CMU into day-to-day businesses? Let’s consider a few cases.

Power plants and Distributed Control Systems

Imagine you are responsible for running a large power plant. You employ hundreds of workers who oversee all of the processes going on inside it and report any incidents or deviations to their superiors and, eventually, to you. You get a very detailed, complex view of how such a critical piece of infrastructure is doing in real time. You could say you are flooded with information that only highly trained engineers can make sense out of.

Luckily, you invested your company’s money into something called a Distributed Control System (DCS). A system like that allows your plant to run from a central operations room, while distributed controllers handle processes. What you get is an overview of the operations of the entire plant, while your engineers get what they want and need to make sure everything is running smoothly – those details you would have been flooded with.

Systems like that have existed for a few decades now. While growing in terms of complexity and capabilities along with development of both hardware and software infrastructure, they are deployed in practically all power plants around the world, whether the manufacturer is Emerson, ABB or other origin. Enhanced with modern predictive capabilities, they allow these businesses to not only function with less (or no) failures. They can even help predict actual electricity demand and fine-tune the plant’s efficiency to match it, while reducing environmentally harmful emissions through optimization of operational parameters.

You’re able to see operational parameters and sensor readings from various pieces of machinery deployed in a plant. You’re able to use predictions and optimization to get the most out of every ounce of fuel. You yourself as well as your employees are not flooded with information, allowing everyone to focus on doing what only human beings can do – true analysis and creative thinking. Doesn’t that sound a little like IoT?

Some might say IoT is already in full-scale deployment, as the above example from the energy sector displays. However, this is not the case for other industry sectors. While some economy-critical industries, such as energy, have relied heavily on technology for a number of years, others have lagged and are yet to seize the opportunity. Some of them are trying to catch up, usually doing pilot and Proof-of-Concept (PoC) implementations.

Proof-of-Concept flaw

Every couple of days I get new mail on some PoC or pilot installation of IoT somewhere around the world. At that rate, I sometimes wonder, are there any more companies which have not implemented IoT throughout their operations? Mind you, all of them say they are so successful in business cases they are applied to.

The answer is “yes, quite a few”.

Pilot implementations and PoCs are often used as a way to evaluate the potential a given solution has, including return on investment (ROI), cost-cutting and efficiency-increasing capabilities, etc. The biggest problem of those is, however, their scale. IoT promises to change the way we manufacture products and manage processes, but on a global scale. It is very difficult to measure any gains when an implementation covers only one machine, while others inside a facility remain disconnected. We cannot digitize the process that way and therefore we cannot optimize it. It’s like applying ABS (anti-lock brakes) to only one wheel in a car – usually one of the back wheels, as they are under less stress while braking – equivalent to choosing non-critical systems for PoCs and pilots. How efficient would such an ABS be if we needed to suddenly stop on snow to prevent hitting a child that’s running through the road ahead of us?

Since it’s so difficult to spot real value, it is even more difficult to justify further investment in scaling up any IoT solutions and implementations. A 3% Overall Equipment Efficiency increase might cause engineering shift supervisor to cry tears of joy, but unless translated into millions of dollars of annual revenue increase, it will never convince the guys holding the money to put it on the table. Don’t get me wrong – pilots and PoCs are still useful for shaping the final look and application of the solution. It is only dangerous to consider them the finish line. Successful implementation of a PoC or a pilot should actually be understood as reaching the start line, with the race just about to begin.

If we are to consider measuring value introduced by IoT, we must scale ideas larger. We must move from the concept and pilot phases to full scale deployments as quickly as possible. Adding more and more elements to the system will not make it less effective. On the contrary – the more information we feed into those Machine Learning and Artificial Intelligence optimization algorithms, the better they become.

To be able to see at first glance the value from implementing a robust IoT solution we need to apply it to a broader enterprise – be it facility-wide, corporate-wide, or similar. We need to add that ABS system to all wheels on our car, so we can truly experience the shorter stopping distance as it was designed. We need to scale the solution… and scale the investment.

Up-scaling IoT – necessary elements: IoT platform

Since PoCs and pilots usually refer to a very specific case, they quite often utilize home-grown or small-scale solutions for managing all the connected “things”, as well as the information flow and, sometimes, adding a bit of extra analytics on top of that for flavor. While it is sufficient for a given case with little or no variety, more complex implementations might provide a bigger challenge.

If you’re hooking one machine (or a piece of it) up to the Internet and whatever IoT platform of your choice, you can usually implement some basic capabilities and automation pretty rapidly. But that one machine, developed by one manufacturer and sold by one retailer doesn’t necessarily have to be the only equipment you might be willing to connect. Consider hundreds or thousands of devices, created by dozens of manufacturers according to various standards over multiple years. Things get more tricky, even in terms of simply establishing communication between these devices and the platform, don’t they? Now, consider utilizing this heterogeneous data set to do any sort of optimization or prediction. I can ensure you that without an army of data scientists you’re not going to get anything out of it.

Sounds like a problem? Sure does. That is why it is of the utmost importance to choose the right tools when starting your IoT journey. An IoT platform which enables you to develop and scale quickly, while allowing you to connect to basically any industrial piece of machinery out there would be a wise choice. I’m not advertising any one of them, as some are better suited to be used for given cases, while others excel at different ones. I can only tell you it’s not without reason that biggest and most reliable research organizations name Amazon’s AWS, Microsoft’s Azure and PTC’s Thingworx as leaders of the pack. In fact, these companies are often working together (with each other) to make sure the dream of IoT turns into reality, so it’s a safe bet that if you’re working with one of them, you might be utilizing everyone’s achievements.

OK, we have an IoT platform of our choice. Now… how to put it in place and make sure it’s running at full capacity, so the enterprise can see efficiency boosts and cost cuts as soon as possible? How to actually use this tool for the benefit of the organization? We need another element to achieve that.

Up-scaling IoT – necessary elements: Partners

Businesses are different from one another. They follow their own processes, create different products, use different component suppliers, and so on. If we tried to make them all operate using a single business model, most of them would fail quickly and miserably. When it comes to IoT it is exactly the same story.

There is no generic know-it-all AI or Machine Learning algorithm that can optimize everything and everywhere efficiently. Each and every case is different and IoT solutions, even the most mature ones, have to be adjusted and fine-tuned to fit and mesh with the environment they are in. They need to know what machines they are talking to and in what “language”. They need to know what are the optimization goals and what can we use as a trade-off to increase our desired values. Finally, they need to be configured and trained using case-specific data, so that they are able to respond to parameter value changes the way we want them to, not the way some other company wants them to.

It’s been time and time again I heard companies saying “we have it all, end-to-end, one-stop-shop for IoT and changing your organization from A to Z”. Doesn’t that lead us to the “single business model for all” problem I mentioned a little earlier? What if I don’t want to change my whole organization to fit a model provided by some know-it-all outsiders? What makes them say they know my business better than me?

If you’re really trying to help companies solve their problems, you need to truly understand the businesses they are in. You need to understand their pains, know where potential for improvement is hidden… basically you need to speak the same language as your customers. Apart from being a technology guru you need to be a subject matter expert, so to speak. Or at least have one on your team.

Since it’s nearly impossible to have subject matter experts covering all aspects in all possible industries inside one organization, it would be wise to team up with others. One company providing cloud hosting, another providing deep knowledge on sensors and machine-to-machine connectivity, another delivering best-in-class software solutions – all working together with experts (including ones inside the customer’s organization) on solving real problems, and not creating new ones. That’s the best case scenario. Is it achievable?

Sure thing. You just need to know where to start. Some more significant vendors form alliances or partnerships, creating whole ecosystems of companies working together, supplementing each other with skills, knowledge and IP on various aspects. Such is the case with PTC’s ThingWorx platform – it’s hard for me to count all of the companies partnering with it, not to mention naming them. Of course there are standouts among them. I am lucky enough to be working in one such standout company, which is directly involved in development of the platform itself. Who can you imagine anyone better suited to deliver it than people who actually create it?

I’m not saying I have all the answers. I’m not even saying I have half of them. What I am saying is that I have the answers from the IoT software perspective. And I know where to look for to find all others.

Up-scaling IoT – necessary elements: Hardware

I’m not a hardware expert, so I’m not going to dive into this one. As I just said, that’s what I have partners for, right? One thing I know for sure: a brief browse through stuff available on the market tells me hardware these days can be both miniature in size and quite cheap.

Is enterprise-wide IoT that expensive?

Justifying any investment in front of some decision boards may be difficult, to say the least. I’ve been there, trying to convince shareholders to invest their money into development of a few software solutions that I then thought might be revolutionary and make us the next Apple/Google/Microsoft/ [insert_your_favourite_tech_company_here]. I needed reliable estimations of the effort (work and money) needed to complete those projects. I needed justification of why (and if) customers would actually buy it. I needed a thorough market research for similar solutions, their pros and cons. I basically needed a business plan, along with quite a lot of technical details of the future solution as well… But it’s probably similar to what many entrepreneurs and startup companies faced throughout recent years. Baseline here – it’s nothing new. It’s expected. That’s the reality entrepreneurs such as myself have to face.

IoT is no different. Having already established there is potential in using it, validated initial ideas through PoCs, the next step is to convince someone to invest their own (or their supervisors, or their shareholder’s) money into going enterprise-wide. Kicking off with the real thing – the hardest part of any project, not only IoT. To get past that point, we need to prove there’s real value.

I attended a conference recently where Dr. Mehrzad Mahdavi of the Society of Petroleum Engineers said that a 1% increase in uptime of operations on a single oil field will lead to creating between $20 and $30 million annual value. That’s from a 1% uptime increase! Now, IoT has in many cases proven to increase uptime by 5% or more. $100-150 million annually. Based only on operations. It doesn’t even take drilling into account. How’s that for a number?

Is your organization not as big as those enormous Oil&Gas companies? Let’s take a smaller case: a power plant in Poland. It utilizes software called SILO, which uses Machine Learning to optimize operational parameters to increase efficiency and reduce harmful emissions. In addition, it can predict a coal mill jam with a 30-minute advance, so you don’t need to keep a large number of servicemen on constant alarm. Value? Over $300,000 annually for a 225 MW block (not a giant).

Not your type of business? Let’s try something different: an Italian packaging machine manufacturer is applying IoT to predict when to replace a cutting blade to prevent losses thanks to arranging component replacement at the right time within normal maintenance activities. According to their calculation each successful unplanned downtime prevention saves their customers 13,000 euro. As far as I am aware, they are on track to not just reach their targets, but even exceed them.

These are just a couple of examples on how IoT can save money or create additional value in various applications. Of course, convincing decision makers (and budget holders) to invest in it will not be an easy thing, but maybe such numbers help them make up their minds. The more such examples, the better.

To invest, or not to invest?

That is the question. The answer might not seem obvious at first sight – there are still relatively few widely known working implementations of IoT solutions which managed to reach broader public. There is a need of something similar to what Pokemon GO or Apple iPhone X ARKit was to Augmented Reality.

What is certain, though, is that all components required to create IoT solutions which can generate huge value are already here, mature enough to pick them up and put them together into a fully functional system. Miniature, low-cost and low-energy sensors are here. Internet and wireless network connectivity is here (and everywhere else). Software solutions, including IoT platforms, are also here, though some more mature and more useful than others.

Technology finally caught up to expectations. The hype can now be justified. The only challenge now is to find a good partner, who can help guide an organization through fully embracing all that IoT brings to the table.

What I can help you with, at least, is the last bit.

Do you have a similar, or a different view on the matter? Please, share your thoughts in a comment.

If you would like to discuss IoT in more of a private environment, please reach out to me directly.