The Internet of Things has long been held up as the panacea to all that ails modern business. Real-time visibility and intelligent response to systemic anomalies have long promised to ruthlessly and autonomously eliminate inefficiency, delivering us into a new era of accountability, performance, and profit. The largest consultancies in the world have entire teams dedicated to “Digital Transformation.” Billions of dollars have been invested in products and platforms designed to collect, transport, and analyze the data that will drive this change. But for some reason, very few organizations are able to capitalize on this opportunity and make the promise real.
All the ingredients are there. I can connect to just about any device, data source, or external system and ingest its data. I can easily move that data around a distributed system of hundreds or thousands of computers. I can craft machine learning models based on that data and use them to power intelligent decisions without human intervention. And I can easily visualize the health of these systems using self-generating dashboards built just for this purpose. Most of these things I can do without writing a single line of code- just drag and drop all the pieces into place and BOOM- it works.
So why isn’t the future as shiny as we were promised it would be? It turns out that the answer to the question is in the question itself:
“Why can’t my human IoT architects design a solution that ruthlessly eliminates all the human inefficiencies from my organization? -Everybody Who Has Started, Then Abandoned, an IoT Initiative
“We can’t solve problems by using the same kind of thinking we used when we created them.” – Literally Albert Einstein
Focus on shareholder value will always derail projects with ambiguous scope and unknowable return. IoT projects are very much a leap of faith. And the monolithic architectures that present the rosiest picture of long-term ROI are also the ones that come with the biggest price tags. If the tendency for solution architects to present comprehensive strategies is at odds with the very concept of progress in an uncertain landscape, how do we break the cycle of false starts, budget overruns, and the onset of existential dread? The following principles are key to getting IoT initiatives off the ground and moving them forward:
1. Start Small
“Boiling the Ocean” is a phrase used to classify those initiatives that are so far-reaching that they are de-facto unachievable. Digital Transformation can very quickly turn that direction- especially if you bring in a giant consultancy to handle it. You can burn your entire IoT budget in exchange for a nice drawing of your end game- at which point you’ll be presented with an implementation scope that all but eliminates the promise of near-term ROI.
Starting small can mean an internal team, a new contract hire, or even the professional services arm of an IoT solution provider. Usually, it’s best if it’s a combination of all three.
What you’re after are individuals with domain knowledge and skin in the game. People for whom a successful pilot project will lead to more work, larger scope, and more budget based on demonstrable ROI. Consulting practices with thousands of employees are perfectly capable of architecting a solution, but subsidizing their overhead is a terrible way to optimize yours.
The scope of your initial project should focus on a single opportunity comprised entirely of systems you can access and control, and with a scope of no longer than 4 weeks. By way of example, here are some Proofs of Concept that I’ve worked on where we were able to deliver within that timeframe and very much on budget:
Optimizing thermostat settings for conference rooms based on scheduling software, number of attendees, current room temperature, current outside temperature, and time of day. Notifying a fleet manager that one of their vehicles has deviated from thresholds for driver behavior, scheduled maintenance, or location boundaries.Scheduling preventative maintenance for a large industrial machine based on current, vibration, and temperature sensor readings. Optimizing outside sales staff productivity by mapping time on site to actual bookings, and developing best practices based on those insights.
There are very likely proprietary solutions that solve each of these problems, but the point is not to buy a series of proprietary solutions. Instead, the objective is to become comfortable with the IoT products and platforms that allow you to ingest, analyze, and act on your system’s data on your own behalf. That comfort becomes leverage as you identify the next system for optimization. Limiting the scope of these initiatives will build the confidence that future initiatives can also succeed. It’s all about building a track record of being able to deliver.
2. Make It Pretty
Technology sales are ostensibly- rationally- about time, money, and quality. But the reason most deals close has little to do with any of those things. Instead, it usually hinges on your ability to make your customer look like a hero. Not the company they work for- the actual human being you’re selling to. They want the same thing we all want- to be good at what we do, and to be respected for the recommendations we make.
In IoT, making your customer look like a hero means making it easy for a layperson to understand what the system is doing, how to manage it, and how it can expand beyond its current scope. In theory, this means considering all the features, extensibility, and enterprise-class functionality before eventually choosing the right software. In practice, it’s even simpler: people buy dashboards. After all, the best way to show them what they’re buying is to show them what they’re buying.
“Will no one rid me of this turbulent IoT architect?”– Henry II, had Thomas Becket been asked to sensorize Canterbury Cathedral.
I have enough experience to know that no matter what I want to connect to, I can probably do it. Getting the data is never the problem, nor is moving the data. Presenting the data in a way that’s easy for your customer (and even more importantly, their boss) to understand is the easiest way to show potential value, but it’s usually the last thing you work on. I’m to the point that I rarely draw complex architectural diagrams until after the deal is closed. I have reference architectures that suffice for most technical discussions, and always make it into the annex of my presentations. But a great UI that people can click through and “feel”? It takes very little time to create, requires very little technical discovery, and closes ten time the number of deals in a fraction of the time.
It may seem a little backwards, but we’re visual creatures. And if you’ve ever had to do technical due diligence with an army of internal developers, you know that those meetings aren’t fun for anyone involved. Selling the joy of control- even if it’s just simulated joy- is way easier than selling a functional specification.
3. Plan for Scale
“Start Small” and “Make It Pretty” feels like I’m failing to appreciate the value of a system that captures and analyzes everything. But this article isn’t debating the value of a comprehensive IoT strategy- it’s proposing a less daunting framework for getting from here to there. Single PoCs and a great dashboard for each of them are no match for a Master Data Strategy.
Most domain experts can intuit how inputs to a system can be used to predict outcomes. That’s what experience gets you. But the scale of IoT means that it’s unlikely that such intuition will serve your organization going forward. There are some fourth derivative analogs in every system that will eventually be included in optimization strategies, but nobody will be able to tell you what they are without leveraging complex algorithms or machine learning models to identify them.
“Start Small” and “Make It Pretty” are designed to lower the impact that “unknown unknowns” have on your first steps into the realm of IoT. Once you get the first few under your belt, the next few become easier… but it’s important that you know how to present this landscape of captured data in an accessible fashion. Having lots of data you can’t mine for insights, also known as dark data, is WAY worse than having too little data to mine for insights.
Planning for scale includes understanding what sort of tools you’re going to be using to analyze and present the data, what sort of technology you’re going to use to store it, and how you’re going to accommodate new sources and types of data as you integrate new systems into your IoT solution. This is where having a small internal team with a comprehensive understanding of your company’s business is extremely helpful- even if you employ external consultants, your internal experts should always serve as guides to shape those efforts.
IoT is inherently complex and difficult, but the promises it makes are not unattainable. The technology and tooling exists and is production-ready right now. Guided by the strategies above, success in IoT can morph from a cost-prohibitive monolithic initiative to a series of small, hardscrabble victories on a dynamic playing field, where failing is just as valid an outcome as winning, and victories become building blocks for the future.
“The universe is infinitely large, but we tend to launch rockets at one planet at a time.”- Me, Trying To Get People To Focus