Implementing an IoT solution for your business doesn’t have to be a beast of a task if it’s planned correctly. Here’s the 5-step roadmap we use to design and implement IoT for businesses, with bonus content on keeping your data safe during the process and calculating the ROI.
One of my favourite quotes of all time is: "Genius is the ability to reduce the complicated to the simple." Problems aren’t always simple, but the people and processes who can make them as logical as possible are the real geniuses in my book.
Implementing an IoT solution for your business doesn’t have to be a beast of a task if it’s planned correctly. In fact, you’re the one who can help make it as simple as possible because few people know the frustrations and problems in your business better than you do - you just need a little creativity, an understanding of the opportunities and cost savings, and a plan of action to make it work.
Here’s the 5-step roadmap we use to design and implement IoT for businesses, with bonus content on keeping your data safe during the process and calculating the ROI.
1. What processes can I automate?
First thing’s first - work out where IoT is going to benefit your business. You want IoT in places where you’ve identified a need to create data so that you can then react from it. You can do this by identifying which of your business processes rely on recording laborious observations of objects in the real world.
As a couple of examples from two different types of businesses:
Maintenance checks on industrial fans: Many factories use large fans to ventilate the space. These fans are checked regularly for abnormal function (or: someone has a quick look at each of them every couple of weeks to make sure they aren’t “rattling a bit”). Why send a maintenance team around to each fan, with pens and clipboards (or even a mobile app), when a vibration sensor on each fan can phone ahead for you? Not to mention more regularly, more accurately and with less subjectivity.
Stock take: These two words meld together to form every retailer worker’s worst nightmare. Many stores rely on a barcode system to keep track of the quantity of available stock, and these digital records become so corrupted by humans over time that humans must periodically count each item to calibrate the system. What if sensors could count each product for us? Instead of doing this over three hours each month, what if they could do it immediately, every day? If we know where each type of product is supposed to be, we can achieve this with proximity sensors or weight sensors.
We make sense of things with our senses.
The reason why IoT yields automation opportunities is because it can replace many tasks that have relied on the use of human senses: hearing, touch, sight, smell and taste. The first example implies a device that can feel the way a mechanism is vibrating. The second example suggests a device that can feel the weight of a group of objects and make a smart guess as to how many objects are sitting on it, or see the distance between the back of a shelf and the nearest product and make a smart guess as to how many products are missing.
It can be hard to find these automation opportunities from inside the business - if you have trouble, get in touch with an external expert to break orthodoxy and evaluate your business for productivity opportunities with a new lens.
Okay, so I know what I can automate. But what should I automate?
Before you figure about how much time you’re going to save with your IoT solution, you should start by figuring out how much time you’re currently not saving. Use the following guide:
From this, you can start to build a business case and work out the core functions it needs to perform and the requirements across the layers of IoT – sensing (collecting data), connectivity (transferring data), business intelligence (mapping data to the process).
An example of this is Redbull automating orders when its fridge stock levels get low. One way of doing this would be by creating IoT hardware that measures the weight of the fridge, triggering an order when the fridge falls below a certain weight, or by measuring the distance between the back of the fridge and the Redbull cans, triggering an order when the distance between is large enough.
PROGRESS: At this stage you should have identified the problem, the cost savings, and the IoT process.
2. Identify and test the potential risks
Once you understand the core requirements, you want to work out which part of the process is the highest risk technologically. In general, this is the part of the process that you haven’t done before or is the highest risk to achieve.
First - identify the data you need to collect and what informs/triggers this process. Then you can work out how you can obtain or create that data in different ways. In the case of the Redbull stock in a service station, the missing data I want to test is in the changing weight of the refrigerator.
I usually like to break it down to a function level and go, "OK, so with sensing the weight of the refrigerator, where have we seen automatic weight sensing before?" You could look at automated truck weighing bays, and others using a similar technology. Relating it like this will help you understand the risk more meaningfully so you can work out what your tests are going to be.
Here you’re testing your biggest barriers, not the stuff you’ve seen done before. If I tested the whole thing it'd be $100,000 at a test level. If I test just the core risks, it could be $20,000.
There's a big difference between taking a gamble on $100,000 of experimental stuff and $20,000 on the stuff that will really make a difference. It also changes the testing timeline from months to weeks.
PROGRESS: At this stage you have tested the big-ticket risks in your process.
3. It’s prototype time - plan and produce it
Now you can plan and produce your prototype, based on the risks you’ve identified in step two. You don’t need to build out the software or hardware end to end, you only need to build and test the parts that are new to the equation.
The first and easiest step is to start IoT developments from the ground up rather than using heavy frameworks and prebuilt “all in one” or “one thing does all things” solution. The more general it is, the higher the overhead on data and energy, simply because they are providing for a number of uses and scenarios rather than your specific one.
There are always three core considerations in IoT. The first is data management. The more data you produce, the heavier it's going to be on your communications network, so your costs increase because you’re transferring more data.
The second is energy use. Just like a laptop, the more powerful it has to be, the less battery life you get. It’s the same with IoT. IoT becomes really frustrating when you have to wire it into walls because it makes implementation long and difficult. So you want to keep your data to a minimum - and you don't want to have to plug it into a wall.
The other core influencer is the communication system - whether you’re running the chip set for wi-fi, 3G or 4G, or something more substantial like 6loWPAN, Zigbee, LoraWAN or Bluetooth (if that sounds like gibberish, check out some explanations here). You have to pick the right thing and aim for something that uses the least energy - that gives you the most flexibility.
Prototypes should be kept under $30,000. If it's over that you're testing too much and you're not going to be able to control it or learn as much from it.
PROGRESS: You know how your prototype stores data, how it is powered and how it talks to networks and to you.
4. Your minimum viable product and the all-important pilot
Here’s where it all starts to take shape. You now have the thing that collects data and you can feed that into the data platform. It’s time for the business intelligence system.
The business intelligence system is where you make sense of the data - it’s where you get to see how often things are occurring, then work out how to make predictions from it. You never want to just get data and have it go into a visualisation or dashboard. You want a pilot. So you need to ask yourself at this stage: “Can the stuff I tested take the extra steps to integrate into my systems and be an actual switch for business processes and business change?”
A lot of people get it wrong and focus on getting the big data together to identify trends and put it into a visualisation or dashboard, but don’t create actions for it to implement change.
This is where you get to tell the story of the product and sell the vision of what it will be. Perception is important here as you only have the MVP and may not have the “fancy” or “sexy” elements to get the pilot over the line and into implementation. You may need to work with stakeholders to understand what shiny pieces of functionality they see as critical to proceeding.
PROGRESS: You now have a business intelligence system and a pilot that shows you data trends but, more importantly, how it changes your business’ processes and actions.
5. Implement your IoT technology
Now it’s time to solidify your hardware, add any extra features to your software MVP, and get it into production.
You can then measure production costs, how much hardware you’ll go through in a year, the maintenance costs, asset cost to manage and the transition cost. Once you’ve identified these and can justify it all, you're open to IoT.
When it’s time to implement, you need to identify the implications to the workforce and organisation to inform a change management strategy. Now you can manage the transition and integration across the organisation.
PROGRESS: IoT is now working in your business and making real business change.
BONUS: How (and when) to secure your IoT implementation
IoT security is really important because you have so many of these devices that need to be managed over the air, from anywhere. But knowing when to apply security is just as important.
If you were to implement security measures in the prototyping phase, you'd never get anywhere. You'd spend all your time on security and encryption when you’re not even collecting the right data or know if it's the right thing yet, so timing is crucial.
Think of security in terms of need. Maybe the data from one Redbull machine doesn't need to be protected very well because the data isn’t really “hacker level” useful information yet, so you could use entry-level security, similar to wi-fi encryption.
But the amount of Redbull in that store plus all the stores around it could be useful for a hacker, and that's when you would appropriately increase the level of security for that.
To do this, you need to go back through the 5-step process and work out where you should place things in your network based on the level of data.
Ask yourself how useful the data would be if someone got their hands on it. Usually the further up the chain you go, the higher the security level. At the point where this integrates into Redbull's order system, you're going to have very high-level security.
BONUS: Calculating the ROI
In this case, we’ve directly integrated the fridges into Redbull's reordering and dispatch system, and it has cut out the cost of the reorder agent who takes the phone call and the admin assistant who places the dispatch.
I’ve actually jumped all the steps of the service station employee doing the inventory and the stocktake, to him placing the order for Redbull, to Redbull taking the order, invoicing, etc – all those steps have been removed. Figure out those costs on an annual basis and watch the ROI increase in year-on-year forecasts.
Even better, think of all the things the reinvestment of those resources could do in your business.