When corporations began large-scale adoption of newly available cell phone technology two decades ago, there were more than a few missteps. Security, roaming, overages, and maintenance were still unknown variables. IT teams would often receive a big surprise in the form of a large unaccounted-for data bill, making it difficult for IT teams to plan and budget for employees’ real-world needs. Over time, IT teams learned how to successfully support and manage cell phone programs, and today, it’s an ordinary competency largely devoid of budgeting surprises.
Now heavy industrial companies are facing a similar challenge as they increasingly implement Industrial Internet of Things (IIoT) technologies to drive efficiency and growth. Manufacturers commonly underestimate the long-term costs associated with connectivity, networking, and the maintenance needs of IIoT. As a result, they fail to consider solutions that will enable them to build products more cost-effectively and often overlook other products that have significant potential for new revenue streams.
Consider the case of a major transportation manufacturer that was looking to leverage maintenance and performance data from its vehicles. The manufacturer had just released a new fleet of vehicles equipped with IIoT technology and received a surprisingly large data bill from its cell carrier. It turned out that a connection server needed to transmit the data would occasionally go down, but the manufacturer had no way to temporarily pause data transfers when this occurred. As a result, the vehicle would continue to try to send data packages—often hundreds of times—unsuccessfully. When multiplied across an entire fleet, the unforeseen data overages the manufacturer faced could quickly total hundreds of thousands of dollars.
Today, manufacturers that want to build IoT products and related services that not only reduce costs but also generate revenue need to address total life cycle costs—the hidden factors influencing costs in the three to five years after a product’s launch. Take a look at the three main ways these costs occur and how you can address them in your own company:
Data Connectivity
IIoT products rely on connectivity but they aren’t always used in locations within the range of cell towers or Wi-Fi. As a result, connectivity costs can add up quickly. Data is the most uncertain component of IIoT programs because it’s difficult to predict how much data a product will use. Software defects, server issues, network latency, and infrastructure challenges can increase the quantity of data being transmitted—and the associated cost—tenfold. Until a product is tested in the field, manufacturers don’t have a good understanding of how much data will be needed, which makes pay-as-you-go contracts an unpredictable cost.
Manufacturers can overcome the ambiguity of data connectivity by negotiating fixed-fee cellular contracts. If the budget allows it, a flat-rate unlimited plan will set a predictable expense for your data costs and free your IT and engineering teams to focus on more important aspects of the project, such as user experience or new features. Keep in mind that the protocol chosen will have a big impact on cost. A full 4G LTE radio will incur higher operational costs than LTE Cat-M1, and NB-IoT spectrum is usually the lowest-cost option.
If the system requires strictly managed bandwidth costs (for example, an extremely large number of low-cost devices), some external controls and monitoring is needed. Failure modes, including exponential backoff algorithms, should be tested extensively, and cloud services should be monitored for anomalies.
Cloud Infrastructure
Surprise bills from cloud service providers are so common that an entire cottage industry of consultants and SaaS products has sprung up to provide cost analytics and optimization. Unit pricing in the cloud is extremely complex, especially with modern microservice and platform-as-a-service offerings. To effectively manage costs, manufacturers need experienced data engineers on their teams.
Devices should not transmit or store more data than is absolutely needed, and data should be regularly purged when it’s no longer necessary. Edge computing technologies can filter, aggregate and compress data before it ever reaches the cloud. Automatic scaling must be designed into the data processing architecture so that costs don’t outpace demand.
Ongoing Maintenance
While many manufacturers are familiar with the self-service maintenance model, in which users can complete some service themselves in the field, IIoT products typically cannot be serviced by the user because they’re just too complex. Simple activities like rebooting a device or updating software must be completed in person by technicians, which can cause costs to skyrocket 50 to 100 times if not managed effectively.
To avoid this, manufacturers should consider enabling automatic over-the-air firmware updates on wireless networks that can keep products up to date without the need for onsite technicians. IT business units must also reorient to Agile software development, an approach that drives rapid release cycles, to support their products. Agile training is critical and will represent a major cultural shift for the entire organization. The benefits are worth the investment; you’ll be able to ship faster, higher-quality products while gaining valuable feedback from customers.
It’s also important to develop IIoT products that deliver easy-to-use, delightful user experiences, just like the ones your customers experience on their favorite mobile apps they use every day. An experience that is designed to make the user more self-sufficient means lower costs for support and maintenance.
Though these problems may seem substantial, they’re quite surmountable. Meticulous planning at the earliest stages of product development can enable substantial cost savings within three to five years and position you to maximize new streams of revenue.
The best way to do this is to start small. Start with just 10% of your final vision and test it thoroughly to understand the data and the network issues you’ll face in the field before launching at scale. It’s a slower strategy, but for every month you rush, you’ll pay for it five times over the next five years. Don’t plan any big launches until you’ve taken one complete value stream (one customer, sensor, feature, data analytics report, and any other necessary elements) all the way through the product’s full life cycle. Only then should you add additional elements.
Starting small is the single biggest protection against uncontrolled life cycle costs. Working slowly and deliberately at the beginning of your IIoT project results in a smoother and more strategic IoT product development process. In the long run, that will enable your product to grow quickly, reduce costs and generate more revenue.
Dylan Tack is a principal engineer at Portland, Oregon-based digital transformation agency Metal Toad, where he develops machine learning, Internet of Things, security, and DevOps solutions for heavy industry, entertainment, and healthcare sectors. He holds a BS in electrical and computer engineering from the University of Iowa.
