Five reasons Internet of Things development will take longer than expected

New McKinsey report pours a dose of cold reality on the Internet of Things vision, noting that an intense re-alignment of IT systems and people will be needed.
Written by Joe McKendrick, Contributing Writer on
Like herding cats, fully embracing the so-called Internet of Things requires being able to pull together a host of variables and keeping everything in sync. The right technologies need to be in place that can gracefully handle the flow of data in various formats. It's much better, of course, if there's some standardization. That's the easy part, of course -- suppliers, partners, internal users and customers also have to be on board.
Photo: Joe McKendrick

In short, if your organization is committed to achieving the possibilities always-on, real-time, sensor-driven applications can provide, then there will be no shortage of work at your company for some time to come.

That's the word that comes from reading between the lines of a recent report from the McKinsey Global Institute. The report, authored by a team of McKinsey consultants led by James Manyika, says embarking on the path to IoT value requires progress in three areas: improvements in basic infrastructure elements (lower-cost, more capable hardware components and ubiquitous connectivity), improvements in software and data analytics, and the development of technical standards and the technological solutions for interoperability.

Above all, interoperability is key. McKinsey prognosis: not there yet. Some of this will be out of enterprises' control, dependent on cooperation between vendors and the work of standards bodies. What will be within enterprises' control will be laying the groundwork internally, as well as with business partners. Participants in an enterprise IoT formation will need to work more closely than ever, and adopt the same or compatible solutions, to enable data to flow freely between sites and applications.

Low-cost, low-power hardware is required. McKinsey prognosis: some progress made. "One of the basic requirements of IoT is to have the capacity for millions of devices, machines, and computers to talk each other, sometimes across large distances," Manyika and his co-authors state. The costs of an IoT project could spiral into infinity without access to cheap components. "Today many applications are technically solvable, but the high cost of components such as sensor nodes -- with communications and power supplies -- makes implementation impractical." IoT efforts will be helped in this regard by ever-cheaper microelectronics, they add. And while RFID tags are relatively cheap -- about 15 cents each -- they need to get cheaper "to make them practical for tracking low-value inventory in retail, manufacturing, and shipping." Plunging cloud computing and storage pricing also helps -- "the price of storing a gigabyte of data on a public cloud service fell from 25 cents in 2010 to .024cents by late 2014."

Ubiquitous connectivity will make the world go 'round. McKinsey prognosis: not there yet. "Many applications that require more complex analytic computing of data from diverse sources will need ubiquitous connectivity, which is not yet available, particularly in developing economies," the report states. "Even in advanced economies, wireless data service can be patchy and unreliable outside urban centers--where many factories, warehouses, and other industrial buildings are located. In the United States, some farmers load sensor data onto USB drives because they cannot count on wireless data networks. In developing economies, coverage."

Analytics software -- and analytic people -- are needed to help have it all make sense. McKinsey prognosis: we're working on it. Analytics is where IoT goes from being just a bunch of data to information of value. However, skills in this area are in short supply, and today's generation of analytics software "has not progressed to the point where it can be easily applied in every case--one reason that so much of the data that is collected goes unused," Manyika and his co-authors assert. "The hard work of developing and tuning these algorithms for the peculiarities of specific use cases is largely still undone,and the skills and capabilities to do this work remain in short supply.There are also gaps in capabilities in user organizations that prevent full implementation of big data analytics in IoT applications." Add to this the fact that most corporate data still resides in silos, a problem IT teams have been trying to crack for well over a decade now.

More attention needs to be paid to privacy, confidentiality, and security. McKinsey prognosis: this will be a lawyer's banquet. There are compliance issues that come into play across IoT implementations -- for example, healthcare data streaming in from devices have HIPAA implications. "Some IoT consumer applications, particularly those related to health and wellness, collect sensitive personal data that consumers might not wish to share," Manyika and his team point out. Even corporate operational data may be at risk if portions are streaming across networks. In addition, simply having thousands of devices and sensors in the wild magnifies potential for hacking and data breaches. "Each device increases the "surface area" available for breaches, and interoperability expands the potential scope of breaches. Every node is a potential entry point, and interconnection can spread the damage." Another question that has yet to be answered: who actually owns the data? This must be clearly spelled out in contracts, or else the lawyers are going to have a field day.

Editorial standards