Piercing the Internet of Things hype: a reality check

By 2020, there will be – according to forecasts from Cisco – 50 billion connected devices. Gartner makes a more ‘modest’ estimation of 26 billion devices. And as for the value of the Internet of Things (IoT) market, by 2025 it is predicted to grow to USD6.1 trillion (McKinsey) or even USD7.1 trillion (IDC)!
These forecasts for the IoT market are almost as impressive as the huge data flows themselves, but are they realistic too?

Even the most conservative predictions assume a 30% annual growth rate, whereas the worldwide internet penetration shows a 3% growth rate. Based on elaborated data by ITU, World Bank and the United Nations, 46.1% of the world’s population have internet access nowadays. But merely counting the things that have the ability to connect, without taking into account the actual application, is like cracking a nut with a sledgehammer.

So is the IoT market overhyped?

We’ve passed the point of silly use cases – remember the toaster communicating with the fridge – but we still have a long way to go to reach the purposed objectives and we may wonder whether the expectations will be redeemed in the short term. Five years ago we saw similar figures, but they haven’t materialized to date. Today, we’re in a beginner’s phase of IoT with a big focus on the physical set-up: how can the connectivity be ensured? Which sensors should be used? What are the infrastructural needs? What do the devices themselves look like? Nevertheless, I’m convinced of the potential of IoT to change all industries. In the long term, IoT will lead to adjacent business models based on new technology.

Making sense of the data

While the various infrastructure providers are fighting for market share and the number of infrastructure standards is growing, some innovating companies are running pilots aimed at making sense of the gathered data. We’re on the eve of embedding an analytics layer in IoT applications. That’s what will bring about the real value. However, if you focus on the use cases where there’s already enough complex data available to enable value-adding analytics, they still only represent a very small share of the market. But I firmly believe that running analytical pilots with the collected data is a best practice and organizations that do so will gain a competitive advantage. Companies that are piloting with IoT data analysis are not only looking at the cost aspect of the necessary hardware, but are also seeking ways of boosting their profits by developing new services based on the technology. That’s a different approach.

We’re on the eve of embedding an analytics layer in IoT applications

The level of IoT adoption varies per industry. Heavily data-driven industries like Telco and insurance took their first steps in the IoT domain several years ago. Insurance companies are already tracking and tracing their most valuable investments, such as drilling rigs, because the cost-benefit analysis is easier for these use cases. As the price of the sensors continues to decline, we are seeing more and more new applications, such as the connected car. Insurance companies are experimenting with a behavior-based model, whereby drivers with a safer driving style pay less for their insurance. In this case, if you only want to analyse the driving style to calculate the insurance premium, you don’t need real-time analytics. But if you want to exploit extra services – e.g. to warn your customers when they park in a dangerous neighbourhood – then that’s when real time data analysis proves its worth.

Innovative business models

In the application domain we’re seeing various players, including new entrants, trying the first applications. In June of this year, the Municipality of Antwerp, Belgium, announced a concession agreement for the installation of smart water meters in the city. In the first phase, the water company Water-Link – togethe with the joint venture of electricity supplier ENGIE Fabricom and crane manufacturer Hydroko – will test 1,000 smart meters for a period of one year. No network technology suppliers are involved in this project, which illustrated that new players are eager to get a slice of the cake. The radio frequencies for IoT are licence-free which makes the investment limited. On the other hand, the opportunities are huge. By using sensors to record water consumption, Water-Link will be able to easily identify leaks, to detect fraud and to automatically bill its customers. The customers, in turn, no longer have to worry about the administrative aspects and they can deal with their own water usage rationally because they have access to real-time statistics.

The radio frequencies for IoT are licence-free which makes the investment limited

Another example is that of an electricity company that can predict whether a customer is at home, based on the electricity usage. I’ve heard that energy suppliers are in talks with courier companies about selling such data, which courier companies could then use to calculate optimal routes.

Who owns the data?

The above example shows that the battleground will mainly focus on the data. But who owns that data? That topic will be the hot potato for the next few years. Will it be the network technology providers, the car producers, the utilities organizations… or the customers? The ownership, control and leverage of the data will become central to how the IoT moves forward. I think that customers will play a major role in the IoT story. Once they realize what their data is worth, they will be more eager to share their data with those organizations that give them a better customer experience or overall value.

For example, if a telecom operator can determine that a customer has almost reached his data limit, the company can offer him a promotional offer in line with his behaviour. But IoT applications can also relate to preventing fraud in money transfers. In terms of data management, IoT also presents a challenge. Not all data sensors are 100% reliable, so data quality checks are important. For example, if there’s an alert about a temperature of 200°C in a car engine, does that indicate a technical problem or is the sensor defective? Furthermore, operational real-time data management is a whole new area. Traditionally, analytics are based on historical data to a great extent.