Fast-forward 27 years, and I get asked a lot about the most recent form of the internet – the internet of things (IoT). And while I think the current possibilities of IoT are fascinating, sometimes I wonder if we are today with IoT where we were with the internet in 1990: We will look back and say, “We thought we had the internet back then. Look at it now.”
I spoke more in-depth about IoT recently with Intel's General Manager of Internet of Things Solutions, Kumar Balasubramanian; and Timothy Chou, author of Precision: Principles, Practices and Solutions for the Internet of Things. You can watch the full webcast, How and Why the Cloud Completes the IoT Opportunity, on demand.
Today, I would like to share the top three best practices that companies can apply to prepare for the future of IoT.
1. Build a data-driven culture.
If you work on IoT applications, you have to be ready for data – lots of data. You must create a data-driven analytics culture, because IoT generates data and depends on analyzing and acting on it. That is a cultural shift for many organizations. If your company orients itself toward service, you need to build domain knowledge. And the best way to do that is through data.
Take the VR Group, for example. The Finnish railway used sensors to continuously monitor the condition of its trains. That data allowed it to shift from scheduled maintenance checks to predictive maintenance. It was able to gather the data and know when a part would need to be repaired in real time. But in order to gain insight like that, you need to develop analytic models for early failure prediction and for prioritizing failures, and you need to manage the life cycle of these models.
Just as it did for VR Group, analytic insight creates opportunities for greater productivity and efficiency, and helps you understand your business in a new way.
2. Analyze the data.
We live in an analytics economy, and it is fueled by data. Wherever there is data, there needs to be analytics. Data without analytics is value not yet realized.
The VR Group could have tracked data about its trains. But if it did not analyze the data to understand which parts would break, that data would be worthless.
If you go from regular maintenance to predictive maintenance, you need to be able to observe, model, predict and decide. That requires a sophisticated model and advanced analytics. You need exploration, visual and predictive analytics, and machine learning to make sense of the data and build an analytics infrastructure.
3. Act on the data – quickly.
A common model in IoT is to observe data in machines, devices, appliances, etc., transfer it to the cloud or the data center, and act on it there, often through batch processing. While this works for some applications, I believe that it is generally a broken model considering the rapidly increasing number of connected devices and the need to act on data at the right time – which often means in real time.
You need to be able to handle data with the speed at which it comes off the sensors. Data has latency and expiration dates. If applying analytics to data creates value, acting on it too late can lose us that value. We need to act on data when we see it, when we touch it and when we aggregate it. We need to understand the right time to perform the right type of analytics.
Think about the finance industry. If a credit card transaction is fraudulent, you would want to know as soon as the card is used. Even if you decide not to stop a potentially fraudulent transaction at the point of sale, you need to know right away that such a transaction occurred. That is the type of reactivity that analytics can bring to every industry.
We have taken huge strides in the last 27 years. I’m fascinated by the opportunities that the internet of things brings. With cognitive computing, machine learning and artificial intelligence, a new world of possibilities lies in front of IoT. The journey has only just begun.
I am really looking forward to when IoT changes from the “internet” of things into the “intelligence” of things. Given the pace at which greater smartness is embedded in software through analytics, who knows how close we are?