Although many organizations are evolving their use of sensors and remote controls to leverage basic analytics and data collection, they aren’t taking advantage of this information in the best possible way. The key is to gather actionable insights from a variety of devices and sources to make intelligent decisions. But many organizations don’t understand the various ways that devices digest and use IoT data.

The objective of IoT is to make the world interact with you while reducing your direct output of interaction with the world. In other words,  it’s supposed to make stuff easier. Tony Fadell, the former CEO of Nest, offered great insight in a Forbes article. “What we are interested in is the ‘thoughtful home,’ Yesterday I used a switch to operate devices in my home. Today I can use my smart phone. But does that really make my life any better? The home has to be thoughtful and understand your habits.”

He’s making a great point – stating that machine-to-machine interaction is cool, but it’s not what makes the future of “smart products” remarkable. Businesses investing in IoT development should consider how their products process and share the most valuable information, to deliver the most life-improving results.

To better understand IoT data, we need an approach to IoT analytics that mimics how we interact with the world today. Let’s think about IoT architecture as a human body in a three-tier context. The IoT ecosystem needs a brain to tell the body what to do, a spine to follow those commands, and neurons to send information from the brain to each part of the body. We’ll take this analogy in reverse order, starting with the first tier.

Tier One: Edge Analytics as Neurons

Edge analytics are based on the device itself. We can think of this as a single neuron that can filter information coming in, and send it up the chain for processing. This allows a sensor to determine what information needs to be sent upstream. Edge analytics are important because devices may need to make decisions, but aren’t always connected to the larger ecosystem. How then, can those devices make decisions based on real-time data? Devices that perform edge analytics, also called edge devices, can make important decisions quickly that only affect a small part of the IoT ecosystem. Small, inexpensive computing devices like Raspberry Pi are powerful enough to perform analysis directly on the device, enabling the device to take some level of action, even if it’s disconnected from the larger IoT ecosystem.

Tier Two: Streaming Analytics as the Spinal Cord

The second tier, streaming analytics, controls the reflex actions that don’t require deep computing to make decisions. This is the spinal cord of our IoT body. For humans, the spinal cord responds immediately to reflex signals, interrupting them in order to process information on behalf of the brain before sending a pain signal. That’s why we pull our hands away from a stove if we accidently touch it while it’s hot. Similarly, streaming analytics address tactical decisions by comparing analytical data with historical information in real time.

The advantage of streaming analytics is that the IoT ecosystem can make changes in real time rather than waiting for batch processing of large amounts of data. In addition, streaming analytics rides incoming data, discarding data that doesn’t need to be retained in long-term memory (data warehouse). There are some great cloud solutions out there such as Amazon Kineses or Azure Streaming Analytics along with open-source, cloud-based or on-premise applications such as Apache Spark and Apache Storm.

Tier Three: Cognitive Computing as the Brain

Other decisions require a deeper set of data and analysis with a wider view of the ecosystem. This is where the third tier comes in—cognitive computing. Let’s think of this as the brain. Cognitive computing can mimic human thinking, analysis, and strategy. It is performed in real time, by accessing long-term information and weighing it against incoming data. IoT vendors are turning towards cognitive computing because it can help identify patterns from large and diverse data sets and make human-like decisions that can’t be achieved using traditional business engines.

CognitiveScale and Intel are investigating the use of combine sensors, contextual data, and cognitive computing to help drive new strategies for healthcare, maintenance, traffic management, home automation and other industries. Intel is a particularly noteworthy example because its recent purchase of Saffron indicates that the company realizes that the real money is in intelligent interaction. Selling sensors just isn’t going to cut it anymore. Cognitive computing serves as a learning mechanism for the entire ecosystem by not only ingesting internal data but continuously feeding outside sources and enabling the full ecosystem to make the strongest decisions.

If your company is considering investing in IoT, look beyond simple analytics, data collection, and control. Look at how you can pull contextual information from mobile devices, direct data from sensors, and other data from third-party sources to make intelligent decisions on-device or deeper in your IoT ecosystem. In the end, if it’s not actionable, it’s just data.