Internet of Things (IoT) devices collect lots of data. This data can be quite useful and a substantial secondary value beyond the device’s primary intended use, showing things like equipment fatigue, employee performance and even product usage patterns.
But here’s the trick: The data coming from IoT devices only is useful if the business is taking advantage of the data. That’s where many businesses struggle.
“View data like you view money: it has value,” says Marcia Elaine Walker, principal industry consultant for manufacturing at enterprise analytics firm, SAS. “Are you letting it sit idle earning no interest? Or are you maximizing its value to get compounded returns?”
If your IoT data isn’t delivering the returns you want, here are six keys for getting more from it.
1. Pre-Process IoT Data
The volume of data generated by IoT devices can be staggering. There were roughly 8.4 billion connected devices in the wild last year, and Gartner research predicts that this total will balloon to 20.4 billion devices by 2020. Each of those devices generates data, often on a minute-by-minute basis.
One key for getting a grip on all that data and ultimately putting it to use is having the devices themselves pre-process the data and only transmit what a company needs.
“This helps to either pass only mean, median or some other aggregate value,” says Alex Bekker, head of the data analytics department at software development services firm, ScienceSoft, “or completely avoid passing data to the data storage and processing it at the system’s edge.”
2. Keep More Than Exceptions
Don’t throw away all the data from IoT devices, though. Routine data from IoT devices helps set a baseline that can help a firm uncover when there are exceptions that should be addressed, as well as malfunctioning IoT devices that are throwing off bad data.
“We often hear customers want to reduce bandwidth and storage by not transmitting data when everything is OK,” says Tolga Tarhan, chief technology officer for AWS cloud consulting firm, Onica. “I’d encourage customers to remember that having a sampling of ‘good’ data is critical to building machine learning models that can detect ‘not good.’”
Knowing the baseline helps businesses get more from the signals coming from their IoT devices.
3. Augment With Existing Data Projects
A third way to get more from IoT data is to use it with areas where a business already has data. Instead of needing a complete picture from IoT data, using the data to augment existing reporting and analysis is a quick and easy way to put IoT data to good use.
“To maximize IoT data, start with the data you have already,” says Walker. “It’s easy to become paralyzed if you don’t have a complete picture of every variable or you don’t have sensors everywhere. Analyzing existing data often points to areas where additional sensors or connectivity will provide the most return.”
4. Contextualize for Deeper Insight
Maximizing IoT data also can mean enriching it with non-IoT data such as unstructured data from social media or handwritten notes from maintenance records and call centers.
“Taking it all into consideration gives you a more comprehensive picture of the value you’re working to achieve,” notes Walker.
“IoT data is somewhat uninteresting until it is combined with context,” adds Tarhan. “If you create a connected industrial machine, the data coming from that machine is much more useful when you know the environment the machine is deployed into.”
This enhancement need not be done during data ingest, Tarhan notes. The correlation can be made later in post-processing.
5. Add Redundancy
IoT data is different from regular business data because it is sampled, and it originates from devices that have failures and experience noise. When IoT devices fail — which unfortunately happens periodically with any machine — the data gets messy or unusable. This can derail the utility of IoT data projects.
One way out of this challenge is through data redundancy built into the data model. This can be done with code or by including alternate sources of data that can serve as a backup and a check-sum in case of data failure. Businesses that want to get the most from their IoT data would be wise to factor in this redundancy.
“Clever algorithms or redundant sensors can help find a way out if a data flow interrupts for some reason,” says Bekker.
6. Build Use Cases with AI
Because not all IoT data is leveraged or even retained, figuring out the use case is a key component of maximizing the data utility.
Artificial intelligence and analytics platforms can help parse IoT data and assist businesses with selecting the right data and understanding where more sensors might deliver a noticeable analytical benefit. Let the machines help figure out how to maximize IoT data.