Think big but start small: Tapping into rich analytics with IoT

from Think big but start small: Tapping into rich analytics with IoT
by Jerry Lee

If you’re a business owner or enterprise beginning to leverage the Internet of Things (IoT), chances are you’ve started connecting your operational technologies – including machinery, building HVAC, and other assets – to your current software systems. As a result, you’re likely collecting a ton of new data and think see the potential transformational value in there…somewhere. One of the most difficult questions to answer when starting out with IoT is how to take vast amounts of raw information and create real business intelligence from it.

The combination of IoT and data and analytics may seem like a “chicken vs. egg” dilemma, but it doesn’t have to be. Companies who have adopted IoT with the ultimate goal of optimizing physical processes or providing predictive analytics solutions still have an opportunity to use data and analytics to advance their business, even if an implementation has stalled after the technology is in place. It’s a more common issue than you might think, as businesses new to IoT often lack the necessary expertise to move to the final step of figuring out how to work with the data they’re collecting from their Internet of Things (IoT) initiatives.

For this reason, much of the data being collected by new IoT deployments today is ignored or thrown away. In some industries like fleet management, the useful life of the data may be too short to justify keeping it. In others, like broadcasting, files may be too large to store for long periods.

The truth is, we’d rather see our customers keeping and using their data. Existing and future data can be used for trend analysis and benchmarking present performance against the past. Microsoft research has found that about 23 percent of enterprises use data for basic reporting functions – and that’s a good start. Keeping data on hand leaves room more sophisticated analytics later, as business needs and capabilities grow.

For businesses looking to go beyond simply collecting and storing data, a recent study by Keystone Strategy* shows that there are three typical stages of analytics capabilities companies employ to analyze their IoT data (assuming they are analyzing their data at all):

1. Reactive analytics, which involves companies analyzing data in a batch processing approach, such at end of day or end of month, so they can benchmark and assess trends. This type of data analysis is mostly looking backwards to identify trends (asset utilization, uptime, etc.).
2. Proactive analytics are often initiated by companies that realize the importance of their operational data, and want to start processing their IoT data in real time. Since the value of operational data decreases over time, the more quickly a company can report, the better. With real-time processing, companies can also drive real-time alerts and actions, so they can leverage learning to initiate new business processes (for example, a machine failure triggering a service ticket request). Basic reporting such as machine utilization, failure reports and real-time dashboards can provide immediate value to IoT adopters, and really help make that next step towards servicing customers.
3. Predictive analytics enables companies to calculate future probabilities and trends, diagnose possible outcomes and make recommendations. This often requires analysis of big data sets that contain both structured and unstructured data to uncover hidden patterns, market trends and other useful business information. Only about five percent of the enterprises surveyed are at this level, but this is where real business transformation begins to occur, unlocking new business models and revenue streams as manufacturers, for example, are able to become predictive maintenance providers as well.

Typically, the progression of data analysis above mirrors the same evolution we see from our IoT customers, from reactive, to proactive, and ultimately to predictive analytics using machine learning.

With a vision in place for data usage, the steps needed to get to rich analysis will become much more clear. At this point, it’s all about execution, which involves two more fundamental elements: the right technologies, and the right people.

Having a cohesive, end-to-end technology platform to build your IoT solution can remove a huge amount of churn. Microsoft’s solution to this necessary element is the recently launched Azure IoT Suite. This offering is comprised of preconfigured solutions, which enable you to quickly and confidently get started with a solution that is tailored to meet your business needs, then scale across your entire company.

But even with the best technologies in the world, in the end, like anything in business, it’s all about people, and few companies today begin with the right skillsets internally to do the kinds of analysis they envision. Today companies that lack the ideal skills in-house can partner with burgeoning array of consultants, vendors and partners with deep experience in their industry, and the data science capabilities to put all the pieces together. You can read much more about how to get started with IoT here.

Learn more about how to create business intelligence with your data using IoT at http://ift.tt/1qXMgCt.

*Source: Keystone Strategy, “IoT Influence Model Research,” March, 2015.