Needed: A Real-World Analytics Plug-In For Digital Operations

For Amazon, Uber, and just about everyone in between, the digital transformation of business is presenting opportunities for success that, for many, are both exciting and bewildering. The theory is straightforward. Business now has access to detailed data that wasn’t possible a few years ago. And we have networking and cloud-analytics technologies that can mine that data without requiring massive Capex investments.

And we’re only at the start. Machine learning and other analytics algorithms are improving with each day, as are the insights gained from the analytics’ access to ever-expanding historical data.

The opportunity? A future of near-perfect continuous improvement for the core operations of business, one that maximizes performance, efficiency and value for the business, its stakeholders and customers.

It seems like a big leap forward, and it is.

According to PWC’s Global Digital Operations Study 2018 – Digital Champions, it means pulling together detailed information from manufacturing, procurement, product management, R&D and all other sectors of the business; from ERP, CRM, MES, security and other technologies; and analyzing them along with data coming in from the outside, from customers, IoT sensors, mobile devices and other sources.

There’s no shortage of technology solutions available today that promise to do the job, or part of the job. And there’s no shortage of executives who are investing in these technologies. But there’s a danger that some investments may be badly placed. As ZDNet points out, “Digital transformation sounds great until the core business slows down and shareholders want better returns.”

What’s needed is a real-world, concrete plan for getting there from “here,” depending on where “here” is for any given company. Also needed: a way to see positive results in a relatively short time, to help in getting and keeping employees and ecosystem partners on board.

Step One: Operations Visibility
Bringing real-time visibility to operations is an essential first step toward transformation because it can create a baseline for linking operations health to business performance indicators such as customer experience and satisfaction. An analytics engine takes in data from sensors, log files and other operational sources, then correlates it with higher-level performance metrics, giving operations center engineers, IT managers, executives and others dashboard views with drilldown capabilities. It’s a visual representation of the business in real time, and a foundational point-of-departure for subsequent analytics tools and techniques.

Step Two: Anomaly Detection
Also for operations center and IT professionals, anomaly detection builds on real-time visibility by responding to the growing numbers of new operations assets and technologies that provide health and performance information. Anomaly detection uses AI and machine-learning algorithms to find patterns in nuanced and complex data sets and cancel out false positives or meaningless alarms. For instance, anomaly detection analytics can help the operations manager group seemingly unconnected anomalies to a specific set of network elements, or to a cluster of machines on a shop floor.

Step Three: Change Management
The next level of analytics sophistication, change management, is able to create and analyze dynamic populations of devices, sensors or other assets. it is invaluable to finance executives, business analysts or others who want to mediate the effects of large-scale changes. An example: a mobile-phone provider becomes aware of a scheduled operating system revision to one of the cell phone models used by customers. Change management analytics quickly identify the customers who will be affected – the dynamic population – so the provider can contact them proactively to push any needed corrective updates.

Step Four: Incident Lifecycle Management
As the digital transformation grows it will generate vast quantities of new data and create more complex data interactions. The analytics framework will respond by drawing on its historical knowledge of past events, and using machine-learning models to track recurring incidents over time. It then uses that information to triage the events that matter most, using predicted business impacts to find patterns among the ever-more-complex interactions of new data.

Step Five: Dynamic Failure Prediction
This follows onto implementation of the previous four steps, and gives users an analytical tool that can collect and interpret pre-incident signals, extract patterns and trigger proactive identification and resolution. In addition to helping operations and IT managers, it is especially valuable to IoT devise manufacturers who may then provide add-on monitoring services for their customers. For example, it would be used to monitor robots for predictable failure signatures, then, upon finding a signature pattern, schedule proactive maintenance to avoid costly downtime. Similarly, it can be applied to operational processes, monitoring, predicting and then resolving problems before they create workflow stoppages.

Growing Sophistication
Importantly, these steps represent a growth in sophistication and utility, with each building on the abilities of the previous one. Operations transformation will demand such an evolution, as organizations learn about their own specific needs and operational abilities.

It’s also important to put this plan into action sooner rather than later to build confidence and visibility within the organization and among investors and other stakeholders. The best way to do this is to look for an analytics framework with a library of low-code/no-code application components that can be assembled within a visual development environment. This facilitates fast iterations and helps you build solutions that can be used and operated without requiring expert knowledge, a resource that will be in short supply, thanks to the advent of digital operations.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>