By Kim S. Nash for the Wall Street Journal.
Caterpillar Inc. is nurturing a growing line of subscription data services that analyze information gathered from sensors attached to 500,000 of its machines in use worldwide. The idea is to offer customers tailored advice about how to run building and mining projects more profitably.
Selling data services to customers isn’t new for the 92-year-old manufacturer of tractors and bulldozers, which in the 1990s began a telematics program. While rising commodity prices and mining activity led Caterpillar in April to report its first quarterly revenue gain since 2015, the company is working to find new sources of income as other markets remain depressed.
“The data is as important as the machine providing it,” Tom Bucklar, director of internet of things and channel solutions, told CIO Journal.
Caterpillar rents and sells equipment and all new machines are connected, letting customers choose data services to subscribe to. The company also is working to retrofit older machinery with sensors and analytics technology.
Now, more extensive services based on machine learning tools, launched under the Cat Connect name in 2014, are becoming a bigger business for the company, Mr. Bucklar said.
Mr. Bucklar’s team is focused on providing insights tailored to the financial, operations or safety goals of customers. One mining company might want to avoid downtime of essential machines like wheel loaders, which put rock and dirt into trucks that haul it away. A construction firm might want to work more efficiently during daylight hours so crews don’t have to work in dangerous dusk hours with lower visibility.
To create specialized services, Caterpillar must gather and analyze many more data points than its original telematics system, and plug them into in its proprietary algorithms. Blending machine performance data with, say, information about the dirt, water and metal content in a machine’s fluid system, plus feedback from visual inspections, produces a fuller picture of the equipment’s health and how and when it can be operated, Mr. Bucklar said.
That analysis can be valuable to customers, he said. For example, if levels of dirt and metal rise, filters can clog and cause temperature spikes that can lead to outages, he said. Proprietary algorithms let Caterpillar predict performance and recommend interventions.
To help Caterpillar’s sales staff, Mr. Bucklar’s team can run analyses to show customers how they could use the technology. A mining customer recently spent $650,000 to repair and rebuild a key machine that failed unexpectedly, resulting in 900 hours of downtime. Caterpillar plugged relevant variables, such as usage history and performance metrics, into its proprietary algorithms to demonstrate how the data could have warned the miner before problems occurred. The situation could have been a $12,000 repair with 24 hours of downtime, he said.
Determining how much to charge for such insight is difficult and often depends on negotiating with the customer, Mr. Bucklar said. Unlike a single tractor with a set price, the value of knowledge differs for different customers and at different points in time, he said.
Caterpillar data scientists and external partners build proprietary models to produce what can be “high-value” predictions for customers, he said. Yet there is no standard value for the prediction of machine downtime, for example.
“If our largest, most complex machines tend to be the most important machine for a customer, they likely can’t afford downtime on it,” he said. “But if a smaller, simpler machine broke down, they could easily rent another one. There’s not as much customer value in that prediction,” he said.
Setting prices for data services involves talking in depth with customers about their own financial, operations and safety goals, he said. Some reports are included in the cost of a machine, while others are bought by subscription. Updating software monthly with new features developed from Caterpillar’s machine learning systems might command another price, he said. “It’s an ongoing process.”
Via the Wall Street Journal.