Before using data visualization in machine learning, General Electric Power would always manage its financial workflows in a manual, time-consuming, labor-intensive manner. Business process analysts would hold a meeting and ask employees to describe their work.
“But there’s inherently a problem with that,” said Jeff Cowan, GE Power’s senior IT director of continuous improvement and optimization. “Humans are conditioned to think of the most important things that they do. They don’t necessarily think of all the activities that they do automatically. They tend to leave a lot of stuff out.”
As a result, the analysts would have to watch employees process payment orders or send invoices to suppliers. “And 10 times out of 10, what’s documented and what’s observed are very, very different things,” Cowan said. Mapping that process would take about one week, and the analysts would then look for areas of improvement. But that involved a lot of guesswork.
“Most likely, you already knew what you were going to focus on anyway,” Cowan noted. So, in reality, the process was more about the certification than about actually improving the process, he added.
