When Alcoa decided to split its business into two separate companies in September 2015, the aluminum giant was not only met with shareholder flack, but a big data problem.
In creating the separate downstream company — North Shore-based Arconic specializing in high-margin aluminum and titanium products for the aerospace and automotive markets — Alcoa had more than 500,000 technical documents that needed to be separated.
Filtering through that volume of data for information could feel like the modern-day equivalent of employing the Dewey Decimal system in a library, rather than using Wikipedia, said Christopher R. Evans, vice president of Troy Hill-based software company Savvior, which employs just under 20 workers.
“The data is difficult to get to. There are only certain people in an organization who know how to access those systems,” he said. “It’s not easy to know the right questions to ask.”
But that’s exactly what information technology companies are trying to build — smart systems that help employees not only find information, but make sense of the records. Largely, it’s a battle against knowledge loss, or a lapse in critical insight that companies face typically as a result of employee turnover from either retirement or restructuring.
To supply Arconic with relevant systems, intellectual property and databases, Alcoa used Savvior’s OmniView product, an information management system that creates a custom central repository for records.
Alcoa and Arconic did not respond to interview requests for this story.
While OmniView, created last August, organizes explicit knowledge — important records and large data sets — it also works to locate and reuse the more interpersonal, tacit knowledge created by a worker over time — know-how that is difficult to transfer, like the ability to quickly pick up the phone and call the right expert on the correct topic.
That’s tricky for information management systems to replicate, especially with a widening gap between workers near retirement and those fresh out of college.
In Pittsburgh, a bifurcated workforce persists — divided between a young generation of tech employees and baby boomers nearing the end of their careers, said Deborah Good, clinical assistant professor of business administration at the University of Pittsburgh’s Katz Graduate School of Business.
“There’s not many [workers] in the middle. That’s where the historical knowledge needs to be conveyed,” said Ms. Good. “The bigger the company is, the more opportunity there is for knowledge loss.”
In Pittsburgh, 22 percent of workers are over age 55, compared to 19 percent nationally, according to a 2016 report by Burning Glass Technologies and The Council for Adult and Experiential Learning with the Allegheny Conference on Community Development.
When seasoned employees leave, they take tacit knowledge with them, explained Linda Argote, professor of organizational behavior and theory at Carnegie Mellon University’s Tepper School of Business.
“Repositories are best for explicit knowledge, but they can provide a pointer to people with tacit knowledge,” she said. “Even if the info isn’t in the repository, you can find the person to reach out to.”
The OmniView platform harnesses machine learning and artificial intelligence.
“You need an expert to find a specific document,” said Mr. Evans. “Machine learning, here combined with search and reporting, gives a high level of accuracy ... it can show subject matter experts by percentage of knowledge.”
In practice, that could mean a traditional repository might pull up two material science experts who both are listed as knowledgeable on semiconductors. One might be a researcher who has written expansively on the topic, while the other is a chemist. Based on your behavior over time, OmniView would determine which expert would better suit your needs.
“When machines are operating, they’re collecting detail transactions,” said Kathleen Hartzel, chair of accounting, information systems management and supply chain management at Duquesne University.
“If you look at Alcoa or Arconic now, their data would be their sales and transactions on the shop floor ... it would be how much product they’re producing, or the temperature at which they produce the product,” she said. “Metadata would tell us what those individual transactions are about and what the data as a whole communicates.”
Ms. Hartzel, who holds a Ph.D. from the University of Pittsburgh, said traditional information systems still can be problematic because workers don’t always have an incentive to enter data.
“From a management perspective, if I have expertise, do I want to take the time to enter my expertise, which in turn, diminishes it?” she said. “There has to be some kind of management strategy that people know about that encourages people to take part in this sharing behavior.”
Automation of any sort can just make it easier.
“If you can automate the capture of data, that’s a big deal,” Ms. Hartzel said. “You’re looking at tech that takes data, looks for patterns and makes projections on what you’d anticipate a process to do.”
While at least three global companies have purchased their own OmniView product, not all businesses have the same needs for data.
It depends on the industry, said Param Vir Singh, associate professor of business technologies at CMU’s Tepper School of Business. “If a company is huge, they like to build it themselves and tech companies build it on their own,” he said, noting businesses that produce software, like IBM, have the know-how to create their own systems.
Costs are also a consideration. The price for an OmniView system is contingent on the time and effort involved in integrating the systems, customization and other factors that may be nuanced around user experience, Mr. Evans said. Most clients host the platform on their own servers due to privacy concerns.
Mr. Evans recognizes that knowledge loss doesn’t pervade small companies in the same way it does major businesses, and those companies may not need a product like OmniView.
“Here’s where it’s not a problem ... when you can walk down the hallway and get your answer in 5 minutes,” Mr. Evans said with a laugh. “A 10-person company doesn’t need this, but a 10,000 person company does.”
Courtney Linder: clinder@post-gazette.com or 412-263-1707. Twitter: @LinderPG.
First Published: August 11, 2017, 1:30 p.m.
