ISPs Have Always Had Data Silo Issues and That Won't Change

Industry experts warned Tuesday that siloing of data remains a stumbling block as ISPs try to use it to increase efficiency. During day two of a Fierce Network virtual conference on automation, executives said companies are partitioning AI away from other units, but it won’t always be this way.

Silos “are a really important issue,” said Monica Paolini, Senza Fili principal. If we have silos that prevent using data across different domains, applications and business units, the value of the data “is just not there,” she said. “We might actually end up worse off.”

Robert Jakubek, UScellular senior vice president-information technology, said he has long been involved with questions about application silos. It feels like 25 years later this should be a done deal, he said. Yet silos remain “and it’s still a problem for us in industry.”

Lots of work has gone into eliminating silos, Jakubek said. “Everyone is looking for low friction ways of getting information,” he said. But the volume and the importance of data today versus the 1990s make the issues more difficult to work through, he said: “Now there are definitely business models that are built upon getting information out of that data.” Handling data is more complex than in the past, Jakubek said. That includes the complexity of the algorithms and how ISPs are using data, he said. There's also "complexity with privacy and security.” Silos will never completely disappear, he predicted. “The challenges of the future become greater.”

Ending silos is challenging in big corporations, admitted Chris Hristov, AT&T assistant vice president-network engineering and automation. Marketing, sales and engineering teams all want to look at data differently, he said. “Everybody wants to create their own separate data infrastructures, their own separate platforms, and they want to look at that data uniquely.”

Building separate platforms adds costs, Hristov said. “You end up funding platforms for one organization, platforms for another,” he said. AT&T is making progress, but in any large organization, silos will be “an ongoing problem,” Hristov said: “It’s just a function of large corporations in managing data.”

Silos mean that it’s difficult to integrate data within a company, said Rode Kirk, Microsoft global sales director-media & communications, and that data is difficult to transfer, he said. In addition, silos mean “inconsistent” data standards, he said: “This inconsistency can create problems for data analysts [and] decision-making, because that data may not be compatible with data in other data silos," he said. There’s also “reduced visibility across all of these different data silos.”

Kirk also stressed the “significant” cost of managing and maintaining data in silos. A company like AT&T has “vast amounts of data -- different data silos, different [business units].”

Kirk said ISPs are still figuring out how they will use increasingly sophisticated AI in areas like sales and marketing, operations and engineering, “but also externally in terms of what services they provide to their customers that are emboldened by AI.” Companies also want to use AI “in a way that has the proper guardrails in place.”

For now, top executives are focused on AI, but eventually it will become part of what the chief information officer oversees, Kirk said. AI is “so new that you need extra focus, extra attention.” The CEO and other top executives in the C-suite are asking what AI means for changing the company, its products and services, “and what new markets can we open up based on this AI capability?”

AI will become “part of our standard tool set that we’re all going to have to have, regardless of what role you’re in,” Jakubek said: “Much like today, you have to have computer skill sets, [eventually] you’re going to have AI skill sets.” Providers will want to integrate AI across the company rather than having one department that serves as “the AI czar,” he predicted.

Ten years ago, industry executives were having the same conversation about data scientists that they’re now having about AI, Hristov said. “I feel like this is a rerun of a movie.” Companies are separating AI from other parts of the business today partly to control costs, he said. “With AI, there are a lot of good ideas, but not all of them” add value.

Paolini said recruiting talented workers will remain an issue for carriers. She said she talks to a lot of college students “and they don’t seem to really feel like telecom is the greatest place to be,” she said. “We have an aging workforce and that aging workforce doesn’t necessarily have all the AI skills they need.”