Steve Stine's new job focuses on making sure business units have they data they need in the structure and place they need it.

November 2, 2017

8 Min Read
AT&T's Stine: Driving the Data That Fuels Business Success

Software may be eating the world, but data is in the driver's seat, and so it shouldn't be surprising that AT&T now has a chief data officer. What might be surprising, however, is Steve Stine's take on his new job.

Stine, who previously led an automation solutions team for John Donovan, now CEO of AT&T Communications, was named CDO by Melissa Arnoldi, herself recently named head of technology and operations for AT&T Inc. (NYSE: T). Despite his most recent focus, Stine tells Light Reading in an interview that automation is one goal of his data organization, and not its driving force.

"We would love to find, via the data, where are some other potential opportunities to make transformative improvements to not only the experience but the cost structure," Stine says. "But you know when you do it, data analytics, you might find that you don't need to automate, maybe you need to change a process, to change how you do a product etc., so the outcome isn't always automation. It's certainly one that we consider as we look to create efficiencies, but that's not necessarily the end game."

Instead, Stine says, the primary goal is to make sure the AT&T business units have the data they require, in the proper structure and at the right place, to enable analytics or other actions that let them run their businesses effectively. That is essentially what he terms "the data supply chain."

Figure 1: AT&T Chief Data Officer Steve Stine AT&T Chief Data Officer Steve Stine

"In some of this, we don't know when we're doing that, whether or not there will be a process improvement finding or an automation opportunity, so it's really making sure that the leaders and the analysts that reside within the business units have at their fingertips what they need," Stine says.

That is an ever more complex job. By the latest numbers, AT&T has 168 petabytes of traffic across its network daily, and that generates many different types of data in very large volumes: performance data, customer-specific data, operational data, network data and more. All of that is stored in various places -- data lakes, data centers and increasingly at the edge of the network -- in multiple different formats and structures. One of the basic decisions that has to be made is how specific sets of data need to be structured, handled and retained -- or not.

"If you were thinking about raw data being the least mature form of data and business-ready data sets being at the very top of that, then all data doesn't have to climb to the highest order of maturity to create value," Stine notes. "You've got to make that connection with the business on how much you need to do, because there's a cost associated with all that -- storage, tooling, scientists to work on it -- all those things. That's what we're still working through."

Stine's group builds on the work done by AT&T's Big Data organization to date, and works in close conjunction with AT&T Labs and Foundry organizations, he notes, as well as with the business units.

"I'm just fortunate enough to take what goodness was already being done within these various areas and bring those together to find a way to do more with it," Stine says. "They were doing analytics and doing data set creation based upon specific projects. We're trying to take a look at what is already out there, what's in our inventory, and then what's a systemic or programmatic way to go ahead and structure more data as we go through this. Then, we can start looking at opportunities to vet for insights between combining data sets, so you know the compounding effect that you wouldn't have normally got."

Next page: How data gets used

How data gets used
Perhaps the best way to look at the data organization's work is through the lens of specific use cases. While working on his hyper-automation assignment from Donovan, for example, Stine and his team explored how to optimize AT&T's truck dispatch operation. The company's 70,000 trucks traverse various geographies for diverse reasons -- installations and repairs being the main two -- in all kinds of weather and traffic conditions. Routing those vehicles more efficiently not only saves money by increasing productivity but directly impacts the customer experience as well. For example, instead of annoying new customers by asking them to set aside a day or half-day for a service turn-up, AT&T might be able to more precisely define when its crew would arrive, down to 15-minute segments.

"Just small increments of productivity means more jobs per individual per day and that improves the customer experience, but it also certainly helps the business in total from a resource and capacity planning standpoint," he said.

But the factors that must be taken into account for this to happen are many and changeable: weather, traffic, the specific facilities and task involved, the technician's training, the available truck and how it is equipped, as well as past performance. There is no way for a human dispatch capability to collect all of that data in real-time, repeatedly throughout a day, and keep up, producing the kind of productivity results AT&T wants, Stine says.

"We started to look at, how do you take an optimization engine, think machine learning, and start to bring all those things together, so all day long while you have a to-do list that's in queue to line up for dispatch, how are you always ensuring that you're updating that, so best tech, best customer experience, all of those things are accounted for in real-time?" he comments.

As AT&T is assembling this data, machine learning is also built in so that lessons learned in a given week's dispatches are applied to the operation immediately to impact the next week's processes.

The network operator is also working to aggregate all the data involved in network incidents - whether those are fiber cuts, equipment issues, malicious attacks causing congestion, or something else, Stine says. On AT&T's global network incidents occur regularly and potentially impact customers, but also internal elements such as radio towers, backhaul networks and video equipment, he adds. Information about any given incident will go into customer care centers for businesses or consumers, but also into other network operations centers. Customer calls may also be coming into the care centers.

"We said, 'Let's aggregate all this data'," Stine says. "In real time, this might take several hours for everybody to come together and coalesce around what it is specifically, thus who should we send to repair that cut fiber. We can now do this in nanoseconds."

That means in the instance of a fiber cut, for example, trucks can be rolled faster from the construction organization and the right truck can be sent the first time, while messages to all affected parties are also sent out immediately with accurate information regarding restoration times, he says. Customers can be proactively contacted and, when possible, services can be protected from impact.

Want to know more about fronthaul and backhaul options for wireless networks, especially 5G? Join us in New York City on Nov. 10 for Light Reading's 5G Transport and Networking Strategies There is still time to register.

AT&T was already doing a lot of this, using some automation, but without the data aggregation piece, he says. "There was no aggregation of the data with the intent of even, when I found out what it was, of saying every time I see a like signature to this, what should we do, and how should we react in order to really start to narrow impacts to customers or eliminate impact to customers if possible from these activities that go on," Stine comments.

The data organization also has the ability to continue to better inform how software-defined networking is used to improve network self-optimization that already happens today, to keep capacity available and maintain performance. "We know that every day we've got an opportunity to learn more," Stine says. With more data, comes more information and AT&T has a policy under its head of technology and operations Melissa Arnoldi of "relentlessly sharing what we're all doing to make sure that the broader benefit is received throughout the business."

That tight integration extends to working with the Chief Security Officer Bill O'Hern's organization to ensure data security as well as with AT&T's privacy organization, to ensure privacy and security of data, in motion and in storage, is built in from the outset and carefully maintained. Stine also says there is strategic importance in knowing when not to keep data -- especially if it isn't needed for compliance reasons -- since keeping data around incurs cost and also increases potential vulnerabilities.

— Carol Wilson, Editor-at-Large, Light Reading

Subscribe and receive the latest news from the industry.
Join 62,000+ members. Yes it's completely free.

You May Also Like