Champions
of Data + AI

データドリブンな革新を推進するリーダー

EPISODE 23

How Data and AI Are Revolutionizing Telecoms

In this episode, Andy Markus, the CDO at AT&T, dives into how AT&T leverages analytics and AI to improve the customer experience. He shares his perspectives on change management, business value realization and talent development. Andy also sheds light on AT&T’s AI use cases and governance frameworks as well as their experience using Databricks to drive innovation in a regulated industry.

headshot
Andy Markus
Chief Data Officer, AT&T
As SVP and Chief Data Officer, AT&T Services, Inc., Andy Markus leads the strategic vision for data platforms and advanced analytics that empower AT&T to double down on data for business decisioning and to unleash the power of AT&T’s unique and differentiating data to better serve customers.

Markus is an accomplished data analytics professional with over 28 years of experience in providing creative and privacy-compliant solutions to real-world data issues. His expertise involves creating and cultivating large consumer data platforms; direct-to-consumer marketing and advertising audience targeting; predictive analytics; building, motivating and managing staff; and developing strategic plans to accomplish key organizational goals. Previously, Markus was SVP, Data Management and Consumer Analytics at WarnerMedia. He led the Enterprise Data Solution efforts involved in executing the data supply chain across all parts of legacy Turner, HBO and Warner Bros. He oversaw data architecture, development, audience activation, and consumer identity management and was instrumental in enterprise data governance and privacy activities.

Before WarnerMedia, Markus was vice president, Database Marketing Services at Time Inc. Prior to Time Inc., Markus was assistant vice president with Management Decision Systems.

Markus graduated from The University of Alabama with Master’s and Bachelor’s degrees in Statistics with a minor in Computer Science.

Read Interview

Chris D’Agostino:
Andy, thanks for being part of the Champions of Data and AI series. It’s great to see you again.

Andy Markus:
Hey, Chris. It’s really great to see you. Thanks for having me.

Chris D’Agostino:
Let’s jump right in. So why don’t, if you can give the audience kind of a quick introduction to yourself, you know, your role as the chief data officer at AT&T. And without disclosing any proprietary information like, you know, tell us a bit about some of the things that you’re working on that are helping AT&T drive forward with customer acquisition, you know, customer 360, that type of thing.

Andy Markus:
Yeah, absolutely. So I’m the chief data officer for AT&T. You know, we have so much data at our fingertips at AT&T, we’re really transforming the company to be data first by first. You know, I think the movement across AT&T to really embrace that vision is super encouraging. And we’re doing so many wonderful things for the company and for our customers, you know, using data and AI. I mean, one of the coolest things that we’re working on right now is an effort to really suppress robo calling. So, hey, you’re a mobility consumer, whether you’re an AT&T consumer or not. It’s a you know, it’s a hassle, right? You know, we all get burdened down with that nuisance that we have from robo calling. And so we’ve taken an approach to try to tackle that with data. And, you know, we’ve used a lot of, you know, standard techniques with AI and ML and then added, you know, graph based solutions to that to suppress, you know, that nuisance robo calling activity. And today, so far this year, we’ve suppressed over 7 billion robocalls and we just released new functionality, I think it was two, three weeks ago where we’re suppressing an additional 40% more. So, 5 million robocalls additionally suppressed per day. So it’s really awesome to see, you know, great technical work, you know, in data and AI really drive value to our customers and to our consumers. And hopefully if you’re an AT&T consumer, you’ve noticed a drop in your robocalls, too.

Chris D’Agostino:
That’s really cool. I mean, most of our viewers are always keenly interested when we talk to data leaders such as yourself, you know, you are a very senior role at a major, major corporation. People are very interested in your path and how you got there. And so the question is, you know, is this something that was by design? Like, have you been pretty methodical in your career selection and what roles you’ve taken within organizations? Or is this something some of our leaders, you know, have always had a love for data, and then the chief data officer role is kind of emerged in the last few years and sort of frankly, sort of just fell in their lap and they were the right person at the right time. How did it play out for you?

Andy Markus:
So, Chris, I mean, it’s been a gradual, you know, evolvement for me. So, you know, I have an undergraduate and graduate degree in statistics from the University of Alabama, so “Roll Tide” to anybody that has an affiliation with Alabama. But, you know, out of school, you know, started, you know, building models, first of all, to predict, you know, risk associated with, you know, with lending, right? So built some of the first commercially available credit scoring models and they were cutting edge at the time, but they’re commonplace today. And, you know, from there, I moved into modeling on the publishing side of the world, you know, subscriber acquisition, you know, subscriber churn and retention management. And at that time, I kind of took over responsibility in this publishing world for, you know, the analytical database. And so that moved me from not just the analytical side of the house, but also to the data side. And so I’ve kind of had both feet in both worlds ever since. That was about 30 years ago. Now, a lot’s changed in those 30 years. I mean, that first analytical database I worked on was written in assembler. I can read and write a little assembler. It’s kind of a true dark art, you know. But it’s all changed, right, but the same focus on customer, on driving value for the business, is still there, right? So, you know, statistics is kind of converged into data science. You know, we’re no longer running on prem systems where we’re running our data environment in the cloud. We’re using great, you know, technology like Databricks. But you know that that core focus and the focus that I’ve had over time in my career is to use these technologies to really, you know, serve the business better and to serve our customers better.

Chris D’Agostino:
Yet, when I talk to people in your role and the chief data officer role, I think it is fascinating to me and I appreciate the balance between the business acumen that a leader has and understanding the business and what the key business stakeholders are looking for from data leaders. The technical acumen is really kind of my passion, right? And so you talk about, you know, assembler and like, you know, having been very deep in the tech and then, of course, the interpersonal skills and the leadership skills. For the tech skills, tell me a little bit about how you feel that’s helped you in your career having such a strong technical background?

Andy Markus:
You know, I think it’s foundational, and I’m not going to say that some people can’t do it without it, but for me, it’s been essential to really understand a few things, to understand what I’m asking people to do, both, you know, how much time I really think it’s going to take. You know how you know what I’m really asking to divide off in an exercise to perform a function. I think since I’ve been there and done that and I’ve lived through that process, I can guide the team better. And on the other side, I think it’s really helped me on the vision of what our target architecture needs to be. You know, all throughout my career, right? I mean, because I’ve been there and done that, you know, I know the trade offs of architecture, of the technical side of it. And so, again, I might not be the expert on it, but I understand, you know, what, we’re giving up, but we’re getting in the balance to make those the right decisions for the architecture of our future. And I think we’re in a great spot. And, you know, working with you guys, working with Databricks is a core part of that.

Chris D’Agostino:
That’s great. Yeah. I mean, one of the things that I noticed in my career over the last seven years, in particular with the move to the cloud, is our business stakeholders that I interacted with in my previous role before Databricks they were so much more technically savvy than business stakeholders that I had worked with, say, 15, 20 years ago. And now it’s a lot more collaborative, and your stakeholders understand these technologies a lot better than they did say, you know, 20 years ago. And so to me, I think the challenge for data leaders and technical data leaders in particular is to always continuously learn and, you know, frankly, to try and stay ahead of the curve because, you know, you want to be hopefully the trusted advisor in the room to your business partners.

Andy Markus:
Yeah, I totally agree. And I think not only on the technology side, you know, are the business leaders, you know, becoming more fluent, I think on the analytical side, too, I think people are understanding maybe not the theory of what we’re doing, but for sure , the application. And that’s super helpful, too. So you’re not starting from scratch. I’ll tell you a very quick story. I was in a meeting this week with John Stankey, who’s the CEO of AT&T, and we’re talking about a project we’re working on. And he made a little joke to the side about our square. I thought, oh, that’s awesome. You know, the CEO just, you know, made it kind of an inside joke about, you know, how strongly our square was on this project. And I loved it. I was so impressed with that. I’m not sure everybody else in the room got it, but I did.

Chris D’Agostino:
Yeah, that’s great. That’s great. Well, let’s let’s shift gears a little bit and talk about some of the unique challenges that AT&T faces as a regulated industry. Right. Help us, the audience understand, like, how do you view that balance between ensuring that compliance, but then making sure that your data teams have access to the broader set of data to improve the user experience?

Andy Markus:
Our general guideline is that, you know, if we use data in a transparent fashion and we’re using data in AI to really deliver value to our customers, you know, we’re on the right track and we work with our privacy teams, our legal teams to just ensure that we’re staying within the guardrails and they’re great partners. But really, you know, if we keep that as our core tenant, our guiding principle, you know, we’re often mostly, you know, straight in the fairway and where we need to be. And so, you know, we’re doing that. And all that said, you know, you’re right. We’re striving to be innovative. You know, we’re trying to reinvent who AT&T is so that we embed, you know, data and AI at the fabric of what we do. And we’re evangelizing this to all of our business leaders. And they’re so receptive. I mean, I think we’re really at a pivotal point in AT&T where, you know, we do great things with data and AI but it’s really going to become the foundation of what we do, of how we operate the business. And, you know, I’ll tell you, we did an analysis recently with some external partners and we had them great us where we are on the AI maturity scale, you know, I mean, we always we could great our own homework, but, you know, we’re always blinded by that pretty picture that we all like to paint. But we had our external partners kind of greet us and said, you know, no holds barred, give us a good assessment. And it was a really great story. I mean, it was something we were encouraged about. We you know, we were very. Where we need to be on a maturity scale. We’re really great in terms of our industry. We’re really great in terms of the whole landscape, honestly. And we’ve created a you’ve seen it as an environment that’s very nimble state of the art, is really effective and leveraging Databricks is a core part of that. It gives us the power and the flexibility to drive what we need to drive to achieve our vision.

Chris D’Agostino:
Yeah, I mean, I had the opportunity to be on campus, I guess it was last month, a few weeks back. And I, you know, I really appreciated just you and your leadership team talking to your broader data teams and the engineering teams around the goals for the transformation and showcasing some projects that were really changing things for your customers. And tell us a little bit about your approach you talked about evangelizing these things. You’ve talked about, you know, really trying to keep that balance between ensuring compliance and innovation can be met at the same time. How much is communicating to your team part of that equation for you and your leaders?

Andy Markus:
Yeah, I mean, it’s foundational. And when we talk about day to day, I obviously so I run a chief data office organization. You know, we are data and AI experts. We’re doing the right thing. I think we’re where we need to be. But AT&T, I think, has the potential to be a data and AI powerhouse. And we can’t do that if there’s only one or two teams great in data. So what we’re doing is we’re really evangelizing this across the board. I mean, our view and our goal is to democratize data and AI skills across all parts of the firm . And if we’re all great, we will be that powerhouse, will be that great company. And so that’s what we’re doing. The event you were part of was part of that democratization. Right. Its this great tool set shouldn’t just be available to data engineers and data scientists. Let’s find a way to expose it to smart people, throughout the business, to people that need it to do their job better. And so that AT&T, no matter what part of the business you’re in, is really leveraging the best tools to do what we need to do.

Chris D’Agostino:
There’s obviously, you know, the challenge for finding and hiring and retaining the proper talent. And, you know, there’s obviously, obviously a huge shortage in that area. Can you talk a little bit about what your approach is to that? And in particular, given that it is such a well-established company, the opportunity for upskilling existing employees to help them along that path, to really contribute to the data initiative?

Andy Markus:
You know, it’s another great question. I mean, you know, it’s been fascinating over the last let’s call it three or four years. We don’t have as many people moving. And on both sides of that slope, you’ve got people moving all the time, right? You’ve got talent, you know, changing positions throughout, you know, throughout the country, throughout the industry. But what I think we’re really doing, we’ve been playing right within playing that zero sum game, right? You know, we’ve got a fixed amount of good AI talent. It’s not enough to fill the demand. It’s just been moving from place to place. So what we’ve got to do, I think we’ve got to grow the AI and data talent. And the way we do that is we work with, you know, our academic institutions to get more people into that pipeline. And then to your point, we also unleash people within the business who have the capacity to do this with the right guardrails in place. And so two things specifically we’re doing on that. On the academic front, you know, we’re working with we just started a program with SMU, which is a great university, and we’re working with SMU to help them recruit underrepresented students into an AI program. And we’re working with their leadership, with the professors there at SMU so that the SMU provides the students with the theoretical foundation that they need to be a great as a scientist. And we have embedded our technology stack into the training so that the students learn tools, they learn the current AT&T Tech stack , and then they go to a bootcamp where they come to AT&T. They work hands on with our data scientist with those tools, and they solve real world problems. And then hopefully by recruiting kind of a group that’s been underrepresented, bringing them through this process, we’re creating a pipeline that just opens up a new world for AT&T and probably others, too, right, because these people are gonna be incredibly trained, you know, when they finish this program. So I think we need to do more of that. We need to find ways to get more talent into the pipeline. The other thing we need to do is unleash such people in the business. So we call it mobilizing the citizen data scientist. Right, the people. And it’s not our term, but it’s an industry term. But the people in the business that are not data scientists, but they’re subject matter experts. And if they had the right toolset and could create AI, effective AI, responsible AI, they can really bring their subject matter expertise and combine it with the AI and really supercharge the company. So we’re doing that. We, we talk to you guys about it. And again, Databricks is kind of woven into this. We’ve created a platform called “AI as a service” and it’s code based today. But the idea is to turn it into a low code/no code environment so that those smart people, whether they’re in finance or H.R. or anywhere throughout the business, can use this process, you know, attached to the AI data, the AI world that AT&T has, the real time features that we have build responsible but really effective models without overreliance on the data scientist. And so I think we’re pretty close on that front. And, you know, we’ve committed that. I think we can change the number of people within AT&T creating AI in the next, let’s say, 3 to 5 years by five X. And so that’s a big number. Yes. I think those are two things kind of non-traditionally that we’re doing to kind of change that zero sum game that we all play.

Chris D’Agostino:
So one, I mean, I think it’s fantastic the outreach to the academic community and really creating that pipeline right then, especially in underserved communities. The other aspect of this is you’re starting to shift from perhaps a centralized model of building off that data platform to really more of that federated approach of that Low-Code No code citizen data scientist, does that then put, you know, sort of shift the responsibility for maybe creating reusable libraries, some of the things that are harder to get done and get right, perhaps that your teams are building those and then, you know, sort of enabling or arming the different citizen data scientists throughout the different lines of business to be able to leverage those libraries and start doing more with, you know, drag and drop approaches to A.I.

Andy Markus:
100%. I mean, I think we’ve got to have the same framework. But more importantly, I think we have to work on the same data. And so from the data side of the world, we’ve always talked about a single source of truth, you know, democratizing that data across the company on the AI part of the world. You know, it’s often been that lone wolf approach. You know, everybody’s on their own building, their own models, their own solution. But things don’t often come together. So with the feature store, you know, now we have, you know, we call it like our water cooler for data scientists. We have that spot where all data scientists come for the best features across AT&T, and I think that’s foundational to this vision. But yeah, we’ve got to create that framework and the data that all data scientists across AT&T, whether in my team or in other teams, can use this to be effective.

Chris D’Agostino:
So let’s chat a little bit about ethics and AI and you know what you and your team are contemplating there or implementing today. Can you talk a little bit about your guiding principles for how you approach ethics?

Andy Markus:
Yeah, it’s foundational to what we do. The AI is a service platform that we’ve created, we’ve got an ethical component built in. So it’s a best in breed solution. We call it SIFT, the system for integration of fairness and transparency in AI. And what we have in the process is that every A.I. model goes through the SIFT process. That’s the vision that we have, not in practice today, but I think we’ll have it in place in the next, let’s say 12 to 24 months. Every A.I. model goes through the process. It gets SIFT-ed, per se, and SIFT does two things. One, it evaluates the model for bias, and two, if bias is detected, then it goes through a bias mitigation process to still try to hold as much prediction as possible but by removing the bias from the model and through the process, we’ve engaged our legal and privacy team. So everything is documented. And then we have the model documented as well. So we believe even as we democratize AI and that we build in the citizen data scientist. All of our AI should go through a fairness check, a bias check. It’s foundational that that AI is responsible.

Chris D’Agostino:
So as you know, like, you know, Databricks has coined this Lakehouse architecture concept. And, you know, without this being, you know, an advertisement for Databricks, I mean, we definitely appreciate the partnership. But just in general, the notion of a Lakehouse style architecture where you can load data in and run more and more use cases, more and more workloads despite what they may be, whether they’re, you know, data engineering, data analytics through sequel, machine learning, model training, all of that. How do you see the Lakehouse architecture helping AT&T?

Andy Markus:
I mean, it’s really important for us. I mean, it gives us flexibility. You know, it gives us the, you know, the ability to go from storage to analytics quickly. It gives us the ability for many groups to work on the same data, you know, in disparate parts of the company. Right. So it’s really been a core part of what we do. I mean, we use, you know, as, you know, other technologies as well. You know, I think we have what we feel like is today a best in breed architecture. And our goal is to always be flexible, always be interoperable, and always be looking for the future. And I think the Lakehouse concept does set us up in all those directions.

Chris D’Agostino:
Yeah, it’s great. Well, why don’t we, why don’t we turn to some closing questions and get your thoughts for those leaders out there that aspire to be the chief data officer of an iconic company one day. So what career advice would you give to people that are really looking for a career along sort of the path that you have? You know, if they’re starting out in their career today, what advice would you give them in terms of project selection, career navigation, that type of thing?

Andy Markus:
Yeah. I mean, you know, I think we’re talking to a lot of, you know, data and AI people that would tune into the podcast here. You know, the guidance I give to our team at AT&T and I’ve given to my teams in the past is that, you know, data and AI, you know, is in a science competition, you know, in industry. Right. It’s not just about doing cool things in technology. It’s about driving value to the business and to our customers. So really understanding, you know, how the two connect, I think is super key for somebody that wants to get into a CTO type role. Right, because it’s understanding both worlds. You have to understand the business aspect. You have to understand the science aspect. And it’s the marriage of the two that really creates the Holy Grail. And so that’s something I can’t stress enough. I mean, you’ve got to have great scientist, but to really be I think to be an effective CTO, you have to, you know , combine the two. And the other thing I would say, and it’s not technical in nature, but it’s really more about partnership. I just think, you know, spirit of partnership is so important to really succeed and to grow kind of a strong network. And I think that’s something if we all strive for partnership, doesn’t matter if it’s within our company or outside of our company, it just creates, you know, that path for longevity, you know, for things to stay longer. And, you know, something I’ve always done in my career and it’s just to get, you know, advice I always give to people is to really, you know, strive to be equal partners with everybody that you work with.

Chris D’Agostino:
Yeah, that’s great. The one thing that I’ve noticed in the role that I’m in, in talking to leaders like yourself, is it really feels like AI is changing the culture of a lot of organizations. With that in mind, like what do you see three years down the road in terms of what is going to do for organizations? You know, holistically?

Andy Markus:
We’re really invested in, you know, integrating AI into the fabric of what we do to relook at all of our operations, how we run our business, and to reimagine it with data and AI in mind. So I think that’s, you know, over the next three, five years, that’s just going to grow even exponentially. You know, we’ve you know, we’ve looked at the value that we create with AI today across AT&T. I’m not going to tell you the number, but it’s very impressive. But I think we’re still just scratching the surface. And I think, you know, by having more people across the business closer to the business with the capability to solve problems with data and AI, it’s going to really just, you know, just help us reach that potential that we have. But one of the things I think that’s really going to change is kind of the automation front. So, you know, we think about A.I., we think about automation. They go hand in hand. But, you know, automation for the most part, you know, has been, you know, almost sequential automation in the past, really cognitive or intelligent automation, you know, at least at AT&T, we haven’t seen that grow as much as we would as we will. I think we’re, I don’t know how many people know this, we’re Microsoft’s largest automation client, so we use automation to its fullest today. But I think the advent of really smart and cognitive automation is going to be a game changer so that we’re not just making sequential automation decisions, but that we’re making really smart automation decisions that, you know, the outcomes become much more complex. So I think that’s going to be a real big game changer in our company.

Chris D’Agostino:
Awesome. Cool. Well, Andy, thanks for your time today. It’s been great seeing you again. Of course, virtually, you know, we’ll see each other in person, I’m sure, before too long. But thanks for being part of Champions of Data and AI.

Andy Markus:
Hey, thanks for inviting me. Can’t wait to get back together in person.