How Comcast IT moved from order-taker to innovator

Rick Rioboli, CIO of Comcast since 2017, has a background in product engineering, a perspective that comes in handy as he leads in customer lifecycle management, data, and enterprise IT. In a recent conversation, we discussed how IT can strike a balance between order taking and designing the perfect platform. We also discussed how to combat data sprawl, drive a culture of innovation, eliminate technical debt, and bring a servant leadership mindset to the CIO role. What follows is an edited version of our interview.

Martha Heller: How do you describe your role as CIO of Comcast?

Rick Rioboli: I break down my role into three parts. One is the lifecycle management of our customers, which covers everything from when we first market to customers to when they buy our products, to when we activate them, bill them, and service them on the back end. That’s the biggest part of my job. The second is classic IT enterprise management, and the third part is data.

Can you give an example of your work in lifecycle management?

If you are customer who is considering the Xfinity service, for example, you might go to our website to search for the product, build a cart, and pay. That order flows through our back-end provisioning and activation service, which sets up the product in your house. My team drives that entire customer experience.

Let’s say you opt out of self-service and request a technician to come to your house. My team also built the mobile tool that helps the technician do the installs, and we manage the billing systems, online self-service, and tools that the agents use if you decide pick up the phone and call us.

What technologies are critical to managing that lifecycle?

We are using a number of digital technologies to drive the customer experience, but it all relies on clean data. When you have clean data, you can apply machine learning to the customer experience. People sometimes think of machine learning as something magical that will solve all their problems. But without clean data and feedback loops, you cannot train the machine.

What is one practical application of machine learning that you are using now?

The Xfinity Assistant is an AI-based self-service tool that my team built internally. We took the natural language processing technology that we had already built for a TV voice remote and transitioned it to our Xfinity Assistant. 

If you are on our website, and you begin to type a question about your bill, for example, the AI tool pulls in data to understand the context around your problem, figures out what you’re looking for, and makes a recommendation for how to resolve it.

If you click on that panel that asks “Do you have a question about your bill?” the machine knows that its recommendation was accurate. That’s a powerful feedback loop to let the machine know that, next time, we should put up the same message.

What are the challenges around having really clean data?

Up to now, we have delivered what I consider to be digital lipstick, which includes great capabilities like digital stores and self-service. The challenge of taking digitization to the next level is having an accurate set of data across the enterprise.

What happens in many companies is this: a piece of data gets copied, it is enriched with other data, copied again, then enriched with even more data, and then you wind up with data sprawl. This can happen to a wide range of data from telemetry type data about machine performance to ordinary billing data.

How do you manage data sprawl?

The first line of defense against data sprawl is giving people access to a single source of truth in a highly reliable, well-governed, and high-performing way. You need to nip data sprawl in the bud.

It is easier to manage data sprawl if you create a self-service data platform, where only the people who need it can access and enrich the data in a highly governed, reliable and rule-based environment. That approach is more valuable to your business partners than saying, “I’m locking down the data.” 

How are you driving a culture of innovation at Comcast?

When I first became CIO of Comcast, I saw that we had a culture where IT would get a bunch of technology requests from our business partners and we’d implement them. I was seeing all of these independent projects running in little silos.

To create a culture of innovation, we had to stop taking the orders and think less like renters and more like owners. Why run 15 different projects around a sales platform? Why not build one reusable set of capabilities?

We built the Xfinity Assistant on the same middleware and data layers we had already built for the voice remote. We became more innovative when we started to build new solutions on reusable layers.

What is the delivery method you are using?

My general philosophy is that when you’re building solutions in IT, there are two extremes: just taking the orders from the business and delivering or stepping back and building the ultimate platform. I tell the team all the time, “You should have a target architecture, but don’t expect anyone to fund it.”

What I mean is that every team should have a target architecture, but they need to be delivering near-term business value as they move toward that architecture. If they do work that delivers near-term value but moves them away from that architecture, they have to raise their hand and say, “’This is an unnatural act, it does not get us closer to our architectural goals, but we are doing it anyway, and we will need to pay some technical debt to get back on path.” That’s how we keep ourselves honest and find that sweet spot between the extremes.

What leadership skills do you rely on as Comcast’s CIO?

When I took this role nearly four years ago, my first message to my team was about servant leadership where, rather than the boss telling everyone what to do, the pyramid is inverted. Servant leadership is asking “What challenges are you having that I can help you with?” versus, “Here are your goals and this is what you need to do to achieve them.” Servant leadership is a different mindset. It’s what I try to practice myself and what I look for in my leadership team.

How do you test for servant leadership in your interviews?

Once I am confident that the candidate understands the technology in our space, the rest of the interview is about EQ [emotional intelligence]. Do they have self-awareness? Do they have empathy? Can they put themselves in other people’s shoes? Do they naturally try to understand other people’s challenges? Can they coach rather than direct?

But the ultimate acid test is when I ask myself after an interview, “Would I be excited to work for that person?” If the answer is no, then why would I ask the people in my organization to work for them? That criteria can knock out 75 percent of the people I interview.

One question I ask toward the end of the interview is, “If I asked someone what it is like to work for you, what would they say?” I get more out of their physical response to that question than anything else. How long did it take them to answer? Did they shift in their seat? Their body language is a telltale sign as to what kind of leader they are.