PODCAST | General Motors, CDAO: We are Over-Investing in First Party Customer Information

Tom Pohlmann: Hello and welcome to the CDO Magazine interview series. I'm Tom Pohlmann, Chief Marketing Officer with AHEAD, and we are partnering with CDO magazine, MIT CDOIQ, and the International Society of Chief Data Officers in a series of informative interviews. Today, I have the great pleasure of talking with Iwao Fusillo, Chief Data and Analytics Officer at General Motors. Iwao, thanks for joining. How are you today?

Iwao Fusillo: I'm doing great and thanks so much for the opportunity. I'm looking forward to our conversation today.

Pohlmann: A lot of digital leaders I speak to — whether they're focused on analytics or core IT or cloud, or something like that — they're all kind of grappling with the concept of scale, how to scale their efforts. Obviously, talent development is part of that. But how else do you think about scaling the adoption of data and analytics for better business outcomes at GM?

Fusillo: Let me focus on digital as an example. If I were to look at digital leaders and how they are using data to scale customer marketing on one hand and customer experience on the other hand, the talent picture is quite positive. The infrastructure picture is being disrupted right now. And why do I say that? It's the world of third-party cookies. What we're seeing is increasing pressure on the world of third-party cookies to where we may see that world not only be disruptive, but it may soon disappear. And so traditional digital leaders are becoming quite innovative.

They're looking beyond their traditional partners and players like Adobe who have long been in the digital marketing space but, in a land fueled by third-party cookies, are thinking differently. Digital leaders are looking beyond traditional players. Nielsen, the TV measurement company, might not be the most obvious player from a digital marketing perspective because they're the TV measurement company. Maybe they are a technology to look at. I'm not sure where that space nets out, but I do think really smart digital leaders, really smart chief data and analytical officers, really smart performance marketing professionals, will figure it out. Companies like General Motors are advantaged because we have more information. So, in the short term, to ensure that we can scale at GM, we are over-investing in our data pools on first-party customer information.

We're making sure to data engineer and quality-check it and make it industrial strength so that in any world, cookieless or with cookies, we’re advantaged to scale rapidly.

Pohlmann: Interesting. Very cool insights. Just a couple more questions. I think there's some confusion — and it's not a bad thing necessarily — around the definition of data and analytics that I'm seeing in the market. I kind of see a continuum of data and analytics capabilities. We've got entry-level, just better quality data, better reporting, and the next phase is better insights leading to better decision-making. And at the third level, I don't see a lot of companies using data and analytics for more advanced real problem solving and, in some cases, using it to even define the right problem to solve. You worked at a lot of companies, most recently with GM. Where do you see most organizations on that spectrum and why, and where are you focused as a leader at GM?

Fusillo: That's a great way to articulate the data and analytics space. I really love the question. I think most companies do the first level very well, the data and the reporting. I think many are graduating to using that data and reporting for decision-making. So, the middle part of your spectrum I agree with you. I think very few organizations have gone on the road of predictive and advanced analytics, both to solve problems and to identify growth opportunities. And I think much of that comes from really two things. It's business leaders truly understanding the power of predictive analytics on the one hand, and on the other hand, business leaders — particularly those who may not have traditionally been trained in data and analytics — do they trust it? Take machine learning and artificial intelligence — for all of the hype that we hear in the marketplace about ML and AI, I certainly marvel at how few large companies have actually deployed ML and AI broadly across their organizations. You have a use case here and there, and a proof of concept, but how many large companies have deployed broad scale?

I think there are two impediments there that can be overcome. We're moving fast and furious at GM on these things. With ML and AI adoption, there are two things that have held the market back. The first is, many companies jumped straight to ML and AI model development, and they got discouraged when it didn't work. Because it takes a lot of investment. You’ve got to hire seriously expensive talent. You have to build infrastructure, and they get discouraged when their first ML and AI models don't work. And the reality in many companies is, the underlying data sets were not yet properly governed, quality checked, or engineered — a theme we talked about. There's a new term out there called ML Ops — ML operations. And it's a process by which you start with the data. I think Google actually has written a number of papers about this; they're great reads for anyone in the audience.

They rightfully point to studies that show that we often see a 10 times the return from investing in data governance, data engineering, and data quality versus investing in the model. That's one key impediment. The second is explainability. For those companies that do ML Ops well and can get past the data piece, they get discouraged when senior leaders don't trust the AI black box. And then what they do is, they default to your earlier point — reporting and descriptive analytics. What AI really needs — and the technology is not out there in any production way — is a glass box. It's a glass box for senior leaders to clearly see the data inputs and the logic that creates the outputs.

I've met a number of startups who are beginning to scale this glass box technology. It's pretty hard to stand up. But I think that one's going to be a huge game-changer when it moves from startup to more production-ready solutions for companies like GM to use. It's a challenge.