Data Science Leaders | Episode 35 | 27:12 | January 25, 2022
Data in the DNA: Breaking Down the Autonomous Enterprise
Get new episodes in your inbox
powered by Sounder
Is your team mining all available data to inform your business strategy and grow revenue? Is your company prepared to compete against others who are?
If you’re like most, the answer is probably no.
How can you future-proof your organization and take steps toward an autonomous enterprise?
Janet George is an enterprise AI leader and author with experience across companies including Oracle, Apple, Accenture, Yahoo!, eBay, and more. She joins the show to discuss the meaning of autonomous enterprise and the process required for true transformation.
- What is an autonomous enterprise?
- Where are companies falling short in their data transformation?
- The investment and first steps required on the transformation journey
- How to prioritize data projects for a larger impact on revenue
Welcome to another episode of the Data Science Leaders podcast. I'm your host, Dave Cole, and our guest today is Janet George. Janet is a Group Vice President of Autonomous Enterprise. She has worked with a bunch of companies in Silicon Valley, including Oracle, Apple, eBay, Western Digital, Accenture, and more. She's currently co-authoring a book to assist companies in their AI transformation. Janet, how are you doing today?
Great. Thank you, Dave. It's a pleasure to be here with you.
Awesome. You've worked with many of the large names in Silicon Valley so you're bringing a wealth of background. The question in my mind, right at the outset: what is an autonomous enterprise?
When you think about an autonomous enterprise, what you want to do is think about it as actually being powered by ML and AI at its core. It’s in the DNA of this company to treat data as a first class citizen. What that translates to is really having all of the modern technologies in place: to mine your data at scale; grow your top line revenue and cut some major operational cost.
A lot of companies right now are undergoing digital transformations. When I ask folks who are in that process about it, I feel like I get a different answer every time. Is that your experience as well?
Oh yeah. Very much so. It's not clear. We've been doing digital transformations for a very long time now. For the last decade, we've been doing digital transformations. But if you ask me about these companies’ preparedness to handle AI and ML as an existential threat in the next five years, the answer is no. Are these companies storing and mining data to actually inform their business and grow their top line revenue?
The answer is no. They may be doing it in small snippets but they're not doing it as a whole. You can think of these trillion dollar companies as actually mining data as part of their DNA. Take Amazon, Apple, Google or any of these bigger companies. They're all data-driven companies. They are prepared to handle the ML/AI existential threat. That's what we are talking about when we ask if these companies are doing AI transformations or data transformations different from traditional additional transformations.
That makes sense. You've said existential threat a couple of times. What threat specifically are you referring to?
When you're not powered by using data, you're basically going into your market a little bit uninformed. That data isn’t driving your business. Let me give you a specific example of this. Do you have a really good understanding of your customer base? Is your data telling you what these customers are consuming on a daily basis—not the first level of BI analytics, but double- or triple-clicking deeper down? Really understand your customers.
Let's say you were a pharmaceutical company and you sold a lot of products. In the world of AI you might need to change your business model. You do that by using AI to understand how to provide personalized help to your customers. It’s quite different from menu-selling products. Let's take a traditional storage company. Let's say you were basically just selling storage products. In the world of AI, you're not just doing storage products. You're doing a lot more. You're going up the stack and you're going to actually mine the data and look at what your customers are using on a day-to-day basis. Where is the need? What do you need to give them in order to serve them?
So, it's using AI, machine learning and data to better understand your customer. Companies have been creating data warehouses and doing analytics for years. I feel like you're advocating for something slightly different. You said that it’s about going beyond your typical BI, which involves looking backwards and doing descriptive analytics. When you are working with companies, where do you feel most of them fall short of that vision of the autonomous enterprise?
Let me give you another example. Maybe I'll talk about the energy sector so it can be really clear. Let's say in the traditional digital or data transformation, you were looking at data, mining it and producing dashboards. The dashboards basically told you, "Hey, I've got all these wind turbines and they're in Iowa and they're massive wind turbines." You look at how these wind turbines are performing. When are they going up? When are they going down? What's actually happening? That would be your traditional digital transformation. You have dashboards and BI—a good sense of the health of your company.
In an AI or ML world, you're not just looking at when your wind turbines go up and down. You’re rather trying to predict. What are you predicting? For this example, that might be icing conditions in winter so you can save up to $500 million in operational cost. You're doing temperature testing, temperature sensing or wind predictions. These things give you a lot more depth of insight about your wind turbines, rather than just reporting on the health metric of the turbines. You look at your business far more holistically.
It sounds like the difference is using data from your company, that you've been collecting for years and years, to better understand your business and use AI machine learning to actually make decisions autonomously. That would mean doing so without human intervention. That makes sense, but then what role do humans have in the autonomous enterprise? When I think of that term, I worry that there will be no jobs. We're just all going to be replaced with models. Is that what you're saying?
No, that's not what I'm saying.
I'm saying that humans will never be replaced. They'll not be replaced by machines because we'll just keep going higher and higher. Look at the car industry. There are cars that run autonomously, but then is a human required or not required? They're still in the car. It's just going to save us time. They're not going to do the driving but they're still in the car. What we do as humans is what’s going to change.
Today I'm spending a lot of time mining and looking at data, format quality issues and schemas. It’s still so far away from being an autonomous enterprise. If I move closer to that spectrum of being an autonomous enterprise then I'm mining my data automatically. I'm streaming this data, looking at my data quality checks through machine learning agents that are actually going to look at that data. I'm no longer going to be very schema-bound on the data. I'm not going to enforce the schema from the get-go. I'm going to have a dynamic schema. My data will onboard itself rather than have 100 people sitting there onboarding it into these traditional application tools. Think about it. That's what I'm calling an autonomous enterprise.
You're cognitively automating many of the first and second order tasks that are manual and very time-consuming. We're not talking about level five where the enterprise completely runs on its own and without people. That's not what we are talking about. When we say autonomous enterprise, we are saying data is really pivotal to the organization. It's a data-driven culture. We are going to use intelligence, not just data. We're going to convert data into intelligence that's then going to drive our markets, our customers, the way we operate and the way we build products. It's going to inform every aspect of our business.
That's what an autonomous enterprise is. It's going to be powered by data. This wouldn’t just be data that is being created internally in the company, but data that's also created externally. You've got third party data, diverse data sets and data coming in from the Internet: product reviews, product information, customer usage patterns, information about the weather affecting your business, if it does. If you're in the energy sector, weather is certainly a big component that affects your business.
Depending on all the aspects of data that's affecting your business, affecting your competitive state in the marketplace, you're going to mind that data. Today we don't operate our businesses looking at every aspect of the data. We do it manually and without involving science. We do it sort of haphazardly. That's what's going to change in an autonomous enterprise with the leaders versus the followers.
Got it. So if I want to be a leader, other than reading your book, what recommendations do you have? Where should I start? Let’s say I want to create more of a data-driven culture than I have today, or I want to understand that gap of where we need to be versus where we are today. What should we do next?
The first step to being a data-driven company is to really address the culture. Are you ready culturally as a company to become a data-driven company? If you don't address the culture, no strategy is going to be executed well. Culture will eat strategy for breakfast every single day. We want to tackle the culture first. After that, let's do a data maturity assessment model.
How far along in data maturity are you: level one, level two? Have you adopted modern technologies like the cloud, autonomous security patching, autonomous provisioning, autonomous everything from a software stack? Do you have the right modern technologies in your company? Are you still powered by primitive technologies like sitting in a data center, bare metal, operating on old compute storage technologies, networking, everything old, versus newer cloud technologies, edge technologies etc.. And then your tool: are you really operating on a data science platform? Do you have these platforms in place? Are you solving for outcomes? What outcomes? Are those of strategic importance to the company?
You mentioned culture being critical. What specifically are you referring to? How do you arrive at that culture? Are you talking about changing the mindset in terms of how decisions are made, like being more data-driven, less sort of finger-in-the-wind?
Yes, that's right. That's what I'm referring to.
Also, it's a board level commitment. The board has to be very involved in the transformation process of a company. It can't be just the executive team that's deciding to suddenly become a data-driven company. This is a board level strategic planning, where the company has to decide to become a data-driven company, to better compete in the existing markets, go into new and adjoining markets, adjoining customers etc.. Once that conversation and commitment happens, there is investment that needs to flow. You can't take a company that's 30, 20 or even five years in operation and suddenly change its business model. You can't do that. You need the board level commitment on the investments.
Once you get the investments then you're prepared to discuss the tactics and maturity. Where are we with our technologies? Are we adopting the cloud? Is it a public cloud? Is it a private cloud? Is it on-prem? These technologies are just infrastructure that enable the outcomes. You really have to focus on what the outcomes are that the company's going after. How do we grow our top line revenue? What does that look like? How will data power the growth of our top line revenue?
Let's say that you are a manufacturing company and all you did was sell storage your whole life. Now you're going to actually become an autonomous enterprise. You're going to go beyond selling storage products and mine the data that is sold inside your storage products, with the permission of your customers. Then you're going to add a whole lot of value-add services for your customers, creating entire billion dollar revenue streams. This is because you've not just become a storage provider that's selling storage products, but you've now actually gone and mined the data that the customers are storing within your storage products; you're offering them a whole lot of new value-add services. That's one new business line that you can create from that.
A second one would be to say, maybe in the autonomous car industry, that you’re going to store all of that autonomous car data into your storage devices. Then you’re going to mine that data and provide near real-time directions with that data, or some other kind of value. This is what we call autonomous enterprises.
Amazon is a great example. They started with books and very quickly went into the cloud business, then Kindle and now it’s all the different businesses. Now they're getting into healthcare. How are they able to do this? If their primary DNA is books or e-commerce, and they're going into a completely separate business, like infrastructure, how are they able to do that? That's because they're very much a data-driven company. Facebook is another example. They started out with social media and now they're getting into metaverse. It’s a completely different business, if you think about it from being a traditional social network.
My understanding is that AWS was born from how Amazon ran their ecommerce business. They realized that what they built internally could theoretically be sold as a service to other companies out there.
Exactly. And how did they know that? They were very, very good—they had an obsession—with their customers. They mined pretty much all the data so you knew what books you were buying. If you bought this book, you might like this other book. So there was a lot of collaborative filtering, machine learning algorithms at work behind the scenes, mining the data, knowing this is what the customers will want and need, and then making decisions around how to grow the business.
They had a whole dynamic process for standing up infrastructure. Providing that as a service was a natural extension, and obviously a great idea, because AWS has been growing in leaps and bounds and it’s incredibly profitable for Amazon.
I think that could be quite daunting though. Not everybody has the talent of, say, an Amazon. How do you see a company getting from A to B? I mean, yes, you spoke about the board level. It has to be a board level decision and there has got to be a certain amount of investment, but where is that investment? Is that investment in people? Is that investment in tools? Where are you putting that?
That investment has to start with modernizing the technology stack. It has to go into the people and processes because it's not important to just bring the technologies in. People have to consume the new technologies and what is being built. The investment is broken down into different aspects of growing that data-driven business. The first aspect would be modernization of the technologies and the tools. The second is investment in the people. How do you up-skill, re-skill, and get your people onboard with the new business strategy of how the company is going? The last one is really about enablement: enablement through processes; enablement through structure; enablement through access of the new data, if you will, that's available to other people.
What role is driving the company towards the autonomous enterprise state? Obviously the board and entire executive team has to be on board with us. There also should be a champion who's really leading this transformation.
This is why we’ve seen this in the last five years, since I was the chief data officer for Western Digital. We are seeing this massive growth of jobs entering the C-suite, which is the chief data officer, chief science officer, chief AI officer or the chief scientific officer. These jobs are coming in massively now. If you just Google and see how many chief data officers have grown in the last three years, the numbers are pretty amazing. Big growth there.
Everyone's hiring a chief data officer. Everyone's trying to get on top of their data and trying to run their business more intelligently. It is primarily the focus of the chief data officer, chief science officer, chief AI officer. It's a little bit different from your CIO. It's partnered with the CIO, but the CIO is traditionally more vendor management and all of that stuff. Your CDO is really responsible for making the business, the DNA of the business, more data-driven.
One thing I hear is that traditionally a lot of the tech leadership roles like a CIO, even maybe a CTO, are typically seen as supporting roles. The business kind of comes up with their ideas and then the CIO makes it happen. If there's some large strategy, the CTO is figuring out what innovative tools are out there that can help the company move towards that overall strategy. I hear your advice to think a little bit differently; a chief data scientist or a chief data officer is no longer just a supporting role to the business.
They also should be coming up with their own ideas, trying to pull the company towards a more data-driven culture and actually putting elements like models at the forefront of decision-making. That potentially could open up new revenue streams and the like. Do I have that right?
That's absolutely right. So, exactly, Dave, what I'm talking about is the chief data officer is not just assigned a supporting role like a traditional CIO, but is chartered with creating the new cash cow for the company. If there's an existing business, that's great, but let's mine the data and actually work with strategy: the chief strategy officer and CEO of the company. Let's figure out some new businesses we can incubate from our position of strength, from our position of core business. What are some of these new businesses that need to take off based on our data, our customers, what we see as top line revenue growth, what we see our competition doing, what see our strength in the marketplace? How do we grow these other businesses that we can go into adjacent markets and play very well at?
Uber is a great example. Uber started out as transportation disrupting the taxi industry completely. Then it was Uber Eats and there will continue to be many, many Uber things as we go on. Who would have thought that Uber would go into Uber Eats? It was a very natural adjacent business when you think about Uber coming and picking you up from your home. Uber is now delivering food. Uber might deliver groceries. Uber might do a lot more. Uber might help you with your post office events—who knows? You know what I mean? It can just go into all of these different things. Ultimately what is Uber out there doing? It's trying to save you on time and transportation. What are all the other areas you might want to go do? Maybe you'd have a monthly subscription and Uber will do everything for you: take you to your doctor's office; do your groceries etc. You've got an Uber at your beck and call. It's not unimaginable to think of a business like that. That would be a great prospect for Uber.
Yeah. For sure.
So let's switch gears a little bit. You sold me on the autonomous enterprise, I think. Many data science leaders out there are trying to get to a place where they're no longer just in a supporting role, but they're actually coming up with new strategies and pushing the business in new strategic directions. Once you've gotten that change, that culture, I think the next step is trying to figure out which of a chief data officer’s 50 projects needs to take priority.
I think you have a particular point of view on this. I'd love to tie into that.
Yes, you're right, because many companies are unable to focus and prioritize on the one, two or three projects that will give them top line revenue. I always think that there's a framework to do these assessments. One of the frameworks that I've used very successfully in the past is for assessing these top strategic use cases. The assessment comes with a certain ranking score. We rank these use cases based on the strategic importance to the company if we apply a specific use case.
Next, we talk about the feasibility of that use case. Can we actually execute on this use case? Is there enough data? Is there enough technological support to tackle this use case? Then we actually talk about ROI. What kind of ROI are we going to get from this strategic use case? Is it tens of millions of dollars? Is it hundreds of millions of dollars? What are we talking about in terms of numbers? Then we talk about the investment required for the strategic use case. We also talk about consumption. If we were wildly successful, what would that consumption model look like? What would that scale look like? We evaluate it on three to four different scoring mechanisms. Then those strategic use cases that surface to the top, based on those high scores, are filtered. You pick one, two or three of them and start executing.
First of all, is it strategic? Does it align with your company's strategic goals? The next question is the feasibility. Do you actually have the data or do you need to go find it somewhere? Is it curated? Is it usable? What state and quality is that? The next thing you need to think about is the ROI. What sort of investment needs to be made in order for this project to go live and then, potentially, what is it? Is it going to be lowering the goal here to become more efficient and lower costs in some way? Or is this actually a revenue-generating product? Is it a new service?
The last thing I heard is just how is this project going to be consumed? When it does go live, when you roll this into production is it, I don't know, an API endpoint? Is it a data product where you're sending a cultivated data feed to an external party? How does it actually reach the end consumer, whoever the intended audience is? For any one of those things, if it doesn't have a high score, it can bump it down the list. If you look at all those things and you rank them on a scale of one to five, or whatever, then that's a good lens by which to help prioritize what you should work on next.
Well, I learned a lot today, Janet. I really appreciate you taking the time. I learned a lot about the autonomous enterprise, and if I want to know more, I mean, you're writing a book. What is the best way for me and our audience to get the book?
I've already gone all the way up to chapter nine in terms of articles. I'm gauging responses, putting them out there in small articles, so people can read each article starting from chapter one. Then look at your maturity assessment, business strategy, culture, talent assessment. How do you assess your talent? Look at how you pick talent. What are the tenants to modernize your infrastructure etc.?
From your modernization, how do you create an advanced center of excellence? What does that advanced center of excellence look like under a CDO or a chief scientific officer? Then what are the components within that center of excellence? How do you pick what to execute on, once you have your center of excellence, and then go from there? Those are the articles that I've already written. Those are chapters one through nine. The others are coming. I'll be winding up with a bunch of use cases that have actually brought top line revenue for a lot of the customers as well.
Very interesting. So, the bottom line is to go to Janet George’s LinkedIn page. If you look at her activity feed, you can see all the chapters there. It sounds like there's more coming, specifically use cases.
Thank you, Dave, that’s right.
Thank you so much, Janet. I really appreciate you taking the time.
Listen how you wantUse another app? Just search for Data Science Leaders to subscribe.
About the show
Data Science Leaders is a podcast for data science teams that are pushing the limits of what machine learning models can do at the world’s most impactful companies.
In each episode, host Dave Cole interviews a leader in data science. We’ll discuss how to build and enable data science teams, create scalable processes, collaborate cross-functionality, communicate with business stakeholders, and more.
Our conversations will be full of real stories, breakthrough strategies, and critical insights—all data points to build your own model for enterprise data science success.