Today’s guest speaker has led technology enabled transformation with global enterprise companies. From her leadership at Booz Allen Hamilton and KPMG to her new book ‘Enterprise AI-Your Field Guide to the New Business Normal’.
Xena Ugrinsky shares why spreadsheets enable modern analytics, what technology enabled transformation can do for your business and how traditional companies can transform to remain competitive.This is HumAIn.
Welcome to HumAIn, My name is David Yakobovitch, and I will be your host throughout this series. Together we will explore AI through fireside conversations with industry experts, from business executives and AI researchers to leaders who advanced AI for all. HumAIn is the channel to release new AI products, to learn about industry trends and to bridge the gap between humans and machines in the fourth industrial revolution. If you liked this episode, remember to subscribe and leave.
Welcome back to HumAIn. My name is David Yakobovitch and today I have a special guest for our audience. Xena Ugrinsky has spent over the last 20 years of her career in technology enabled transformation. The last eight years of those have been focused on the application of data science for corporate performance management. She has a new book dropping on the shelves this month, titled “Enterprise AI-Your Field Guide to the New Business Normal”. Welcome to the podcast Xena.
Thank you, David. Happy to be here with you.
Thanks so much for being here, here as well. I’m fired up about analytics. That’s actually how I got started my career back in the 2010 or so I was in actuarial science.I was doing analytics and spreadsheets and IBM mainframes and using COBOL. And it’s amazing how organizations have been transforming themselves in the last decade. And there’s so much to take advantage of today. How do you think the field’s changing?
Let me share with you that I have been around business and technology since the advent of the spreadsheet. That probably dates me quite significantly, but I’ve had an active role in the implementation of technology and analytics specifically for the last 31 years. So we started out seeing technology as a way to do our jobs more effectively. And first it was really process optimization, sort of moving from standalone, flipping a floppies 10 and having to bring something member Sneakernet, having to bring something across the building to someone else as opposed to email, because email wasn’t available yet.
And then we saw the advent of the land. I can just send somebody this analysis that I did and show them whatever results that I need to communicate. For me the first cloud was when wide area networks were developed and suddenly you could put a server in a building across town and use it as if it were in your building. The transformation has occurred in the use of technology, in three distinct buckets, the first was better individual tools for doing analysis. So there you have your spreadsheet. And then you had basic BI, but if you think about what basic BI is, it’s looking in the rear view mirror.
And there’s a reason why the windshield is so big and your rear view mirror is so small because what’s important is what you are hurtling towards in your car. And what analytics began to enable in the second phase was the ability to do things like driver based forecasting trending analysis. In 2000, I worked with a major beverage organization and it was the first time that we pulled global weather data into our forecasting models. And the third phase and the one that we find ourselves in today is the beginning of applying truly modern mathematical methods in the form of data science to technology.
And the thing that makes this possible today, is that the technology has now become robust enough to be able to enable it. I spent the last 25 years of my career breaking technology because I knew as a business person, what I needed to accomplish, but the technology could not support that vision. And so we find ourselves in the world we’re in today where we are only beginning to tap into what technology has now made possible.
We as humans have to catch up. There are two anecdotes or analogies that I like to tell around this and my first almost 10 years in my career we’re based in finance. So I was a consumer of data and I wound up just getting frustrated because I couldn’t get from IT what I needed, reports were wrong, the data that I was being provided was wrong. And I finally gave up and started building my own solutions for the things that I needed to do.
When I was in finance I’m thinking that 50 miles per hour and it can only give me 30 miles per hour worth of content. I had to figure out how to close that 20 mile per hour gap? I did it by building my own solutions right. Now fast forward to today. Technology can do 2000 miles per hour, and that’s just what we’re aware of at the moment. So if we’re still thinking at about a hundred miles per hour, cause we have evolved in our thinking, we still have a 1900 mile per hour gap that we have to close.And if technology isn’t the hurdle anymore, it’s people, it’s process, it’s culture, it’s organizational structure.
How about spreadsheets? That could be part of the problem.
I spent seven years of my career selling enterprise software, specifically multi-dimensional analytics. And what I found was 80% of the time I was competing with spreadsheets, not some other competitive product.
It’s amazing how here we are in 2019 and analytics has continued to evolve and spreadsheets are still in the game. I know earlier this year, Microsoft mentioned that now spreadsheets are going to be powered by Python. The whole C-sharp and visual basic programming language that powers macros and the behind the scenes for spreadsheets is changing to Python. Do you think that’s good for spreadsheets?
It is because it connects your spreadsheets to a single version of the truth. When I was competing with spreadsheets, What was the predominant complaint by the prospect? It was not that spreadsheets don’t work. It’s that spreadsheets accumulate like Tribbles. And suddenly you have islands of information everywhere with no version control and no idea of what changed.
And this was important during the old lap times,the multidimensional analytics phase that occurred between 1997 and 2005. That was really transformational for business because for the first time they could connect their spreadsheet two and OLAP engine and an all app cube or a set of OLAP cubes. It created a window for them to navigate and have a real time dialogue with their data and that was transformational for business.
At the same time, those applications came fully loaded with almost 500 fairly sophisticated statistical functions that you could apply. So we were doing things like Monte Carlo´s Simulation using spreadsheets in front of an OLAP package. Those were Essbase, TM one, DB two, all lab. Those were the three major products in the market. So the nature of spreadsheets and having ultimately come from finance, you will never take spreadsheets out of the equation, right?
People want to take their information, put it in a spreadsheet or suck it ideally into a spreadsheet and take it off and maybe do some work and then come back and reconnect and have it uploaded or updated in the database, but they’ll never take spreadsheets away. It is too deeply ingrained in how business works.
Now you coin in your new book that we’re entering a new age. It’s the beginning of intelligent enterprise. I want to know what made the enterprises not intelligent prior to now.
That’s very simple. It’s the siloing of functions. If you look at the average organization’s technology portfolio, you may have a data warehouse, you may have an HR application, you will have financial applications for financial reporting. You may have planning and budgeting applications. You may have supply chain management applications, marketing applications, the list goes on and on.
And what we saw in the 1997 to 2007 range, was the beginnings of an understanding across all organizations that siloing was causing a bottleneck of data that prevented your executive team from having the right information at the right time at their fingertips to make a decision.
And so this realization happened in that period of time that you need something that is sitting on top of all of that spaghetti, that allows you to aggregate the data from all of those separate systems in an appropriate manner to deliver it to decision makers.
These decision-makers are in every single company, right? Whether you’re in the sales organization or the operation side of the organization or the product transformation part of the organization. Transformation has to occur everywhere and I think the challenge that you mentioned is being siloed.
I traditionally also come from the finance background, working for the big banks and back office, and they are everything siloed. They would almost be mind-boggling why the same brand new tools were being developed by different software engineers in different teams creating this redundancy of work.
You’ve also mentioned in your book about a transformation roadmap, how we’re moving into a new paradigm. What’s a way that enterprises can start thinking about how to be as effective and efficient as these intelligent enterprises?
The intelligent enterprise is not a new concept. If you Google it, it goes back to sometime in the seventies. And the challenge in getting there is this going back to, what is the Nirvana? What is the vision? At the time, there was already an understanding that we were moving to these sort of siloed data silos standalone data silos.
And how do we break the walls? And start considering the enterprise as one entity that can be optimized. And during the transfer, what I call the first wave of big data, 1997 to 2007 timeframe, that’s where there was a major adoption of software packages that purported to sit on top of everything else and enable one vision of the truth. One version of the truth everybody accesses the same data. I was in finance also, I can’t tell you how many CFO meetings I walked into where the CFO puts up a slide and says, these are the results and half the room is looking at their own spreadsheet and going through those aren’t my numbers like where did those numbers come from?
And so in an effort to solve that, there was basic BI against the data warehouse. One version of the truth, but getting a data warehouse stood up and creating BI that satisfy everyone’s needs wound up being IT´s job while they didn’t want to be the report writers or the dashboard writers of the organization.BI tools began to evolve into more of a surface self-service approach. That was great for a time until I walked into organizations and this is right around the 2000 time mark and I would repeatedly hear, we have like 18 different BI tools and like most of them are shelfware.
It highlights the problem that you have from an IT perspective. You have analytics application proliferation. As these tools become more accessible to the business community, they start buying what they think they need and then the IT organization completely loses control of their tech portfolio and managing costs, for example. So where we then go is into where we are today. So our ability to do analytics at a level of sophistication and scale, we’ve never been at this point.
The things that have to change in an organization to become an intelligent enterprise don’t involve technology at all. That’s why I refer to it as it’s technology enabled transformation it’s not technology transformation. So the evolution towards becoming an intelligent enterprise means you have put an infrastructure in place that allows you to accomplish it. A single view of your organization in a completely interconnected way from an analytic standpoint.
But now we have to turn our sights to the complaints we always will. I’m not getting it from a business person. I’m not getting what I need from my BI solution. We’re not doing budgeting and forecasting in a coordinated way. Now we’re talking about process and ensuring that the process doesn’t just mimic the old process that we re-engineer the process to take advantage of what is capable today.
The second thing is people. We hear so much about the job loss that we are heading into the automation of jobs that will require the rescaling of our workforce. And so if we don’t take that into account on our journey to become an intelligent enterprise, we will not reap the returns that technology promises that we can.
The third and fourth pieces to this. And this is one that really has only risen to the forefront in, I want to say the past 10 years or so, is this concept of culture and they always say culture trumps organization every day and that’s for a reason. So you have to begin to develop a culture of inclusiveness, of culture that moves away from information as power to empowering everyone with the information and further we have to teach the workforce. Not necessarily how to become data scientists, but we have to help them become comfortable with what technology can do for them today and how that allows them to think about their business problems differently. Very similar to the adoption of the spreadsheet.
There were plenty of people who said, no, thank you I’ll just stick with my pencil and paper and of course they were left in the dust by the people who successfully applied spreadsheets, the same holds true here. So those organizations who continue to do things the way they always did them, will fail. They may not exist five years from now because they will not be able to sustain any competitive advantage.
And the last thing I want to say on this topic, and this is really crucial, the kind of massive organizational change that we are describing here cannot be done as a grass root effort. It has to be owned at the top. And so what does that mean? That means that our traditional organizational structures need to be rethought. And it starts at the top with re configuring the responsibilities of the C-suite. Now this is not to say that all of a sudden the chief marketing officer needs to hire a data science team and the HR person needs to hire a data science team. This has to be collectively owned. So I’ll give you an example of why this is so important.
There were two trends happening in the market today. The first one is everybody is running out and hiring a chief data science officer. Or a chief analytics officer and saying, congratulations, you now own how we deploy data science to the entire organization. Well, what’s happening is the big banks are doing it. Multinationals are doing it and those people are only lasting for maybe two years. So let’s talk about why that is. The first thing they’re asked to do is optimize the business, solve, find out what we can apply data science to and go do it. But the reality is that it is going to touch every function in the C-suite.
So it is going to whether you’re the chief risk officer, you need to have a voice in this. What about privacy and bias? And diversity, right? How are we measuring those things? Your chief legal officer, what are our GDPR compliance? Is this cyber security secure? So that brings your CSO into it, right?
Your chief marketing officer. They can optimize through this, but it’s ideally going to be connected to what is sales doing? What are our products? A marketing organization doing our product management, our organization doing, and we find ourselves now in a place where we have to rethink the collective ownership of this responsibility.
With this new ownership on people on re-skilling on culture. As we’re moving into a new paradigm where this is our fourth industrial revolution, our intelligent enterprises, do you think is AI? This new term is AI, the new spreadsheet?
AI for me is still the catchall right now. I’m happy to report that the last two years have provided what I call a firming up of the taxonomy of the nomenclature.Very similar to how in 2010, I think I did my first presentation on what is big data. And you asked a hundred people, you got a hundred different definitions. And that has been the case with AI, with machine learning, with neural nets, with deep learning. And the last two years have really helped us to evolve what that means. So AI as a whole is not going to be the new spreadsheet, AI as a whole describes the transformation of applied data science.
How do you think about AI for organizations today?
I see organizations falling into three categories. The first are those organizations who have been built up and are architected as a data company.So those are the big guys that we all know and hear about every day.
The next group are those traditional organizations that have been early adopters to solve something or to embed it in their product. And I’ll use the sleep number bed as a perfect example. When the sleep number beds started out, they were just a bed that went up and down and your side could go up high and the other one could not go up quite as high and everybody was comfortable and then they made it so that you could also dial in your softness.And really all it was was a mechanical bed. They very quickly realized that if they embedded sensors, they could prepare for a time when the bed itself learns that when you are in your deepest sleep, it is in this configuration so when you start getting rested, it will correct to that configuration to try to get you back to that REM sleep.
And by extension, now they have a strategy or evolved into a strategy where they’re thinking, okay, if we have all this data about what’s going on with the person sleeping, we can identify, for example, a sleep apnea event that they should be checked for. And then by extension connecting it with their fitness tracker or their heart monitor, et cetera, we could conceivably pull together a set of biometrics that allow us to alert a customer to the fact that they may have a cardiac problem going on. So they have very skillfully evolved from a traditional mattress manufacturer to a data company.
Right now that brings us to the third set of organizations. And these are the organizations I spend most of my time with. They know they have to do something. They have perhaps hired some data scientists, maybe a year and a half ago and they’re still wondering from a business perspective, when am I going to see something out of this group that is transformational? And so I devised and part of the book is about this,what is the right way to get started in data science? And it begins with strategy.What are the biggest rocks that you have to move as part of your strategy and how can we apply data science to those use cases?
And what you do is you solve one problem and you build a community around it. So data scientists hate when I say this and I’ve had many,I’ve managed teams of them. If you put a data scientist in a room along with a problem you have about a 50/50 shot of them coming out with something that’s useful for you. They may get distracted by interesting things they find in the data, or they may not arrive at the right decision based on unfortunate facts, but that’s how they’re reading the data. Magic happens.
When you line a data scientist up or data scientists with a business person or a business person who has a tremendous amount of expertise and depth in the subject matter, it becomes a much shorter iterative cycle to get to the right answer because the business person can provide feedback on what should go into the decision or the outcomes that we’re trying to drive towards and investigate, and they can and look at some of the results and say that doesn’t feel quite right to me, It doesn’t look quite right. Perhaps we’re not answering, asking the question correctly, let’s try asking it a little differently.
When you do that and you put that answer into production, whether it touches your supply chain, or what have you, or it decides product mix for you in various geo locations and departments, etc. retail is an early adopter of all of this. At that point, you have built one solution and wrapped a community around it. And you have to educate that community. This is where we begin the re-skilling of the workforce. And so now you may have created what I like to call a data puddle. It’s not a data Lake, but it is the beginning of what you will build upon.
So, when you answer question number two. And it needs to tie again, directly to strategy you want the big impact, very material results to be part of your early efforts. So you answer question number two, you have incrementally created more community. You have expanded the data puddle. A data pond and you do this step by step And before you know it, you have a community that understands how you use the data. So it isn’t mistrusted. I can’t tell you how often. I’ve been called by a customer who says we built a data Lake and no one came. Happens all the time. You were dealing with ownership of information. You’re dealing with a mistrust of something new you’re dealing with behavioral change.
These are all human problems. Not technology problems. And so I often do workshops with C-suites to help them sort of gel on this concept to help them understand the right way to install a chief data science officer, and then how to help that individual interact with the group to drive towards what problems we should be solving. Not putting it on that one individual’s plate, making it a collective decision by leadership, because what that then fosters is communication down from the C-suite that is consistent ,and it begins to break down those barriers and the silos between the information. And it begins to change human behavior.
So there’s honestly, no new magic to this. We are merely at a point where we are forced to solve the problems that we gave short shrift to in the past. So if I buy explanation, when I was selling enterprise software and it would touch 5,000 people perhaps for the budgeting and forecasting on a global basis, integrated with our BI global BI platform, ectc, we had to change behavior, but what was the first thing that someone struck from the order schedule, change management and training. And the second they did that. I was like this project will get them only 60 to 70% of the hoped for return on investment. So that’s what makes the human element so incredibly important.
Xena, your new book “Enterprise AI-Your Field Guide to the New Business Normal” talks about the field guide. Why is now the right time for this new field guide?
The reason I wrote the book was everything I read was chock full of use cases, but nothing told me how to do it. And it is also very few and I think it’s only beginning to bubble up. Now this concept, actually the concept of enterprise AI is less than two years old and when I think about the fact that it’s becoming a topic of conversation, the question is how do you get there? And I am a big believer in organic transformation when you change people, you change the entire organization. So you have to have a methodology that helps bring your population along. And so I wrote the book out of frustration because I didn’t see what I felt is an important success factor for the adoption of advanced analytics.
I didn’t see any primer for how to do it. I just saw lots of examples of data science solve this one problem but that’s not what the winners will be doing. The winners in this race will be transforming so that their organizations are truly data-driven in a way that we’ve been talking about being jaded driven organizations since the mid nineties. It’s becoming urgent and important because if you miss this boat, you won’t exist as a company.
I’m so much looking forward to this month, checking out your new book, dropping on shelves near you. Xena Ugrinsky Enterprise AI- Your Field Guide to the New Business Normal. Appreciate you being here on the humAIn podcast And it sounds like whether you’re an executive in operations, data science, sales, AI. AI is coming to eat the world. I like to still call it the new spreadsheet. I think time will tell to see if that happens but it’s a good time to start learning how AI can change your business seeing that. Thanks so much for being with us.
It’s been a pleasure. Thank you David.
Hey humans. Thanks for listening to this episode of HumAIn. My name is David Yakobovitch and if you like HumAIn, remember to click subscribe on Apple podcasts, Spotify or Luminary. Thanks for tuning in and joining us for our next episode. New releases are every Tuesday.