Navigating Enterprise Transformation: Design and Data for Competitive Advantage

Ed Lovely, Seth Johnson at UXDX USA. Video: https://youtu.be/FkZCq4jDGvg

Readable transcript: edited from the recording's captions for readability (fillers and false starts removed, punctuation and section headings added). Wording is the speaker's own. Timestamps are positions in the video. Names marked [?] could not be verified against the audio.

IBM's reinvention and being client zero

[00:00:10] Seth: Hello everybody. Thank you to Joe, and thank you all for being here. IBM has been around for a long time, more than a hundred years. And in that time, there's been one consistent thread: reinvention. We have constantly transformed our offerings and how we work to bring those offerings to market, and today IBM is the enterprise leader in hybrid cloud and AI that makes the world work better.

[00:00:49] Now, even if you just look at our most recent history, this transformation is pretty obvious. We used to be primarily an IT hardware company. Just seven years ago, less than half of our revenue was software-centric. Now it's more than 80%. So we've dramatically changed our business model, but we've also fundamentally changed the way we work to support this new high-growth focus.

[00:01:26] A big part of this strategy is what we call being client zero. Before we go sell something to our first client, we have already internally tested, refined, and used it ourselves. They're client one; we're client zero. Eating our own cooking improves our offerings, but it also increases our revenue, which drives our productivity, which leads to reinvestment. And now you've entered into a virtuous cycle of growth.

Siloed data and the integrated enterprise data model

[00:01:56] Now, the operational foundation supporting all of this relies on our strategic use of enterprise data. The chief data office, which is where we work, essentially provides data as an asset to IBM for competitive advantage: things like governing, integrating, securing, and democratizing our data to provide trusted and timely business insights across our entire company, and making data ready for AI.

[00:02:32] We began our data journey at a place that's familiar to many enterprises today. You might recognize this. It's just accepted that it's going to be hard to work with data. You've got to know where to find it. You've got to jump through hoops to get it. You manipulate it in spreadsheets. You put screenshots into slide decks.

[00:02:57] Why? Well, in most organizations, data is siloed. It's useful only to a subset of people for isolated analysis, which is problematic, because end-to-end business workflows don't happen in silos. They happen across boundaries. So when data can't flow between those boundaries, it slows things down, requires manual workarounds and gives you limited opportunity for insights, like when the supply chain isn't ready for an uptick in production because it's not actually connected to the sales data. This also severely restricts the ability of AI use cases to provide any return on investment, since they're essentially just point solutions working within one silo, maybe two.

[00:03:51] What we've done at IBM over about the past seven years is develop one integrated enterprise data architecture and a front-end experience for users, my fellow IBMers. We call this EPM: self-served governance, dashboards and AI insights. Now, when data from across end-to-end workflows comes together in an integrated data model, it becomes exponentially more useful to the business. For example, engaging a client might start with a marketing event, progress to a sales opportunity, on to billing and payment, and probably some customer support.

[00:04:37] Connecting data across these workflows, putting it all in one place where anyone in the company can easily get it: that is a game changer. It allows everyone to see insights that had been unseeable, particularly when they're augmented by AI. Competitive advantage.

[00:05:01] I'll end by highlighting just a few of our outcomes. 75% of IBM's enterprise workflows are now optimized with these integrated insights. And we're deploying AI use cases in weeks, not months. There's massive simplification in our reporting landscape, sunsetting more than 20 legacy platforms, with even more to come. And just one example of how all of this comes together: our C-suite preps and runs their operating reviews with live dashboards. No spreadsheets or slide decks in sight.

[00:05:39] Hopefully that gives you some context for our discussion: a single source of truth for the company that's integrated across the enterprise, governed holistically and enabled with new insights for competitive advantage. So with that foundation, let's talk about it. I'd like to welcome to join me on the stage my boss and IBM's chief data officer, Ed Lovely.

Why a chief data office has a design team

[00:06:07] Ed: Welcome. Thank you. Great to be here.

[00:06:10] Seth: Thank you, Ed. Listen, folks, we're going to be getting some questions coming in from all of you, our fellow conference attendees. And depending on where our conversation takes us, we may flash a slide or two to help illustrate a point that we might come to or explain things. But I'm just going to put that chart up there real briefly in the background as we get started, just to give folks an overview of our journey.

[00:06:38] I'm going to start with maybe a self-serving question, Ed. I'm a designer. You're a business executive. IBM's chief data office has a dedicated design team, which might seem unconventional at first glance for such a deeply technical and operational organization. Why do we have a design organization?

[00:07:02] Ed: It's a good question, because most of my peers don't really engage design too much in their enterprise data strategy. It's just not obvious enough to them. The reason we decided to create a design team as part of our enterprise data was really because I grew up in finance and accounting jobs, operational jobs, and working with data, as you mentioned earlier, Seth, was just an arduous task. Getting at data and manipulating it in spreadsheets was just considered part of the job and an expectation. And for me, after living through that for many years, it was unacceptable. We need to make engagement with data simpler and easier, so that users want to work with data as opposed to dreading the moment they have to play with their big spreadsheets.

[00:07:51] That was the idea we had, because we set a BHAG at IBM that we're going to have a single source of operational truth to run the company, which is what EPM has become. A few years ago, we didn't have anything like that. Data was everywhere. So to create this integrated data model, we wanted it to be a delightful experience for IBMers. And we chose to pick, I'll say, the persona that was the hardest one to deal with: pretend the user is a brand new IBMer who's never been to IBM before. How do we make it simple for them to engage with business data and use it immediately to generate insight for the company? And for that you need designers.

[00:08:30] Seth: Yes indeed. Just give us some context here. You said BHAG goal: a single source of operational truth for the entire company. Why is that such a big goal? And I know you talk with our clients a lot, so you have some external perspective on this.

[00:08:44] Ed: I'm very opinionated on this topic, and I talk with clients all the time. Literally dozens of clients. Last year, I met with over 100 clients. This year, I'm up to 75. And most enterprises, believe it or not, have not prioritized their data. If you think about how you run a company, one of the most foundational elements to run any company is data. It's the lifeblood of a company. It's more important today than it ever was because of AI. And what's happening now that we're in the AI era is that companies that haven't spent time on their data are really struggling to spin up AI POCs, proofs of concept.

[00:09:22] The reason is all AI starts with clean data. If you have to spend your time finding, cleansing and aligning your data, you've just set yourself back several months. Most companies are finding that today. At IBM, we're blessed because we knew it was coming. We integrated our data. We created enterprise data standards. Five years ago, we had zero. Today, we have 565. We have a single source of truth. We have 30,000 IBMers that use it to get their jobs done. And it's paying off now as we scale AI.

[00:09:55] Seth: I love it. We'll get to your questions in just a second. I did, however, want to just flash this chart, because I do want to give a shout-out to our amazing design team, our small but mighty design team. We're here at a product design conference, so I thought it might be good context to see what a design team in a chief data office looks like. And here's one of them, and several of these folks are here with us today. Thanks for allowing us to give our team a little shout-out. Booth in the back. We're ready to look at some questions whenever you're ready to show them to us.

[00:10:32] Ed: Hey Seth, I have a fun fact. Do you know who the first design leader was for EPM at IBM?

[00:10:37] Seth: I do. But why don't you share with our fellow conference attendees?

[00:10:42] Ed: Joe Mirsman [?] was our original design leader. I thought that was pretty wild.

[00:10:47] Seth: Yeah. Thanks, Joe.

Keeping the data correct and democratized

[00:10:52] Seth: Okay, we've got great questions coming in. This is interesting: Ed runs his occasional chief data office all-hands using a similar question format. Very transparent leadership. What does EPM stand for?

[00:11:07] Ed: Enterprise performance management. Not to be confused with Oracle's EPM. It's just an internal branding that we decided on, because it describes with clarity what it is.

[00:11:14] Seth: Right. And another way of thinking of it, perhaps, is that a big piece of the EPM platform is the front-end user experience that IBMers are thinking about. Think about this as dashboarding and reporting.

[00:11:27] Fantastic question here. How do you know the data you're fetching is the correct data and that it's not misinterpreted across an enterprise at some scale? At IBM we're talking about 300,000-plus people in 170 countries. I would imagine that across the enterprise, people are using different definitions for different business questions. It's a great question. How do you deal with that?

[00:11:51] Ed: Yeah, it's a great question, because we have so much data. We have 300 terabytes of integrated data that we're refreshing constantly, 24/7. To keep the data straight, we have a data catalog. If people are confused as to what a data element means, they can go right to our data catalog and learn everything about that data element: the fact, the dimensions around it, the metadata around it. So we try to make it as simple as we can, to avoid people getting confused about what data they're using.

[00:12:16] Seth: Earlier we heard a case study from Netflix about how to find exactly what it is you are looking for, and once you are satisfied with that, how to perhaps expand your interest or your awareness of other things that are available. And our data catalog is designed in a similar way. People come to the data catalog having an idea of, okay, I need to find the data or the data source that will connect me with this. And what we would like, and what we're finding, is that while they're there, because it's an intentionally designed experience, they can see, oh my gosh, look, I didn't know that all this other stuff was in here and it's available to me. Oh, okay. Oh, look, I can have that. I can get that.

[00:12:57] Ed: You used the phrase earlier, Seth, democratizing data. Again, think about data and how data is used. Data is used to operate a company and to make better, faster decisions for your company. The more you can democratize your data while securing it appropriately, the faster your company's going to move and the better decisions your company's going to make. So to me, democratizing that data, making it simpler, using that experience to really engage with the data, is what adds value for your firm.

How the initiative started

[00:13:29] Seth: Sure. Kate's got a question to maybe get us on a thread here. Ed, bring us back in time a little bit. How was this initiative to centralize data spearheaded? How was everyone coming together? We were talking earlier about how difficult organizational change is. One part of the business might be all in, but they have to bring everyone else along. I know we faced some of that. How did all this get started?

[00:13:55] Ed: We did. Well, the hero of our story is IBM's chief financial officer, Jim Kavanaugh. Jim had the vision many years ago that those companies that invest in their data, data standards, master data management, data privacy, data governance, will have a distinct competitive advantage as we get into the AI era. And he was exactly right. That was our catalyst.

[00:14:18] Because pre-EPM at IBM, if you were a knowledge worker, and most IBMers are knowledge workers, meaning they work with data to get their work done, you would have to, as Seth described earlier, know where to go, when to pull it, at what rate it refreshes, and validate that if you combine it with other data it has referential integrity. Otherwise you could be really screwing up your data, comparing apples to oranges: in this system it's this way, in this other system it's this way, I need to put them together. And this literally happened all the time. It happens today in other enterprises.

[00:14:51] You'd be in a meeting and you'd have two people showing numbers, and they're a little bit different. We used to have this at the highest levels of our company. Our CFO would show a resource number, and then our CHRO, our head of HR, would have a different resource number. Now, you can imagine how quickly that meeting would then disintegrate, where you don't trust the data anymore. You question every data point.

[00:15:14] The other thing is, when you integrate your data and you make it drillable and you load it at the lowest atomic level, you can view it at any level. If you're a senior executive, you're looking at the very high consolidated numbers, or if you're an analyst, you're getting way down into detail. If it's a single integrated data model, it doesn't matter where you come in at the data, because it's all synchronized. We used to have the problem in IBM that you'd go into a meeting with literally dozens and dozens of PowerPoints or slides, and someone would ask a question that's not in the PowerPoints. So again, you have a bad meeting. You take it offline, you put it in a parking lot or something, and you waste time. You schedule another meeting for the comeback.

[00:15:53] Today in IBM, we work, as we call it, on the glass in EPM.

[00:15:59] Seth: The glass meaning your pane of glass that you work on, your laptop or something.

[00:16:02] Ed: Exactly. We can go right down to the invoice level, the client level, the product level, for any question that comes up, which again makes the speed of decision-making and insight much faster.

[00:16:15] Seth: Yeah, speed. And there's another word that you brought up, which I think can't be understated, which is this notion that it also helps breed trust. You used the example of, oh gosh, the head of HR has one number and the head of finance has another. Those numbers are different from each other. Of course, that screeches a meeting to a halt, but what it also does at scale is breed mistrust. I don't know which number is to be trusted, and that's no way to run an organization, fast or otherwise.

Design at IBM and user-centric thinking

[00:16:47] Seth: Oh, I love this question. From Anonymous: did IBM always value design or put a user-centric approach first? Did you have to advocate for this? How did you demonstrate its value? I'll maybe talk a little bit about this at the company at large, and then see if you have some thoughts from your perspective.

[00:17:07] IBM is 114 years old. For much of the 20th century, IBM was a storied, design-centric company. Paul Rand did our logo. Charles and Ray Eames were hired to do all of our instructional films. Design, for a century or more, has really been woven into the ethos of our company. IBM, like many organizations near the end of the century, lost its way a little bit with design.

[00:17:42] And just about 12 years ago, when our former CEO Ginni Rometty became CEO, she made a proclamation, essentially on day one of her being CEO. Her proclamation was: IBM will win because of the experiences that we offer our clients. And in order for that to happen, she had the wherewithal to say we need to bring design, the practice of design and designers, back to IBM.

[00:18:13] So over the course of the past 12 or 13 years, IBM has massively scaled its design program. We have one of the world's largest design workforces. There are upwards of 3,600 formally trained designers that work at IBM, in every part of the company. And virtually every IBMer, regardless of discipline, has essentially been trained in design thinking. I'm always so pleasantly aghast [?] when I drop into meetings with engineers or developers and they're just naturally and organically talking about users and the user experience. Oh gosh, well, we could do that faster, except that is not going to get us a user. These are engineers. And I think you feel the same way, which is, as you said earlier, why you felt it was so important to bring design and a design approach to this work.

[00:19:13] Ed: I do. And I give kudos to Phil Gilbert. Many of you may know Phil Gilbert. I was a Phil Gilbert disciple on user-centric design. Before my jobs in data, I was working in the CIO's organization, running and building our applications. And that's where it really struck me how the user experience has to be considered at the conception of your data or your application. For us, it was just a natural thing. It was important. I think for me personally, again, growing up in IBM many years ago in accounting and finance jobs, those jobs were so difficult because the user experience just wasn't something that our application design teams put a lot of thought into. And we knew there had to be a better way.

[00:19:57] Seth: You've talked to me about this before, about the late hours, deep into the dark of the morning, just grinding through spreadsheets and spreadsheets. And I would imagine that part of the thinking there was that the experience here doesn't matter. These people have no other choice, they have no other option, it's part of the job. And one of the things that I love about you being such a design sponsor is you essentially reject that notion. No, that's ridiculous.

[00:20:28] Ed: Yeah, because I lived through it, and I have scars.

[00:20:31] Seth: You have empathy for it and you have scars.

[00:20:32] Ed: I do. Deep empathy and deep, deep scars.

Starting with finance

[00:20:35] Seth: I want to get to the question about integrating AI into the design process in just a moment. But to close out this thread, let me hit the second question here. Ed, shed some light on how we went about identifying how the data should be organized, captured and maintained. You talked about the fact that we did say that it should be. How did we do that?

[00:20:56] Ed: Yeah, that's a daunting task for anyone who's starting to build an enterprise data model, because where do you start? Data is everywhere. At IBM, you've got to start somewhere, and we started in finance. Because our point of view was, if your integrated data model is going to be all about business data, data used to run your company, finance is central to that. And with finance data, you have to get it right all the time. Every cell of data has to be right, because we pay taxes off it, we report externally, and we don't want to go to jail.

[00:21:26] So we figured, if we can demonstrate to the finance team that we can build a trusted integrated data model with a great user experience, then we'll be off and running. So we picked the toughest one to start with, and thankfully we got it right. It took us, quite frankly, about 12 to 15 months to get to value for finance, where we hit the tipping point of workflows integrated within EPM. But from that moment on, it's been an avalanche of can you add this, can you add that? So we knew we were on to something when we got finance right.

Encouraging designers to use AI without fearing job loss

[00:21:58] Seth: Yeah. Okay, let's maybe shift gears here a little bit. I love this question that's come in: how can I encourage designers to integrate AI into their design process without fearing job loss? Any successful approaches you recommend? I'd like to share a point of view on this, but I'm also curious to hear from you as well, and maybe even broaden it to non-designers.

[00:22:21] I think something we as designers in particular need to make peace with is the fact that if we are working in an enterprise, particularly if we're working in a large organization, if we're working in a business, I've got news for you. The business doesn't care about design. And frankly, the business doesn't care about designers. The business isn't supposed to care about design and designers. The business cares about market outcomes. So the business cares about design, designers and design thinking only to the extent that those things can help deliver the market outcomes that it needs.

[00:23:09] Part of my advice here for design leaders and designers, and I'm even talking to myself here, is that we have to make peace with the fact that we are working in a business, and that design cannot be the most precious thing to me. What has to be the most important thing to me is user outcomes, which will lead to market outcomes. And if I start prioritizing how I approach my work and my practice that way, that design is not the thing, design is simply the way that I contribute, it frees me to think, okay, well then, design's going to change. It's just fact.

[00:23:57] And part of that change right now is, oh, there are new opportunities. There are new tools. There are new things that I can use. Oh my gosh, this might help me go faster. This might help me automate some of my work. I don't have to focus on doing that anymore; I can do this. Part of that, of course, is fear. Part of that, of course, is change. Some of that is replacement. I don't have to do this anymore. I have to do that. I need less of this. I need more of that.

[00:24:27] But that's no different than what happened 10 years ago when we were in the throes of the mobile revolution, or 25 years ago when we were in the throes of the web revolution. This story has been told dozens and dozens of times before. So my advice is to not fear the change and be protective and avoid it. My advice is to accept the change as a natural course of the way things work. And as we learned earlier from Jimmy at AirPable [?], play with things. Try things. The more you try it, the more comfortable you'll get with it.

[00:25:00] Ed: Yeah, I think it's well said. I think embracing this change is just so important, and it's in every job. It's not limited to designers. We get a lot of questions from our software engineers, because we have code assistants now that can write code and document your code and scan code. They're worried about the same thing: will I lose my job and be replaced by AI? The answer is no. We hope that you'll use AI to enhance your abilities and your output so that you can focus on the next thing, and the next thing, and the next thing. And that's really what AI is all about. It's an augmentation to the human skills that we have today.

[00:25:36] Seth: What's that phrase? This is not mine; I'd appreciate the attribution if someone has it. Your job won't be replaced by AI. Your job will be replaced by someone who's using AI. Maybe that's a way to look at it.

Running the C-suite meeting on the glass

[00:25:52] Seth: Let's see. Oh, great question. We mentioned earlier that our C-suite reviews are run live on this platform, using live dashboards. And correct me, right, no PowerPoints allowed?

[00:26:00] Ed: Not only are they not used, they're not allowed. I'll tell you a little funny story. At IBM, this meeting, we call it the operating team meeting, the OT meeting, happens every two weeks, for two hours, in our corporate headquarters. It's hosted by our CEO and our CFO and about six of our SVPs. Our SVPs are essentially IBM's C-suite, the senior-most people, and no one else is allowed in the room. And preparing for that meeting every two weeks cost the company hours of preparation. You can imagine, with the PowerPoints that all those leaders created, their teams would have to work and churn all night long for two weeks to create 1,100 hours of work. And then they'd go into the meeting, and someone would come up with either numbers that don't match or a question that wasn't in all the PowerPoints.

[00:26:50] So for us, going to the glass, going to a digital rendering of our dashboards, we were serious about it. When I reached out to our OT meeting to say, hey, can we digitize the OT meeting, our CFO said, absolutely, we're going to walk the talk. And we said, this means your helper can't come with you. We even had people who were going to hide behind a bush in the room and just type on the keyboard. We said no, no one's allowed in the room but those six or eight people.

[00:27:20] So we had to train up our senior-most leaders in IBM to use our platform. Now, that's a real indication of whether you have built an experience that's simple. And this is not to criticize the technical skills of our senior-most leaders, but navigating spreadsheets and deep diving and pivoting was not something that they were necessarily comfortable with. But we taught them each how to do it. And that meeting, the operating team meeting, has been run on the glass for three and a half years now. They have no helpers. They run it on their own devices using their own dashboards.

[00:27:55] Seth: Yeah. And it was interesting from a design point of view, because our design team was involved in this. We were interested a little less in designing the dashboards, though that was certainly a thing, literally what the interface looks like. What we were really interested in, and did a bunch of discovery and work around, was what the meeting experience is like and what it should be like. Who's in the room? How are they going to prep? Who determines the order that they're going to go in? So that was an interesting leap into an interesting area of design. We referred to it as meeting design.

Q&A

[00:28:33] Seth: All right, we have time maybe for one more question. Let's see, maybe we'll do rapid fire. Ed, how do you ensure your data is safe while also using AI?

[00:28:42] Ed: We have a governance team that manages governance through a product we call IBM Knowledge Catalog, which essentially allows us to democratize governance, so application owners can scan their data and make sure that it's governed properly with our common tools.

[00:28:58] Seth: Excellent. All right, we're going to end on one last question. Deliver value within 18 months: what tests and theories were engaged to have confidence in that number? We heard a lot about tests earlier today.

[00:29:12] Ed: Yeah. The test was utilization. We tracked monthly average active users, and once we saw that start to spike, we knew we were on to something.

[00:29:22] Seth: Monthly active users. I love this. We're talking about consumption metrics, not sales metrics.

[00:29:27] Ed: Yes.

[00:29:28] Seth: Yeah. Ed, thank you so much for being open to this product design community, and thanks for being open to having us join you on stage. We hope to continue engaging with you throughout the conference. Thank you so much.

[00:29:43] Ed: Thanks so much.