CDO to CDO Podcast: Ken Dunn

March 23, 2021 Chris Knerr

The CDO to CDO podcast is hosted by Chris Knerr, Chief Digital Officer of Syniti.

 

 

+++++++++++++++++++++++++

Subscribe Here:  Listen to all the episodes in the Syniti Podcast Series

+++++++++++++++++++++++++

 

 

Chris Knerr (00:23):

Hello and welcome to the CDO magazine interview series. I'm Chris Knerr, Chief Digital Officer of Syniti, a world leader in enterprise data software. And we're partnering with CDO magazine, MIT CDOIQ and the International Society of Chief Data Officers to bring you this series of interviews with thought leaders in data and analytics. Today, I have the tremendous pleasure of speaking with Ken Dunn, formerly a principal data manager who was directly supporting a number of Chief Data Officers at BP, now managing director of Data Vital, a CDO, MDM, data integration, and data architecture boutique consulting firm. Ken's got a strong background in information technology, spanning over the course of 30 years, broad expertise in data architecture, information management, big data, advanced analytics and information security. As noted, he was with BP for many years, working across upstream and downstream businesses. Ken's also active in the industry standards movement and was chair of the international oil and gas producers digitalization and information standards subcommittee. And president of DAMA Houston for four years. Welcome Ken, terrific to have you on the podcast today, and I'm really excited for our conversation.

 

Ken Dunn (01:44):

Hi Chris. Yes. Really looking forward to the discussion.

 

Chris Knerr (01:47):

Super. So maybe let's dive right in. Can you give us a sense of when you were... Those many years at BP, what was the scope of your organization and what was your brief at BP?

 

Ken Dunn (02:02):

Right. So BP is an enormous company. A global energy company and in recent years, that expanded beyond just oil and gas. So you need to start thinking of these large multinationals that are focusing on production of energy. And that's means that they have gone through an enormous shift in the way they are thinking about their business. And as a data professional supporting that business, we've gone through a big shift as well in terms of making sure we can still support the evolving business. I started off in data when I was back in the days of data warehousing and have followed it through the various iterations of data warehousing and data lakes and data virtualization and, and looking at all the different aspects of data.

 

Chris Knerr (03:07):

Perfect. That's very helpful context. Thank you. And I know we've all heard the tagline 'BP beyond petroleum', so, to your point, a lot of transformation in terms of the thinking and the vision and the strategy, which is very intriguing. So on top of that transformation that BP was going through anyway, and on top of the fact that energy is a cyclical sector, very capital intensive, enormous companies, to your point, very, very mature, strategic business and strategic capital planning processes compared to a lot of other industries. How did the strap planning process and roadmaps survive the turmoil of 2020? Because it's very interesting from an industry perspective that the high level of maturity, and then sort of these unprecedented black swan events that we don't need to go into in great detail, but were you prepared to catch the pitch that was thrown if you will?

 

Ken Dunn (04:19):

Well, as prepare as any organization. I think the oil industry had three things converge in 2020. There was COVID that everyone had, there was the pressure on the price of oil and then there was the low carbon future. And the three things came together in 2020 that really accelerated the plans for change and really made the whole company think about where do you invest? How do you invest? How do you exploit your data? How do you get more value out of that data? And so I think it was more an acceleration of the plans that were in place and a relentless focus on profitability as well. I mean, I think all companies faced that dilemma, but even more so in the oil and gas industry.

 

Chris Knerr (05:33):

Yeah, no, that makes sense absolutely. And it's an interesting perspective on kind of the triple threat that you've faced and obviously the price shock and the demand shock was very significant and that caused many different interest strays and we saw this in our own client base to sort of retrench and have a core focus on profitability. And in some cases, unfortunately even on remaining solvent and illiquid to fund ongoing operations. So just sticking with that theme of the strap planning process, were there... If you think about the connection point of strap planning, capital planning, how data is used in that process, were there some key learnings? Some things that even with the perfect storm and being probably as well prepared as you could have been, as well as any industry could be, were there some learnings from a strap planning process related to how to best utilize data that were significant?

 

Ken Dunn (06:40):

Yeah. So bringing it back, right back to my area of expertise, and that is data, and the importance of the role of the chief data officer in ensuring that data is on the minds of the executives of the company, when they're doing their strategic planning. BP was such a big company that they decided to break the business up. And they had four separate CDOs within the company looking at different aspects of the business. And that made a lot of sense because the impact of all the change in 2020 was quite different on our production business and, effectively, the manufacturing side than it was on the customer side. So if you look at those two separately, on the production side, it was all about efficiency and moving into new areas and then on the customer side, it was looking at the retail sites and how you evolve a retail site into the new world.

 

So you don't want to just offer the old gas, you also want to offer electric charging. And so the retail site is evolving. And if you think about what's coming in the mobility industry, autonomous vehicles are coming, and then you want your retail site to also move into servicing and maintenance of the autonomous vehicles. So those two businesses are completely different. They have separate CDOs that are integrally involved in the strap planning for the company and bring to bear the thinking about how do you exploit your data assets in this rapidly evolving world.

 

Chris Knerr (08:59):

So let me just replay that and then I have a question that builds on that, and I think that's really fascinating. So your point, the needs are very different according to the business area. So, whereas on the retail side, your commercial organization might be kind of leading the charge and thinking about, "Look, the market's changing. I need to figure out how to bring my data assets to bear in order to drive new commercial value, new offerings, and respond to a change in the market." On the upstream or the supplied side of the business, the line of business executives, even commercially, they're going to be much more focused on cost improvement projects or CIPS on economies of scale, on how to leverage, for example, streaming data, everything you have in that asset base to really keep everything running. And so that that can be maintained efficiently and cost-effectively, so it's almost like diametrically opposed thought processes, depending on which area of the business that you're thinking about.

 

Ken Dunn (10:08):

But the common thread is to identify what data is critical to your business and how do you point that data. [crosstalk 00:10:19]

 

Chris Knerr (10:20):

And because I do a lot of this work myself, and I find that companies often struggle with us. So when you identify the data, are you doing that within kind of the construct of one of those large business areas or lines of business on kind of a defined value drivers scheme or value driver analysis. So that you'll start with something like... Pick something germane to oil and gas, like fixed asset utilization, right? So you'll say, "Look, my macro is fixed asset utilization, and then beneath that, I've got N number of value drivers, and then those value drivers are physically made up of data, which feeds those. So that if my data's poorly governed, my data's a mess, I can't properly feed those metrics and I can't understand how my subsidiary value drivers are performing. So I can't get to my fixed asset utilization." Is that [crosstalk 00:11:16].

 

Ken Dunn (11:16):

Absolutely, that's the thinking. And so, yes, you know from the business what those key metrics and key drivers are, and then you look at ways that you can utilize data to improve them. And on the fixed asset side, the availability of our facilities is one of the things that everyone in that part of the business focuses on, including the people that are the CDO and the people that are supporting the CDO. So how can you use the streaming data coming off the equipment to do preventative maintenance, improve the availability of that key metric and make a direct contribution to it?

 

Chris Knerr (12:11):

And have you found the... One thing that's been thematic in a lot of the CDO focused interviews that I've had is educating business leaders and let's say non data technology leaders on the connection of the value of clean data to the KPIs, to the business outcomes, what's your experience been with that? Easy, hard, various?

 

Ken Dunn (12:43):

Critical.

 

Chris Knerr (12:46):

Okay. So no different.

 

Ken Dunn (12:49):

You've got to make sure that your CDO is a full partner at the executive table and that the CDO doesn't just do what the other executives tell them to do. He has to bring new ideas. And let me give you an example. And that a lot of the executives may not have thought about, and that is we have a lot of, or BP has a lot of ROV videos. So videos that had taken from the submersible submarines that are going around looking at the state of the facilities beneath the surface. And, hey, we need to have that video for compliance reasons, but how can you exploit that video? And so the CDO brings to the table saying with AI, now you can do image recognition.

 

So how about we look to see if we can speed up the maintenance process and improve the maintenance process by doing automatic detection of corrosion from this sub sea video. And that's proving to be a very valuable tool for the business to actually utilize this data that we've already got in a new way that people hadn't really thought about before. And so that's something that gets brought right to the executive table and changes the business.

 

Chris Knerr (14:32):

So that's fascinating and let me drill into that a little bit more. One thing that I noticed in preparing it in your bio, you've done a lot of work around data lakes. Around the architecture, the design of data lakes. So I think, partly, you gave me a hand to what the answer, but I'd love to hear more about this. So when you think about a data lake strategy kind of scope approach, first maybe just a simple question, what kind of data goes in the data lake? Is it the cool data that you just talked about? Those video data objects, or is it just transactional data or is it some combination? And what else? Because I think this is something, although that the data lake conversation has matured a lot in the last five years, my sense is a lot of companies are really still struggling with architecture's scope approach. And then we'll come back to the value, which is where we started this thread of the conversation.

 

Ken Dunn (15:39):

The world has evolved beyond a physical construct called a data lake. And it's about time too. I mean, the amount of data that any business has these days is so enormous you're never going to put it in one physical store. What you need to be able to do is to know the data that you've got and know how to access it and how to integrate it. So no, we don't move all the video data out of the video store that had seen into the data lake. What we do is we make sure that it's appropriately tagged so that you can get to it. And appropriately stored in the video store in such a way that you can do AI analysis of it. And I think that this new concept of a data fabric, a virtual data lake, I think that has a lot of potential to really bring in all these new types of data, or make new types of data available for analysis. And that is a critical part of this whole driving value out of your data assets.

 

Chris Knerr (17:05):

I'm with you [crosstalk 00:17:07].

 

Ken Dunn (17:08):

On the fundamentals, you obviously still have to get the fundamentals right. But do it in such a way that you can incrementally expand it into different types of data.

 

Chris Knerr (17:19):

Yeah, no, that completely makes sense. And I agree for the record. I mean the emerging best practices, they're data lakes, not data oceans where you just dump everything. I think what, I sometimes jokingly call the Soviet style of data lake, or master data management, where you bring everything together and manage it centrally is not... That hasn't worked very well in practice. And what I'm seeing is kind of more of a focus to your point on enlightened decentralization so that you have something more like an agent based architecture, you have the appropriate tagging and so on, so that you're not physically replicating data unless you need to. And certainly the technical [I POS 00:18:10] tools and so on, have greatly enabled that. What I'm curious about, at least from what I've seen, that doesn't necessarily solve, and this is your comment on fundamentals, it doesn't necessarily solve the basic data integration or data interoperability problem, right?

 

So just, if you're able to maybe stick with this example, right? I have a video and the video's an undersea video, and it's a video of an asset, right? And the asset is serialized, the parts within the asset are serialized. They have maintenance records and maintenance history. So to do my AI model, right, part of it is it's actually a very simple sort of image recognition, like a radiology type analysis. The boring part of it in a way [inaudible 00:19:04] is figuring out which valve is that? What's the maintenance history of that valve? And gluing all that together. So the thing that, I personally feel, sometimes gets lost in this data fabric discussion is okay, the nuts and bolts of what you called the fundamentals, master data management, do I have good maintenance records? And can I glue all that together somehow? What are some of the best practices that you've seen in dealing with those fundamentals, I guess?

 

Ken Dunn (19:37):

Yeah. So again, there's a lot to that question and as CDOs and architects, we've got to be able to do all those things at once, and you can't just focus on one area, right? You do need to focus on the storage of the videos and the streaming data, but you also have to focus on the tagging knowledge. You've also got to focus on the master data behind it, and master data management is an ongoing area of focus that I don't think anyone's got right yet. And we will never get it right. You just have to incrementally improve the way you do master data. And so equipment master data, and associated tag master data is something that there's a whole program of work around that, but you need to make sure that that is in support of... You understand the use cases in terms of how that master data is going to be used. [crosstalk 00:21:00] focused on the basics, but not exclusively.

 

Chris Knerr (21:08):

In your experience doing this, because I agree with you. I mean, I think master data management or what I've come to start calling data operations, which is kind of the merging of all this together, it's really hard. There aren't very many companies in the world who do it really, really well. If you were to describe whether you think, in the kind of classic triad, is it more people and organization, is it more process or is it more technology, or is it a combination of all three where you've seen it work poorly and where you've seen it work well?

 

Ken Dunn (21:51):

Right. And it's absolutely a combination of all three. The people, process and technology, and you've got to get all three right for it to be successful. And I think that's why we still see so many master data management projects failing. Because IT cannot implement a master data management solution without strong business support. Because you do have to change the business process, you do have to educate your data stewards and the folks in the business about why the business process is changing and the importance of the data. So it's very much.... I, in fact, like to start with looking at the business process, because I have seen so many MDM projects fail because they haven't looked at the business process. Again, I'll give you a classic example from a different domain. And that was in our vendor master data. And someone had this brilliant idea that will improve the quality about vendor master data by managing it centrally. Guess what? The central guys had no clue what the right value was.

 

So they had to go back to the regional folks. BP operates in about 60 different countries. And the central team didn't know all the rules, they didn't know all the local businesses and so they ended up, believe it or not, communicating via email. So, sending an email saying what should be in this field and then a week later getting an answer. And so in fact, the quality went down by centralizing. So the first thing you have to do is say, "Who are the subject matter experts? Where is the expertise? How can we design a business process that matches that expertise?" Once you've got that, then you can say, "Well, what technology do we need to support this business process?"

 

Chris Knerr (24:05):

Yeah, I'm a hundred percent with you and maybe add to that. By the way, I've read the book on master data management programs that have not worked myself multiple times, unfortunately. I've also read good versions of the book where it's worked very well. I think that the business process design, but also the organization design and what I would call them, sort of the executive level governance design. So given the characteristics of the organization and the data, do you want to do centralized? Do you want to do decentralized? You want to do federated? And there are four or five different kind of standard organizational models, but I've found that all too often, people do want to jump right into, "What's my master data management app?"

 

And they want to ignore the difficult alignment work of figuring out, "Look at the end of the day, you want to have all your vendor master data set up so that you have accountability for it." And this is back to the business value conversation, there's some financial or customer innovation or quality value that's being adduced to having that data better, faster, more efficiently. That it's not just for the sake of data quality, which is, I think, another mistake that people make in doing these programs is to try to do it just on evangelical data quality as opposed to data quality to do X or Y business outcome.

 

Ken Dunn (25:31):

Yep. And I think we sometimes, as professionals in the field, get hung up on words like data governance and master data management, but you have to keep going to back to, "Well, what is the fundamental thing that you want?" And that is what you need to be successful is you've got to have someone in the business that is prepared to take accountability for the decisions. So whose decision was it that you should centralize the master data? And I must admit everyone ducked for cover when we started asking that question. So that's why it's so important upfront to get your governance right, to say, "Who is going to be accountable for the decisions that this program takes?"

 

Chris Knerr (26:28):

Yeah. Sometimes people want to centralize it until they find out that they have to do all the work and that the work is very complicated and that they find that they've got buyer's remorse on centralization programs. I think that that thought process is very important. Well, let me wrap up, if I can, with one question. So something that really intrigued me when I was preparing for our discussion, and then in talking to you is you've got a deep technical background, and the majority of your career was spent in technical disciplines, but yet you've got this clarity that I love on connecting technology strategy to business value. That in my view is often, it's not terribly common among senior folks who've gone down a very deep technical track, so what was the magic that happened at some point in your career that was sort of the spark of, "No, it isn't just bits and bytes. It's bits and bytes that add up to customer value or financial value."

 

Ken Dunn (27:39):

I think it was a couple of the projects that we did that were technically successful. And then the business said, "Well, so what?" Because we hadn't sold the value of the work that we were doing. And again, let me give you another example from a few years ago now that is the opposite of that. And that is we were doing a data quality program and we were looking at the quality of the pipeline data. And it wasn't as if we went out with a specific objective, and with hindsight we should have, but what we found was the poor quality pipeline data was leading to paying royalties for the wrong state.

 

And if we hadn't found it and the regulator had found it, there was multimillion dollar fines involved. So, if we had started at the other end in saying, "We really want to make sure that we have high quality, accurate royalty data so that we know that we're paying the right royalties, then we would have been really focused on that problem and actually solve that business problem rather than just, hey, improving the quality of the pipeline data."

 

And I think that to me was a... We were heroes because we had saved the company a lot of money. And when I looked at it, technically, we weren't doing anything different. We sort of got lucky. And then it made me realize that by focusing on the value, then you're going to be seen as the hero rather than someone implementing yet another bureaucracy.

 

Chris Knerr (29:57):

So, that's brilliant. I completely agree. And, it's sort of what they say about congressional investigations, right? Follow the money.

 

Ken Dunn (30:09):

Yup. Follow the money.

 

Chris Knerr (30:10):

It's actually that much [crosstalk 00:30:12].

 

Ken Dunn (30:16):

The risk of not being in compliance these days is also really critical.

 

Chris Knerr (30:22):

Yeah. Yeah. And there's [inaudible 00:30:23] For a long time, and I was formerly at a different Fortune 10 company, everything largely comes down to cost growth or quality and compliance, right? So I always have in my mind when you're starting in an initiative, you're asking for significant investment dollars from businesses that have a lot of competing priorities, and you want to attach your program to one of those outcomes. And then ideally, as in the pipeline example and the fixed asset example we talked about, drill down from there. And I think for folks working in the data space and in CDO organizations, that's very powerful.

 

Ken Dunn (31:17):

Absolutely. Yep. Yep. Growth, cost, and risk.

 

Chris Knerr (31:21):

Yep. Growth, cost and risk. Yeah. Thank you so much for joining me today Ken, I really enjoyed our conversation. Can people find you online somewhere if they would like to connect after watching our discussion?

 

Ken Dunn (31:35):

Yes, absolutely. So my consulting company is called datavitals.us. So go and have a look at our site and see the services that we offer or I'm on LinkedIn as Ken B Dunn.

 

Chris Knerr (31:51):

Excellent. Thank you again so much. For our audience, to see additional interviews, please visit cdomagazine.tech, and very much enjoyed talking today Ken. I hope you have a terrific rest of your day. Thank you.

 

Ken Dunn (32:05):

Thank you so much. Good talking to you.

About the Author

Chris Knerr

Chris Knerr is Syniti's Chief Digital Officer. As a former Fortune 50 Client Executive Sponsor for large-scale data migrations at Johnson & Johnson, as well as a Syniti alliance partner, Chris serves as a powerhouse whose proven success, background and experience helps accelerate Syniti’s data strategy, analytics organization and offerings.

Follow on Linkedin More Content by Chris Knerr
Previous Article
Platform Focus Series: Syniti Knowledge Platform - Advanced Data Migration
Platform Focus Series: Syniti Knowledge Platform - Advanced Data Migration

In this edition of our new Platform Focus video series, we dig into the Syniti Knowledge Platform, our indu...

Next Article
5 in 5 FAQs — Master Data Management (MDM) Edition
5 in 5 FAQs — Master Data Management (MDM) Edition

If you're ready to get up to speed on Master Data Management, you need to check out this 5 in 5 FAQ!