1 00:00:01,570 --> 00:00:09,690 \h Host Voice: Please welcome Eric Katencamp, Director of Information Technology for Publix Supermarkets. 2 00:00:09,690 --> 00:00:11,770 \h Erik Katenkamp, IT Director, Publix Super Markets: Good morning. 3 00:00:11,770 --> 00:00:17,760 \h So you probably think it's a little bit unlikely to hear from a supermarket guy at something like this, 4 00:00:17,760 --> 00:00:23,110 \h but I wanted to start off by telling you a little about myself. I actually started off here. 5 00:00:23,110 --> 00:00:28,350 \h My career started here at Kennedy Space Center about 19 years ago, believe it or not. 6 00:00:28,350 --> 00:00:32,720 \h And I worked in a group called rtsoe2. 7 00:00:32,720 --> 00:00:35,200 \h I don't know if that means anything anymore to anybody. 8 00:00:35,200 --> 00:00:42,540 \h But it was the industrial and safety -- industrial and payload safety 9 00:00:42,540 --> 00:00:46,390 \h processing group and it was a great experience. 10 00:00:46,390 --> 00:00:54,660 \h But I did move on after a couple of years. And went to Publix, where I've now been for 17 years. 11 00:00:54,660 --> 00:00:59,670 \h And I've worked in a lot of different parts of the company, 12 00:00:59,670 --> 00:01:04,170 \h but I currently run part of the information technology department. 13 00:01:04,170 --> 00:01:08,910 \h And you may wonder what kind of innovation we do there, 14 00:01:08,910 --> 00:01:11,820 \h but there's actually a lot more than you might expect. 15 00:01:11,820 --> 00:01:18,780 \h And I spoke here about a year ago at a systems engineering graduation 16 00:01:18,780 --> 00:01:22,700 \h ceremony and talked about some of the innovative things we're doing. 17 00:01:22,700 --> 00:01:26,690 \h And a lot of the things that I've been able to do as an engineer moving into the 18 00:01:26,690 --> 00:01:35,380 \h supermarket industry is bring a lot of the higher mathematics and technology to how we replenish product, 19 00:01:35,380 --> 00:01:40,500 \h how we forecast sales demand, how we decide what products go on what shelves, 20 00:01:40,500 --> 00:01:45,860 \h what -- what shelf to put them on, what mix of products to put in each store, 21 00:01:45,860 --> 00:01:50,370 \h how we get product from California to Florida and how we move product 22 00:01:50,370 --> 00:01:53,470 \h from our warehouses to our stores, et cetera, et cetera. 23 00:01:53,470 --> 00:01:57,760 \h So there's a lot of optimization logic and things that we've employed. 24 00:01:57,760 --> 00:02:01,760 \h And as computing power has gotten more powerful, we're able to crunch tens 25 00:02:01,760 --> 00:02:05,680 \h and hundreds of millions of calculations so that we can get product to the 26 00:02:05,680 --> 00:02:09,770 \h right place at the right time in the right quantities. 27 00:02:09,770 --> 00:02:14,140 \h So I actually was going to take a little bit of a different slant today 28 00:02:14,140 --> 00:02:22,240 \h and talk more from an organizational standpoint how we do collaboration. 29 00:02:22,240 --> 00:02:26,560 \h And I understand that collaboration was a big part of The theme today, as well. 30 00:02:26,560 --> 00:02:28,650 \h And I called this the collaboration myth. 31 00:02:28,650 --> 00:02:32,290 \h I'm going to talk some from my own experience of what has worked and what hasn't worked. 32 00:02:32,290 --> 00:02:36,870 \h I'm going to tell a story about where our information systems department 33 00:02:36,870 --> 00:02:41,500 \h was and where it is now from an organizational standpoint. 34 00:02:41,500 --> 00:02:46,210 \h We've made changes to be more effective. 35 00:02:46,210 --> 00:02:49,110 \h So real quick on Publix. 36 00:02:49,110 --> 00:02:51,420 \h I'm sure most everybody has been to Publix. 37 00:02:51,420 --> 00:02:57,250 \h We have 150 stores in five states across the southeastern United States. 38 00:02:57,250 --> 00:03:00,440 \h We have about 28 warehouses and ten manufacturing plants. 39 00:03:00,440 --> 00:03:06,160 \h It's a complex operation, about 40,000 items in each store. 40 00:03:06,160 --> 00:03:11,610 \h So orchestrating that whole supply chain and network is not a straightforward process by any stretch, 41 00:03:11,610 --> 00:03:15,440 \h and we believe we're on the leading edge of being able to, like I said, 42 00:03:15,440 --> 00:03:18,540 \h get product in the right place at the right time. 43 00:03:18,540 --> 00:03:24,060 \h The myth that I wanted to talk about that I've seen and experienced is that 44 00:03:24,060 --> 00:03:27,040 \h if you put a bunch of really smart people Together and give them an objective, 45 00:03:27,040 --> 00:03:30,940 \h you're going to get great results and they're going to succeed. 46 00:03:30,940 --> 00:03:34,940 \h That is not a given, and I think everybody would probably agree with that. 47 00:03:34,940 --> 00:03:41,500 \h I see organizational structures get set up in a way that assumes that this is true. 48 00:03:41,500 --> 00:03:46,910 \h And I want to go back to when I joined the information systems group there at 49 00:03:46,910 --> 00:03:54,030 \h Publix how we were structured and the type of issues we had and how we solved it. 50 00:03:54,030 --> 00:03:59,760 \h So when I got there, we had a fairly traditional matrix-type project structure. 51 00:03:59,760 --> 00:04:04,600 \h I'm sure most people in here are familiar with that kind of organizational structure, where you have 52 00:04:04,600 --> 00:04:11,610 \h functional groups of like-skilled people with a manager that typically has that skill, too. 53 00:04:11,610 --> 00:04:18,990 \h And those folks in those functional departments are allocated out to projects. 54 00:04:18,990 --> 00:04:23,770 \h And I just show the projects along the side to depict it. You had functional groups going up and down. 55 00:04:23,770 --> 00:04:29,250 \h In our case it was process analysis and system requirements and design and data base and architecture, and I 56 00:04:29,250 --> 00:04:32,250 \h know here at NASA those would be different things. Then we have various projects. 57 00:04:32,250 --> 00:04:39,190 \h So these folks would be allocated out to the projects, and a project manager would lead the work, whether it 58 00:04:39,190 --> 00:04:46,980 \h was an innovative type, innovation-type solution, delivery of a particular project or solution, 59 00:04:46,980 --> 00:04:51,360 \h and what we found is that this was highly ineffective. It was very slow. 60 00:04:51,360 --> 00:04:57,420 \h And I'll just talk about a number of the specific issues that we encountered. 61 00:04:57,420 --> 00:05:02,500 \h And I know this is conventional wisdom, kind of an engineering mindsets, right, that I compartmentalized the 62 00:05:02,500 --> 00:05:08,630 \h like resources and allocate them out to projects to deliver results. 63 00:05:08,630 --> 00:05:15,820 \h But some of the key issue we saw in this environment included, first of all, the managers of these 64 00:05:15,820 --> 00:05:20,370 \h functional groups often times are not engaged in the projects. 65 00:05:20,370 --> 00:05:26,930 \h They might have 20 or 25 people, and they naturally gravitate toward being what I call a resource manager. 66 00:05:26,930 --> 00:05:31,110 \h Something I have no liking or desire to be. 67 00:05:31,110 --> 00:05:35,240 \h Someone that just kind of manages people and pushes paper around and so on. 68 00:05:35,240 --> 00:05:40,250 \h We found they were not deeply engaged with the customer, the end result of 69 00:05:40,250 --> 00:05:43,290 \h whatever project their folks were allocated to. 70 00:05:43,290 --> 00:05:47,250 \h As a result, they weren't really contributing to the ultimate delivery 71 00:05:47,250 --> 00:05:51,200 \h of the solutions that Were being worked on. 72 00:05:51,200 --> 00:05:57,550 \h Another issue is you have a natural problem here where you have people that have kind of divided loyalty or 73 00:05:57,550 --> 00:06:03,670 \h you have people who report up to one manager but have some temporary allegiance to the project manager, 74 00:06:03,670 --> 00:06:08,060 \h the person leading the particular effort they're assigned to. 75 00:06:08,060 --> 00:06:17,780 \h And again, that happens a lot, but in my experience, preventing that to the degree possible is optimal. 76 00:06:17,780 --> 00:06:21,880 \h The other problem we would have here is a lot of conflict arise. 77 00:06:21,880 --> 00:06:26,700 \h So you can have competing objectives in the structure where you have a particular manager, 78 00:06:26,700 --> 00:06:32,000 \h particular person from one of the functions that has a strong opinion about something 79 00:06:32,000 --> 00:06:36,210 \h that should happen on the project that may contradict someone else on the project and possibly contradict 80 00:06:36,210 --> 00:06:40,640 \h what the overall project manager has as their mission. 81 00:06:40,640 --> 00:06:45,440 \h The problem is oftentimes that project manager role has very little authority. 82 00:06:45,440 --> 00:06:52,780 \h So resolving these types of Conflicts oftentimes has to go way up high in the organization and is painfully slow. 83 00:06:52,780 --> 00:06:57,690 \h Meetings, prep for meetings, sometimes getting on schedules a month out. 84 00:06:57,690 --> 00:07:03,330 \h And the decision making process is incredibly slow in this environment. 85 00:07:03,330 --> 00:07:09,380 \h And so we came to a point where we were just overly frustrated with our inability to get 86 00:07:09,380 --> 00:07:15,270 \h work done and make decisions toward the things we were trying to achieve. 87 00:07:15,270 --> 00:07:18,330 \h So I'm going to talk a little about what we did about this. 88 00:07:18,330 --> 00:07:22,620 \h So in -- this is actually I have found in talking to other companies, 89 00:07:22,620 --> 00:07:26,930 \h it's actually quite rare in an IT organization today. 90 00:07:26,930 --> 00:07:32,360 \h And what we do is we went from this matrix structure to more of what we call a line of 91 00:07:32,360 --> 00:07:37,440 \h business structure where teams were defined as more permanent structures. 92 00:07:37,440 --> 00:07:44,160 \h So we took these managers, basically got rid of the project manager role altogether and got rid of the 93 00:07:44,160 --> 00:07:50,270 \h functional manager role altogether, and we defined roles called line of business managers. 94 00:07:50,270 --> 00:07:54,920 \h What's neat about this role is it's an individual who has overall accountability to a particular 95 00:07:54,920 --> 00:08:01,060 \h business area or customer, but they also directly manage the resources -- pay, 96 00:08:01,060 --> 00:08:04,070 \h performance reviews, career planning, and so on. 97 00:08:04,070 --> 00:08:09,600 \h And it allows that person to balance supply and demand and to be totally accountable to the customer and 98 00:08:09,600 --> 00:08:14,110 \h have the authority and the autonomy to make the decisions that need to be made quickly, 99 00:08:14,110 --> 00:08:19,890 \h and that person also lives with the results, balancing short term and long term. 100 00:08:19,890 --> 00:08:27,390 \h For example, immediate project delivery versus helping people grow and go to conferences and learn and so on. 101 00:08:27,390 --> 00:08:29,880 \h Trying to balance that short term and long term, 102 00:08:29,880 --> 00:08:33,060 \h that individual owns both of those as opposed to the previous structure. 103 00:08:33,060 --> 00:08:39,370 \h It was conflict between the resource manager trying to grow their people and the project, 104 00:08:39,370 --> 00:08:42,270 \h trying to deliver and actually get results. 105 00:08:42,270 --> 00:08:47,290 \h So what we have found is that this structure is extremely fast. It's extremely effective. 106 00:08:47,290 --> 00:08:55,710 \h We have grouped within these line of business teams cross-functional people that have all the skills needed to, 107 00:08:55,710 --> 00:08:58,870 \h in our case, deliver information technology solutions. 108 00:08:58,870 --> 00:09:02,520 \h And there's a manager in place that can make the decisions and Drive the results. 109 00:09:02,520 --> 00:09:05,110 \h So the projects go more up and down. 110 00:09:05,110 --> 00:09:11,480 \h The projects are not executed by pools of people from different structures. 111 00:09:11,480 --> 00:09:16,810 \h They all report to the same manager. If we get a new initiative or business area to support, we create 112 00:09:16,810 --> 00:09:20,960 \h another permanent structure with a manager, a "permanent" in quotes, 113 00:09:20,960 --> 00:09:25,610 \h nothing's permanent in this world, but a manager who has people reporting up to 114 00:09:25,610 --> 00:09:32,100 \h them directly so they can be accountable to the customer and own the resources. 115 00:09:32,100 --> 00:09:36,270 \h So I was actually going to talk about an analogy of how we run our stores, 116 00:09:36,270 --> 00:09:41,410 \h but I was thinking more it there's a continuum of types of work here. 117 00:09:41,410 --> 00:09:45,900 \h And you might be saying to yourself, that sounds fine for an operational execution environment, 118 00:09:45,900 --> 00:09:50,950 \h but in a project or innovation type environment, it doesn't make sense. 119 00:09:50,950 --> 00:09:57,540 \h And I was thinking about kind of this continuum of work types with one extreme being kind of wartime 120 00:09:57,540 --> 00:10:03,250 \h military where there's no room for consensus building, there's no room for negotiation and so on. 121 00:10:03,250 --> 00:10:09,730 \h It's a straight chain of command, command and control structure that's required to make decisions for obvious reasons. 122 00:10:09,730 --> 00:10:16,040 \h All the way to the other extreme which would be more like a think tank where folks are there contributing, 123 00:10:16,040 --> 00:10:22,760 \h it's more loose decision making, rapid decision making isn't as critical, and so on. 124 00:10:22,760 --> 00:10:29,450 \h But I would argue that what I've seen is that orb structures gravitate more toward the think tank side, 125 00:10:29,450 --> 00:10:35,820 \h to far to the think tank side and less toward the good, old-fashioned command and control structure. 126 00:10:35,820 --> 00:10:39,100 \h It's like somewhere about 15 years ago the idea of chain of command 127 00:10:39,100 --> 00:10:42,430 \h and command and control structure became like a bad word. 128 00:10:42,430 --> 00:10:46,410 \h Like what was talked about with self-directed work teams and stuff like that, 129 00:10:46,410 --> 00:10:49,080 \h not traditional command and control structures. 130 00:10:49,080 --> 00:10:59,330 \h My experience has been that good, old-fashioned line of command structures work best. 131 00:10:59,330 --> 00:11:04,860 \h Now what's critical in this role, in this structure is that line of business manager. 132 00:11:04,860 --> 00:11:09,960 \h You have to find people that have very good mix of skills to run a 133 00:11:09,960 --> 00:11:14,840 \h cross-functional organization and be accountable to a customer and so on. 134 00:11:14,840 --> 00:11:19,530 \h The first thing is they've got to have technical skills which means they've gone deep, 135 00:11:19,530 --> 00:11:22,050 \h that's what we call it. They've gone deep once into something and 136 00:11:22,050 --> 00:11:26,010 \h really understand technology and the complexity of technology. 137 00:11:26,010 --> 00:11:28,650 \h They obviously have business skills, project management skills, 138 00:11:28,650 --> 00:11:31,170 \h and leadership management and people skills. 139 00:11:31,170 --> 00:11:34,630 \h And I'll admit it's tough to find people that have that complete package. 140 00:11:34,630 --> 00:11:41,020 \h And we actually use kind of a pie chart where we classify everybody and say -- you know, 141 00:11:41,020 --> 00:11:43,840 \h one person has two pieces of pie, another has three pieces of pie. 142 00:11:43,840 --> 00:11:47,860 \h We want a lot of managers that have all four pieces of that pie that 143 00:11:47,860 --> 00:11:52,570 \h can actually run teams from end to end. 144 00:11:52,570 --> 00:11:55,980 \h The other thing I'll add is facilitation skills. 145 00:11:55,980 --> 00:11:57,830 \h We obviously are dealing with professionals. 146 00:11:57,830 --> 00:12:04,320 \h So this command and control-type structure has to be there for decision making and expediency, 147 00:12:04,320 --> 00:12:08,400 \h but the reality is with professionals, folks have to be heard. 148 00:12:08,400 --> 00:12:14,810 \h So it's important that the manager also has the capability to get input from everybody and facilitate discussions. 149 00:12:14,810 --> 00:12:21,060 \h But at the end of the day has to make decisions and do it in a rapid manner. 150 00:12:21,060 --> 00:12:23,870 \h And I'm going to close with this concept called work. 151 00:12:23,870 --> 00:12:27,860 \h I'll tell you just a little about our president at Publix. 152 00:12:27,860 --> 00:12:33,860 \h The president of Publix of a multibillion dollar organization has a high school education. 153 00:12:33,860 --> 00:12:39,030 \h He's one of the most brilliant people I know. He often talks about this concept of work. 154 00:12:39,030 --> 00:12:43,030 \h He has this book actually sitting in his lobby. 155 00:12:43,030 --> 00:12:49,990 \h And he often talks -- it's a "national geographic" book, has pictures of all types of work like this. 156 00:12:49,990 --> 00:12:52,420 \h I'm not exactly sure what types of work those folks are doing. 157 00:12:52,420 --> 00:13:00,450 \h The concept is whether someone is sweeping a floor in a store or building an optimization algorithm, 158 00:13:00,450 --> 00:13:05,350 \h at the end of the day everybody's trying to produce output, everybody's trying to do work. 159 00:13:05,350 --> 00:13:14,290 \h And I've seen organizational structures get too focused on thinking, conceptualizing, and so on. 160 00:13:14,290 --> 00:13:19,190 \h At the end of the day, we all have to have output. We all have to have work that needs to be done. 161 00:13:19,190 --> 00:13:24,840 \h I would just suggest as I've had to do myself, next time a big project or 162 00:13:24,840 --> 00:13:29,590 \h endeavor comes up and you're figuring out how to structure it, 163 00:13:29,590 --> 00:13:35,800 \h at least consider building an actual structure with a command and control environment 164 00:13:35,800 --> 00:13:39,540 \h where there's a manager ultimately accountable and the team directs to them directly 165 00:13:39,540 --> 00:13:44,020 \h as opposed to a matrix environment where folks are there, 166 00:13:44,020 --> 00:13:50,090 \h the project manager doesn't have a lot of authority, and you're spending weeks and months trying to resolve conflict. 167 00:13:50,090 --> 00:13:55,680 \h And I know there are lots of models in between, but the challenge I would offer and something 168 00:13:55,680 --> 00:13:58,990 \h I've had to do is just to -- the thing about structuring work the good 169 00:13:58,990 --> 00:14:03,280 \h old-fashioned way where somebody's in charge as opposed to loose structures