A Mechanical Clutch Isn’t a Steampunk Purse, w/ Bill Sundwall

Rob Collie (00:00:00): Hello, friends. Today's guest is Bill Sundwall, a director of client services here at P3 Adaptive and one of my favorite human beings. Perhaps the most succinct way to summarize what working with Bill is like, our company is now more than 50 people strong. If you tell me that I can only get the opinion of, let's say, three or four people on any particular topic, whatever comes up that day, Bill is going to appear on that list quite frequently. Whether it's about technology, process, people, especially people, or even if it's just something pop culture or funny, you're going to want Bill to weigh in.

Rob Collie (00:00:36): We didn't mention this in the podcast, but, for instance, Bill wrote the script for our LetterKenny video, which was just a lot of fun. If you haven't seen it, we'll link it in the description.

Rob Collie (00:00:46): I think all of those things I just said about Bill absolutely shine through in this conversation. As an added bonus, I think this might be the only podcast you'll ever listen to in your entire life where midway through, the host gets bit on the neck by a pit bull. Okay, okay, it doesn't actually happen while we're recording, so you're not going to hear any violent sounds. No trigger warning necessary. But, yeah, we had to stop recording and come back because there was a quick trip to the ER in the middle there.

Rob Collie (00:01:13): I'm fine. Everybody's fine. The dog's fine. Everybody's great. There's even a little bit of a harbinger in there where I comment on my phone blowing up and I'm ignoring it. Oh, big mistake. Maybe you'll even hear the moment when I have to throw the headphones down and run out. Wordagamis, pop cultural references, and dogs. Oh my. All that and more await you when we get into it.

Announcer (00:01:38): Ladies and gentlemen, may I have your attention please?

Announcer (00:01:42): This is the Raw Data by P3 Adaptive Podcast, with your host, Rob Collie. Find out what the experts at P3 Adaptive can do for your business. Just go to p3adaptive.com. Raw Data by P3 Adaptive is data with the human element.

Rob Collie (00:02:05): Welcome to the show, Bill Sundwall. Another one of those it's been in the making for a long time. How many times did we reschedule this one?

Bill Sundwall (00:02:11): I think only once or twice. The real trick was I wanted to be on here, but it seemed like there were only really two routes available to me. They were either to get hired again and have the recording done before I got hired again-

Rob Collie (00:02:24): Yup, the Liz Rogers route.

Bill Sundwall (00:02:25): ...or get promoted to director. So one of those was better for cash flow. So that's what I went with.

Rob Collie (00:02:32): This is the reason to move up the corporate ladder at P3 is to get on the podcast.

Bill Sundwall (00:02:37): Yeah.

Rob Collie (00:02:38): On the director pipeline even, we're falling behind. When we set this up, we started to set this up with you. You were the only director who had not been on the podcast. Well, now that we get around recording with you, we have two others, two other newly minted directors.

Bill Sundwall (00:02:52): Although Justin was on an early episode, wasn't he?

Rob Collie (00:02:55): Yeah, that's true.

Bill Sundwall (00:02:57): [inaudible 00:02:57].

Rob Collie (00:02:57): That's true. That's true. He filled the Tom role in the booth one time, I think. But probably a little bit more talkative than the 7% LaRock. We should call it the LaRock ratio. It's actually hard to say LaRock ratio.

Bill Sundwall (00:03:08): Actually, what about the LaRock threshold? That's got a real great Star Trek vibe.

Rob Collie (00:03:12): It does. I can dig it. Our first pop cultural diversion.

Bill Sundwall (00:03:16): Somebody make a mark on the chalkboard.

Rob Collie (00:03:18): You're a director of client services at P3 today, but let's go back to the beginning. Where's your first collision with data? When did you discover your latent data gene?

Bill Sundwall (00:03:29): So long story short, right? No, it's not. Quite the opposite. Anyway-

Rob Collie (00:03:34): Short story long.

Bill Sundwall (00:03:35): Yes, exactly that.

Rob Collie (00:03:37): Is there a Sundwall inflation ratio?

Bill Sundwall (00:03:39): We're going to be able to, I think, infer it from the contents of this podcast.

Rob Collie (00:03:43): Fantastic.

Bill Sundwall (00:03:45): We are clearly ballooning it up. So that's great.

Rob Collie (00:03:45): We're going to take the noise to signal ratio and just peg it.

Bill Sundwall (00:03:48): Yeah. So amongst the nontraditional paths to the data career that we tend to celebrate on this show, I'm not going to say I win for most nontraditional, but it's up there. So I was in the midst of a complete career retooling. I had gone off to college on an academic scholarship that I had managed to claw my way into keeping for four years, despite ...

Bill Sundwall (00:04:13): The story I'm telling myself these days is just enough undiagnosed ADHD to where high school was fine but college was a wreck. I know you can commiserate. Made it through four years of that. Scholarship was done, had a big stack of credits that did not amount to a degree. Went back home, collected myself for about 18 months, and moved back to town, got the McJob that you get circa 2009. The gaping chest wound that was the US economy was recovering.

Bill Sundwall (00:04:44): Got myself positioned to get back into school and took inventory of what had happened there and was like, you know what the real problem is? It's academic writing. So if I go back to school, I should do something that does not involve that, and preferably something with a tangible career path coming out of it, which to my mind go into a state school in southwest Missouri. Pretty much narrows you down to vaguely something to do with computers or accounting.

Bill Sundwall (00:05:13): At this point, I'm running scared enough that even general business, I don't know, someone might stick a paper in there, and I would just rather not. So oddly enough, I went the accounting route.

Rob Collie (00:05:23): All right. Yeah. Okay.

Bill Sundwall (00:05:25): So got back in there, was doing that pretty well, and doing school about three-quarter time and working to stay fed, stay housed.

Bill Sundwall (00:05:34): So first weird things about Springfield, Missouri, sidebar, we're an odd hub for industrial remanufacturing. For reasons that have been fairly well-documented, for anyone who actually wants to know that story, go look up Inc. Magazine interviews with a guy called Jack Stack, and you'll learn all you want to learn.

Bill Sundwall (00:05:54): Suffice it to say, for the purposes of this story, that's the way it is. And, tying it together, those companies tend to do summer internship programs that are actually paid. That's a thing we do here in the Midwest.

Bill Sundwall (00:06:05): Got on one of those with a joint venture between SRC, which is Springfield Remanufacturing Corporation, which is the local side of things. This particular one was a joint venture with Case New Holland, which is second behind John Deere, I think, unless Mahindra is making some big move globally in the ag equipment sector. So we're talking tractors, combines, et cetera. [inaudible 00:06:26].

Rob Collie (00:06:26): Sunday, Sunday, Sunday. Sunday.

Bill Sundwall (00:06:31): Nice.

Rob Collie (00:06:32): Bill Sundwall in the accounting internship. Okay.

Bill Sundwall (00:06:34): It wasn't the accounting internship, though.

Rob Collie (00:06:36): Uh-oh.

Bill Sundwall (00:06:37): It was, of all things in the world, a warehouse internship, which sounds like a boot heel of a job, but actually was a lot to do with, here's how it starts turning into data, warehouse slotting for effectively picking pick lists for delivery to the manufacturing line. So like how do we cut down the amount of time it takes? Which turns out to be really fascinating stuff, right?

Rob Collie (00:07:02): Yeah. Yeah.

Bill Sundwall (00:07:03): We're talking drawing out pallet racking diagrams in Excel. That's where I first went back and uncracked VLOOKUP. This was about 2013. Started digging into it from that side, and it just really clicked for me. It was like this is how my brain operates, as it turns out.

Rob Collie (00:07:19): Wow. Okay. I've never heard any of this. There's a few things that struck me there. So there isn't a more computer sciencey type of problem really than the one you just described. It actually combines two of the super challenging NP-complete problems described in computer science.

Bill Sundwall (00:07:37): Do you happen to mean integer programming here?

Rob Collie (00:07:40): No.

Bill Sundwall (00:07:40): Because that's where I ended up, and it was ugly.

Rob Collie (00:07:43): I've never even heard the words integer and programming strung together directly like that. So let's come back to that. So bin packing, optimal bin packing, is an NP-complete computer science problem, and the traveling salesperson problem. The traveling salesperson problem is another NP-complete problem.

Rob Collie (00:07:59): When we did the podcast with Shishir Mehrotra, he reminded me of something that I'd long forgotten, which is that all NP-complete problems are basically the same problem in a weird way.

Rob Collie (00:08:07): But you had both. You had bin packing, stacking things on shelves in an optimal fashion so that you store things efficiently. You also had the traveling around, like the forklift going around grabbing everything, like that problem.

Rob Collie (00:08:20): Okay. So how do you end up ... From the first day that you show up for that internship, how does that lead you to diagramming things in Excel? Was that something that everyone was already doing there, or how do we jump that particular track?

Bill Sundwall (00:08:35): Basically the remit was we want to optimize picking. We would like to understand how to do it in a more general sense, but we're going to start scoped to this one production line, building old-school mechanical clutches, quick sidebar, the sort that don't actually get used in tractors anymore.

Rob Collie (00:08:53): Mechanical clutches. This is not like a steampunk purse. This is-

Bill Sundwall (00:08:58): No.

Rob Collie (00:08:59): But now this internship, though. Was the purpose of the internship from the jump, was it, "Help us optimize this"?

Bill Sundwall (00:09:05): Yes, and it's whatever methods that you find beneficial. And so, as we frequently end up talking about, Excel can do anything at 70%. So why not diagram out pallet racks and then be like, okay-

Rob Collie (00:09:21): Why choose Excel? Why not choose PowerPoint or Visio? It's more graphical.

Bill Sundwall (00:09:25): Because I don't hate myself. I mean there was some Visio involved too, to be prettier, but it was Excel because that allowed me to operate on it as data.

Rob Collie (00:09:33): You'd seen Excel before this. Because of the accounting stuff?

Bill Sundwall (00:09:37): Basically, yeah.

Rob Collie (00:09:38): Okay. All right. So you had an awareness of Excel, thanks to the short time already spent in the accounting. You show up and they say, "Okay, go make it better." So in 2013 was your first fee lookup.

Bill Sundwall (00:09:52): Mm-hmm.

Rob Collie (00:09:52): I mean that seems like yesterday, Bill. I was three years deep in the PowerPivotPro blog at that point, and you're discovering VLOOKUP at that moment. You've had a very compressed development curve in the intervening years.

Bill Sundwall (00:10:04): Rather than get too deep in the weeds on that, because I could stay there for a while-

Rob Collie (00:10:08): Oh, I know you can.

Bill Sundwall (00:10:09): ... let's go ahead and push fast forward to where PowerPivotPro starts to come into it, which would be 18 months later, at which point I'd gotten hired in from the internship sideways. I mean promoted, but also made a lateral move from the warehouse side to the actual supply chain department.

Bill Sundwall (00:10:29): So I mentioned this was remanufacturing, which is essentially a reverse logistics loop on a normal manufacturing plant. So you bring in somebody's busted component. Let's go with engines here because this was specifically what we were dealing with.

Bill Sundwall (00:10:43): So dealer gets a broken down engine from a customer on exchange for a new remanufactured engine. That core comes into a core center where they evaluate it to say is this thing remotely salvageable? Basically, are the major components still good? If so, they offer the dealer back a core refund. That core comes over to the disassembly side of the manufacturing plant where they then take it down, inspect, clean pieces.

Bill Sundwall (00:11:10): Ostensibly, it looks a little bit like a repair operation, except we're taking everything down, doing a full clean, replacing all wearing parts, regardless whether it was the point of failure or not, retesting. Because we're a joint venture, we've got IP from the OEM that tells us what are the original manufacturer's standards. So we can be sure we meet or exceed all of them.

Bill Sundwall (00:11:32): So engines is the newest line in the plant. At this point I'm a supply chain engineer, which basically means I am inspecting the actual engineering department's bills of material for new construction to make sure they're actually feasible from a structural standpoint.

Bill Sundwall (00:11:47): As we've been discovering lately at P3, bills of materials are a special help when it comes to data modeling. In remanufacturing, it's even worse because the BOM is stochastic. 70% of the time we anticipate we're going to use a new part. 30% of the time we anticipate we are going to use a recovered core part.

Rob Collie (00:12:04): Let's slow down here with these big words, stochastic. You're just trying to get a wordagami. I know you. Okay. So stochastic probably has not appeared in the show until today. I happen to know what it means because of reading Adam Harstad's work, touchdowns are stochastic. Tell us what stochastic means, Bill. Mr. Fancy Vocab.

Bill Sundwall (00:12:24): It's a probabilistic process. One of N things is going to happen, and it's expressed as a percentage.

Rob Collie (00:12:31): All right. So the way that Adam uses stochastic, remember-

Bill Sundwall (00:12:34): Which is probably more correct.

Rob Collie (00:12:35): Well, who knows? I don't read books anymore. This is where I get my vocabulary lessons now is from podcast guests. So the way I understood stochastic was that it was uneven. For example, like the yardage accumulated by a player in the game of American football is relatively linear and has a lower deviation from game-to-game than the number of times they happen to cross the touchdown stripe. A player can have 150 yards every game for six games in a row and score no touchdowns, or they can have a total of 25 yards over those same six games and score seven times. Yards were presented as not stochastic. Why is stochastic bad in this situation?

Bill Sundwall (00:13:18): I mean, really, just from a complication standpoint, you can't extend. What do you use a bill of materials for to, amongst other things, one, it's the recipe for what you're going to build. But from a supply chain perspective, it's how you figure out what you should be purchasing as far as new parts over the forward horizon.

Bill Sundwall (00:13:37): Why do we have ERP systems in manufacturing? I mean aside from, yes, you need accounting functions, yes, you need the other things that get bolted on, what do you really need it for? You need it for materials planning.

Bill Sundwall (00:13:49): That's why the core of old school, ERP systems, as far as I've researched it, because I'm that sort of nerd, is that they execute, they instrument MRP algorithms, material resource planning algorithms, which basically says we're going to build in finished goods. That finished good has a BOM that looks like this. We explode that BOM and you need ... We're talking engines still here. You need six pistons to build a six-cylinder engine.

Bill Sundwall (00:14:17): So on this day that you expect to be building five, six-cylinder engines, you need to have 30 pistons in stock. Today you have five. The lead time from the manufacturer is Z weeks. So when do you order that?

Bill Sundwall (00:14:31): So you hold the inventory the least amount of time as possible. So you maximize return on assets from an accounting perspective.

Rob Collie (00:14:38): Damn. Yeah.

Bill Sundwall (00:14:39): So how does this get worse in remanufacturing? Well, you're going to use 70% of six pistons, and you can't really order 4.2. They frown on that. Also, just because it's expressed as a percentage in the bottom doesn't mean it's going to work out that way. Sometimes you're going to use all six new. Sometimes you're going to have six core recovered. So you have to account for essentially how much slop are you willing to tolerate. All that to say how does this hook up to Power Pivot?

Bill Sundwall (00:15:07): Here's the other element. The ERP system we were actually running there, I would occasionally grab a sneaky screenshot as it was loading up because it was not AS400, but it definitely was green screen. It had a copyright date slightly older than me, and I am 47 as we're talking today.

Rob Collie (00:15:26): So this screenshot you're talking about was taken with an actual camera, because the system that you had was not actually capable of taking its own screenshots.

Bill Sundwall (00:15:33): I mean you could grab it with Windows, right? Green screen still dumps to the clipboard.

Rob Collie (00:15:38): Oh, I see. I see. Right, right, right. Okay. I was thinking that the whole system wasn't running windows. It was just a green screen terminal on its own.

Bill Sundwall (00:15:44): No, we were terminaling in some sort of agent VM.

Rob Collie (00:15:48): That's pretty high tech, man, running it in windows. That was a huge upgrade.

Bill Sundwall (00:15:54): It was great.

Rob Collie (00:15:54): It's hard to imagine running an efficient business at all consuming things that are randomly broken. So now I understand the use of the word stochastic here. Again, I like to play the pseudo-physicist card every now and then. One of the theories these days is that the arrow of time only going one direction is because of entropy. You're really, really working against entropy, manufacturing in reverse. You don't even get to control your starting conditions.

Bill Sundwall (00:16:18): Right?

Rob Collie (00:16:19): It's crazy. This business can't exist. You're just making this up. This will never get off the ground. Okay, I believe you that it's a thing. And so, you're getting into Power Pivot. How does Power Pivot help with this? Are you running Monte Carlos or something? What are we doing here?

Bill Sundwall (00:16:38): Not even that far. There was no subsystem within the ERP to capture core disassembly yield. No one had anywhere to put in ... We got three pistons out of this one because, over time, you would want to correct those percentages between new and core in the bomb to reflect your latest and greatest knowledge.

Bill Sundwall (00:16:58): But that data could not be collected except they were collecting it just on paper that someone was typing into a spreadsheet, which was all well and good, but now you've got easily a dozen different engine models coming in the back door and they're only recording half of the equation. They're only recording yield.

Bill Sundwall (00:17:20): So on this particular disassembly, we got three pistons out. Well, that's great. But what is our rate of piston fallout? How many have we lost over the past month, six months, year? How do we reset our percentages?

Rob Collie (00:17:36): Reset? What do you mean reset? How do you even know your percentages?

Bill Sundwall (00:17:41): So the engineering team, when they initially set up the bills of materials, will go through and do a few ... Not even simulated, honest to goodness they'll tear down five or six-

Rob Collie (00:17:48): I see.

Bill Sundwall (00:17:52): ... to get a starting pass.

Rob Collie (00:17:52): They make a prediction, like a forecast, essentially, like, yeah, we tore down a handful and this is the percentage that we saved. This is the percentage that was bad. That's clearly going to hold up ... That sample size of five engines is clearly going to hold up over the lifetime of this business. Totally.

Bill Sundwall (00:18:07): Yeah, that was statistically significant, I'm sure.

Rob Collie (00:18:08): Yeah. Okay. So when you say reset, that means actually make accurate.

Bill Sundwall (00:18:14): Yeah. Like everything in manufacturing, at least in my experience, like it was one of those things you do at standard cost roll, because it's a wild time when the whole business can change. So why not?

Rob Collie (00:18:26): All right. So you're not even collecting the data to feed this analysis. You're not even collecting it digitally. So what happens next?

Bill Sundwall (00:18:33): So someone other than me gets the wonderful job of filling it out on a spreadsheet, but where this leads to Power Pivot is that I needed a distinct count, because I could take the bill of materials for a given model and say this engine should have six cylinders. So essentially we can get the denominator of the fraction that way-ish. But there's a problem. We have yield.

Bill Sundwall (00:18:56): So we've gotten 42 of these cylinders out of engines of this model recently. But how many should we have gotten? Well, for that, I need to distinct count the model number of the engine. Circa 2014, how do you do a dynamic distinct count in Excel? There's one way that doesn't involve VBA. Every time I've tried VBA so far, it works for about two minutes, and then something changes and it blows up.

Bill Sundwall (00:19:21): So I find this guy writing a blog about how do you set up this wild and crazy thing that now exists in Excel? That kind of solves my problem.

Rob Collie (00:19:30): All right. So distinct count. You needed a distinct count in a pivot table in Excel. What the hell are you going to do? So you go to Google, probably, and you land somewhere. Did you land on my blog?

Bill Sundwall (00:19:42): Yeah.

Rob Collie (00:19:42): All right, there you have it.

Bill Sundwall (00:19:43): Or at the very least, if I didn't land there initially, the rabbit trail was such that dynamic distinct count, Excel, led me to Power Pivot. Then it's like how do I do any of this? And that led me to you.

Rob Collie (00:19:57): Yeah. In the early days, actually, like in 2010, we didn't even have the distinct count function. We had count rows of distinct, count rows of the distinct function. But distinct counts a lot faster than count rows distinct and a lot easier to write. Ooh, those were the green field days. Every day brought a new discovery, what you could do with the tools. I'm sure that's still going on today. I'm just not really in the mix as much.

Rob Collie (00:20:25): Okay. So you join at least two or three other guests of this show, and many of our employees who distinct count was the gateway drug. So you discover that. So now you got to go download the ... Is this like 2014, 2014 and a half?

Bill Sundwall (00:20:41): Yeah.

Rob Collie (00:20:41): Okay. So you get your distinct count and that's it, right? You're like another time. It's all good. I don't need to do anything else.

Bill Sundwall (00:20:47): No. By that point I was beginning to understand what measures we're getting to enough that rather than ... Quick sidebar. In foundations, I spend time talking about business math is oddly simple from a mathematical operations standpoint. What are we doing usually? We're adding a few numbers, we're subtracting a few other numbers. When we get really fancy, we subtract something like one year from another year and divide it to get a change percentage.

Bill Sundwall (00:21:15): Not fancy mathematical operations at all. Get those factorials away from here. They don't belong in this house, et cetera. But where does a pivot table choke? That third one. Division in a pivot table is a bad day. For the same overall mathematical reasons that we don't average averages, you want to make division usually about the last thing you do. What does that usually mean? That you fire up two pivot tables on a hidden sheet and do the division, and then somebody moves your cheese and next month everything breaks. So measures look like a solution to that.

Bill Sundwall (00:21:47): So, fundamentally, this fallout problem where we're measuring what was yield over ... I'm bad at sports metaphors, but I will use one for the sake of a podcast. It's yield over at bats. How many could we have yielded? That was a lot easier to implement as a measure. Then just say, okay, so six-cylinder Iveco engines yielded 42% on cylinders in the last quarter.

Rob Collie (00:22:12): So we talk about what's our percentage of home runs that we hit when we're in the red zone, to mix a baseball and football metaphor. How many rebounds does Tom Brady get?

Bill Sundwall (00:22:24): During icing.

Rob Collie (00:22:26): That's funny. All right. So that launched you into a broader DAX and data modeling mindset. What happens next from there? So you spend a few more years at that organization, don't you, or do you help around a bit in manufacturing?

Bill Sundwall (00:22:41): I did, yeah. So I end up wearing several different hats in the same organization. Oh, in the back of that, and one other thing that I'll go deep on, we have limited but extent access to some of Case New Hollands larger data, because they have a tremendous logistic system.

Bill Sundwall (00:22:59): One of the things they have is a system that reaches into 90% of their dealership locations' on-hand inventories and can tell them essentially what is sitting on the shelf of each skew at each dealership, which is pretty tremendous from a what should we suggest that our dealers buy situation. We're able to get a slice of that that's just the remand data that comes to us in a text file every week, pipe delimited, runs about 600,000 rows per file.

Bill Sundwall (00:23:30): So speaking to another foundation's point, any one file is fine, but if you want to do time series analysis, time series analysis with inventory might be important-

Rob Collie (00:23:42): Maybe.

Bill Sundwall (00:23:42): ... you're a bit out of luck. So the current sales director who was our CNH contact in the larger sense sends me one of these files one day and says, "Can you make sense of this?" Because for one, it's pipe delimited, which looks like gobbledygook if you're not used to seeing exotic delimiters.

Rob Collie (00:24:00): It's kind of like the five or six characters on the keyboard. When I introduce the or operator when I used to teach classes, I would say, "Okay, this is the last key on the keyboard you've ever used." It's like you've got to use it twice, too.

Bill Sundwall (00:24:14): Yup.

Rob Collie (00:24:15): Yeah, pipe delimited.

Bill Sundwall (00:24:17): Turning that into anything useful got me his attention, along with some other projects I was working on. So I went from supply chain engineer to product management data analyst. Spent a couple of years there, ended up as financial analyst for a couple of years.

Bill Sundwall (00:24:33): Then we finally rolled that old dinosaur of an ERP system out the door and the IT director needed a couple of business analysts. So I spent the last couple of years working on that stuff, which is where things really leveled up. I got into SQL for the first time because our ERP backended straight into SQL server. What else are you going to do?

Rob Collie (00:24:56): You need SQL server. You need SQL and you need a decoder ring for their halfway intentionally and halfway unintentionally obfuscated database schema.

Bill Sundwall (00:25:05): It wasn't even all of that obfuscated once you understood normalization decently. Here's the thing. It was implemented in SQL server, but it had originally been a progress-based ERP, which is some sort of weird hybrid imperative and declarative language, as I understand it. Somebody's going to tell me how I'm wrong, and that's okay. I am fine being wrong about that.

Rob Collie (00:25:29): Those people are not on this show at the moment. They're going to have to howl in pain while listening, going, "Oh, he so got it wrong. What a noob. Dysfunctional declarative," whatever. I don't know.

Bill Sundwall (00:25:41): Yeah, whatever. The upshot of it being the way they translated it to SQL servers, if you wanted to learn, one, what cursors are and, two, why you should never, never use them, go study that ERP system.

Rob Collie (00:25:55): It's kind of like the related function, learn it, and then immediately resist the temptation to use it. It's a good function. In the hands of a capable professional who knows what they're doing, you can do good things with it. But nine times out of 10, you give someone that fork and they stab themselves in the eye with it.

Rob Collie (00:26:10): It sounded to me, you're about ready in this part of the story, to just start running things. You're going to be remanufacturing God. Why jump out of that world? Why come this way?

Bill Sundwall (00:26:22): That's a great setup.

Rob Collie (00:26:23): I didn't mean to do [inaudible 00:26:24].

Bill Sundwall (00:26:23): So there's this guy whose blog that I've been reading to continually build on these superpowers that I've discovered, that have given me a string of promotions over the past four or five years. He makes this turn from talking so much about the tech side of things to talking about the more human element. You started writing about ...

Bill Sundwall (00:26:45): I'd say if there's one that's more fundamental than anything and one that I keep going back to, it's the plumbing versus taps analogy. Those things were building and informed the way I approached the work I did for other people.

Bill Sundwall (00:26:59): But at a certain point, I found myself sitting in the IT director's office about halfway, two-thirds of the way into my tenure as a business analyst, talking about how, yeah, we've just kicked off this new ERP system. Like any of those implementations it's been rocky, but there are things we need to get into people's hands. We need to worry about that last mile stuff, because then they'll get off our backs a little bit and we can go take care of that other structural work.

Bill Sundwall (00:27:25): I deploy said metaphor without a particular bias to it. He says, "Yeah, I think, fundamentally, I'm a plumbing kind of guy. That needs to come first.

Rob Collie (00:27:36): Oh, I see.

Bill Sundwall (00:27:36): There was the record scratch inside my brain. It just so happened that that was the point at which there were also a lot of blog posts coming out about, "Hey, we're hiring. We would like people to do this full time."

Rob Collie (00:27:50): Okay, I get it now. I get it. Okay. If you've never heard this metaphor, it is probably the first episode of the show you've ever listened to. But just in case, this metaphor that we use is that the BI industry existed, and I think even today, to a larger extent, it still does, on the notion that, yeah, yeah, business, you're thirsty. You need water, you need faucets. But let's not be too hasty now. Let's not talk about building you actual reports and things like that, or dashboards that actually help you. First, you've got to get responsible. You've got to go build the proper foundation for all this stuff first.

Rob Collie (00:28:25): That's the plumbing. This plumbing project is intentionally designed to run forever. It's a feature that it's never done. It's inherent in the business model of data consultancies all the way back into the farthest dust bin of history.

Rob Collie (00:28:43): What I saw with these new tools and the ways that they actually worked was that you could actually orient around the faucet first. The plumbing that you have today might actually be good enough. It's certainly good enough for right now and it's a great proof of concept. If you do want to improve your plumbing, there's a reason to improve it as opposed to like, oh no, no plumbing for its own sake.

Rob Collie (00:29:03): So I can imagine, if the roles were reversed and you were writing about this and I was in your role in the manufacturing industry, the remanufacturing industry, and I'm like, "Oh yeah. Okay, yeah. Okay. Faucets first," that makes sense. It's possible now. It should have always been faucets first, just that the technology of the past wasn't built to support that way of working.

Rob Collie (00:29:23): Then you go and you talk to someone important at your company about that and they say, "Yeah, yeah, yeah, we're a plumbing first." You're like, "Oh, shit. How do I continue?" I could see myself acquiring a certain bounciness wanting to maybe land in a different place. What year are we talking?

Bill Sundwall (00:29:42): That's '18.

Rob Collie (00:29:44): 2018. That's fantastic. From first VLOOKUP to jumping ship. Jumping ship is the wrong metaphor. That focuses on leaving something. Let's be more kind to our current day. Jumping in the ship. That's a better way to describe it. So 2018, how close are you to your five years is what I'm working on here?

Bill Sundwall (00:30:03): I apply in 2018. In fact, I remember it pretty well because I got the test, I got the interview of death, and I stayed up till about 3:00 AM getting through to the first-stage answer, and then got up four hours later and drove to Gen Con.

Rob Collie (00:30:23): So it's 2018 and you're driving to Indianapolis.

Bill Sundwall (00:30:27): Yes. The irony is not lost on me.

Rob Collie (00:30:30): Yeah. I'm in Indianapolis at that point. What month was this? Do you remember?

Bill Sundwall (00:30:34): Yeah, it would've been tail end of July, early August.

Rob Collie (00:30:37): Okay.

Bill Sundwall (00:30:37): But things go cold because I'm not sure whether this is the structural go cold and I just handled it low key or I actually asked a weird question about how one of the metrics was to be calculated and someone said they were going to kick it up the chain. That's where things went silent for a while.

Rob Collie (00:31:00): We weren't quite as organized in 2018 as we are today. I guess it's like the old Mitch Hedberg joke. Here's a picture of me when I was younger. He goes, "Every picture of you is a picture of you when you were younger." When we say we weren't as organized back then, I mean, yeah, we were just becoming more organized over time. So, of course, we were less organized last week than we are this week. Plus, our hiring needs back then, Bill, and the rate at which it moved forward was a stochastic process.

Bill Sundwall (00:31:25): Ooh, there it is. It's a call back.

Rob Collie (00:31:28): Yeah, it wasn't linear. It wasn't smooth. It was just like, how many people do we need to hire this week? Zero. How many people this week? 0, 0, 0, 0, 1. Now it's a lot smoother. We're basically hiring all the time. We used to have this ... Kellan used to call it the inchworm problem.

Rob Collie (00:31:43): Kellan is a low-key master of metaphor. He's given us the aquarium effect, the Tetris problem. The inchworm problem was like a version of chicken and egg. You need more demand for services in order to hire more people, but you need more people to take on the demand.

Rob Collie (00:31:58): And so, it was like we'd grow one side of the business a little bit, the demand generation. You'd add incrementally 10% to demand, and then you'd be like, "Okay. Now I need to go add 10% to supply." It's like this back and forth process, which you can imagine the inchworm stair-stepping effect. I can certainly imagine us going quiet for a bit. How long will we be quiet, Bill?

Bill Sundwall (00:32:20): I think till April.

Rob Collie (00:32:22): Wow. See, that's stochastic.

Bill Sundwall (00:32:23): By that time, putting another eight, nine months on it, I was agitating for just something different at that point and decided to throw the Hail Mary. Reached out one last time and that got the ball rolling again. Three years later, here we are.

Rob Collie (00:32:38): Fantastic. So for those of you listening, please, please, please understand that we are not like that today.

Bill Sundwall (00:32:42): Yeah, absolutely.

Rob Collie (00:32:45): If you're interested in coming to work here, you're not going to be sitting drifting in limbo. That's not how it works anymore. We've gone non-stochastic in our hiring process and we are a lot more organized. We put a lot more effort into it. We're a lot more systems involved.

Bill Sundwall (00:32:58): It's a scaling problem, right?

Rob Collie (00:33:00): It is.

Bill Sundwall (00:33:00): Because, essentially, once you get to a certain point, you can tolerate the gap of hiring capacity before you can fill it a lot more easily. One incremental person on 50 is a lot different than one incremental person on 10.

Rob Collie (00:33:13): Yeah, five incremental on 50 is a lot less of a deal than one on 10, even though the ratios are the same. It's like Godzilla is impossible. When you scale up an animal that works at a certain size, maybe there's a bipedal version of Godzilla that existed someday that was like six feet tall or something, two meters tall, that would work. By the time you get to Godzilla's height, however, because the cube-

Bill Sundwall (00:33:40): Yeah, square cube wall.

Rob Collie (00:33:42): ... of volume, square cube glossy, there we go, that monster would have to be all bone to hold up its own weight. There wouldn't be any room for muscles or nerves. So this is in reverse. You can't just scale and expect everything to be the same. There's an economy of scale as you get bigger. Then suddenly things just start to work better. Again, you've got to go backwards with this metaphor. It's like the remanufacturing version.

Bill Sundwall (00:34:08): Yeah. I forget whose it is, whose quote it is. Are you getting to the quantity as a quality all its own?

Rob Collie (00:34:14): Well, that's a Stalin quote. The hardest thing about growing our company and hiring is that as we get bigger, we need to keep the quality bar of our candidates super high. Everyone we hire, they don't work ... On an island, we support each other. You have to be pretty broad-spectrum competent. We never even know ...

Rob Collie (00:34:43): It was probably part of the test, Bill, actually back then. We were probably just saying, like, leave the candidate on the doorstep without any food or support or encouragement for 48 hours. If they're still there, then they're worthy.

Rob Collie (00:34:55): So when we hire you, we don't know that we're hiring one of our future directors. Now that we've seen this story so many times, we also have to keep that in mind. We're not just hiring consultants. We're also hiring ... Especially at this phase in our lifecycle, we're hiring our future leaders. So our quality bar combined with quantity has a quality all its own. Take that, Stalin.

Bill Sundwall (00:35:17): Love it.

Rob Collie (00:35:18): So you come on over to P3, principal consultant. That's where you start out. What's the road from there to director, as if I don't know?

Bill Sundwall (00:35:27): Just dig in and do the thing. If I use the word luxuriate, that makes it sound like I'm slacking. No, it's just like that welcome home environment. It is just, by and large, easy to do good work when doing good work feels as good as it does.

Bill Sundwall (00:35:44): I know that coming in just from the prior experiences I had had in terms of the project work. But it turns out training's that way too, if I do say so myself. Try to pass on a lot of the narrative bits of training that I've developed tribally for myself.

Bill Sundwall (00:36:01): Just thinking about a manufacturing environment. A lot of people come to us from a more knowledge work kind of environment to where the communication bar is a little different there. The people I got to work with, but also had to work with in a manufacturing environment, you really have to think about how you communicate the results of knowledge work, because you've usually got to build the pyramid from the bottom.

Rob Collie (00:36:26): Can you give me an example that highlights that? It sounds like you're saying, if I understand, that there's a higher premium on precise communication in the manufacturing industry than there might be in the average industry. Am I hearing that correctly?

Bill Sundwall (00:36:40): Yeah. You're headed in that direction. Here's how I frame it up. I mean this amounts to giving away a little bit of the secret sauce-

Rob Collie (00:36:47): Don't do it.

Bill Sundwall (00:36:47): ... but it's hard to implement without the framework. So what I tell PCs, as they're gearing up for their first training, I usually frame it around who's that person that you know that was "good" at Excel, finger quotes here, but in a way that just made them dangerous, the sort of person that the fact that the color column in our dataset starts with some values that are N/A, they would throw the whole thing out because they know enough Excel to know if they see an N/A, there's an error somewhere. So it can't possibly be valid.

Rob Collie (00:37:18): I would say that in your ... This is a sidebar. In your day-to-day life, you're probably a lot better about that stuff than I am. This is actually why sometimes I miss some things, that on second listen, I go, "Oh, I really missed an opportunity there." There was something really important and profound, and I didn't quite understand it all. I've got these other corners of my brain that are having to watch the whole thing.

Bill Sundwall (00:37:39): Sure.

Rob Collie (00:37:41): And so, talk about ADD. I've got a play.

Bill Sundwall (00:37:43): Metacognition costs.

Rob Collie (00:37:45): Yeah. That's right. That's right. So I've got an overhead. So who's that person that's good but dangerous to get the N/As? Okay, so let's pick up from there if we can.

Bill Sundwall (00:37:52): So our sample dataset that we use for foundations has a column for product color, and there are N/As in there. This person that is good enough at Excel to be dangerous is going to say, when you present them the results, "Well, that has N/As in it. I know enough about Excel to know that if there's an N/A in my pivot table, there's some sort of error back in there. So clearly your dataset must be an error. You need to go back and fix that." That's the jumping off point to, like, that's where we first encounter DAX and all of that, is in correcting that problem.

Bill Sundwall (00:38:23): But the story about that person, that psych profile, if you like, that's something I encountered a whole lot more, I think, especially ... It tends to resonate with everybody, but to me, there were a lot of those people in manufacturing, that were good enough at the tooling, but good enough to be dangerous.

Bill Sundwall (00:38:43): I mean I'm thinking of one particular person, and this is the person that I had to tell multiple, multiple times ... He was at a line manager level and would email reports up to the C-suite that would inevitably crash because he had live VLOOKUPs to files on his desktop. No matter how many times you tell that profile, yeah, you've got to paste values before you send that out or it's going to do it, it's still going to happen.

Bill Sundwall (00:39:13): Getting to the core of how does this toolset allow you to steer into that skid and think about what your problems are going to be ahead of time in this communication and refine it. Not just think of what is the next virtuous cycle question going to be, but also what is the doesn't quite get it cycle question going to be?

Rob Collie (00:39:34): So I have an example about that. Unlike you, I love writing papers in college. Bring it on. Give me that 16-page philosophy paper. It needs to be 12 pages. Can I turn in 16? Is that okay? The professor rolls their eyes like, "Yes, I guess it's okay." All they can see is the increase in work for them to read the thing.

Rob Collie (00:39:53): When I got to Microsoft, my first job there was as a tester. I got some really good training on how to file a bug report. It was a very black and white set of rules in what you had to do. First of all, include all kinds of environment variables in your bug report. Like I'm running on this operating system, I'm running this version of the product, this build number, of the product, and then you had to include numbered steps to reproduce the problem. First do this, then do this, then do this. Then sometimes you feel like you're just overspecifying. Launch the app. Next, next, next, next all the way to screen seven. You couldn't skip things.

Rob Collie (00:40:33): So that was tedious, but at the same time I got a kick out of it. There was something kind of ... It scratched a niche, having an excuse to be that almost pedantic. Then really crucially, after the end of the steps, like the last step is the one that triggers the problem, you then had to say result, colon, like what happened, then also importantly, expected, colon, what you expected to happen instead.

Rob Collie (00:41:01): If you were just told these just really quickly, as opposed to being trained ... Like I just had walked in the door at Microsoft. I had gone to the kingdom, I had. I had left college and I was on the big stage now. I was working on Office, of all things. It was just monolith. There was a certain gravity to being taught this that got past my immune system. In normal life, I'd be like, oh, come off it. This is really tedious and stupid. I just want to cut to the chase.

Rob Collie (00:41:29): But that methodology saved us so many times. It was important that every one of those things be there. Does that rhyme at all with your experience in manufacturing?

Bill Sundwall (00:41:37): Yeah, that's essentially how you have to think through the communication with, again, these not as vested in knowledge work stakeholders, is really try to ... I'll throw words at it here. Try to cognitively empathize with what process they're going to walk through.

Bill Sundwall (00:41:59): You don't have to know a person a lot to think through what they'll be thinking through about this sort of stuff. It's not usually deeply emotional stuff, but they're going to think through it a different way than you will, and embodying that is critically important.

Rob Collie (00:42:17): I had a computer science professor in college who would say that to us repeatedly over time. As computer scientists, we are trained to communicate with the dumbest thing in the world, which is a computer. It's going to do only what you tell it to do. For better and worse, it's going to do exactly what you tell it to do. He said, "So, ladies and gentlemen, why are we so genuinely terrible at communicating with one another?" He's like, "You've got to be better than this." That's one of those things that it just sounded quirky and funny at the time. It's like I can retroactively now digest that wisdom.

Bill Sundwall (00:42:55): Thinking back on it, this is very distant to the Power BI world, but one of the last things I did, we launched a new manufacturing execution screen attached to that ERP system we had implemented, which an MES is a thing you very much get in manufacturing where you want to present a very small slice of system capability to people actually out on the line. There's very little in the need of reporting or deep, like go look at where inventory is stored or any of that kind of stuff that they need to see.

Bill Sundwall (00:43:27): So there's a cottage industry of people building these MESes, selling them at a cheaper license than a full license for the ERP. Then you bolt them on, you farm them out onto the floor.

Bill Sundwall (00:43:39): The point I'm getting at with that is there, especially, I think what I took away from that, to your point about communication with human beings versus the computer, is in educating people on the line how to use these sort of things. More often than not, it was the case that they were assuming some meaning of greater depth to an action than was actually taken.

Bill Sundwall (00:44:00): They understood it to be doing something that in fact it was very close to if you had the requisite background and you were reading the button for what it said, you knew, okay, so this is just going to move X quantity of units to the next step in the router, which is the appropriate technical term.

Bill Sundwall (00:44:20): It's not doing anything else. It's not clocking you off of that step. So you're not booking time anymore. It's not pushing any other inventory through the system. It does exactly what it says on the box, but I think unpacking and trying to anticipate what people will think will happen that does not happen is maybe the core of the whole thing.

Rob Collie (00:44:42): I've been getting phone calls and text messages and this has been the ... Apparently, my family's like, "Oh, he's in the basement doing a podcast. Let's call him." I think I missed the meta point in that MES story.

Bill Sundwall (00:44:54): Sure.

Rob Collie (00:44:55): What's the meta point there?

Bill Sundwall (00:44:56): I think that more often than not, it is not a matter of a different understanding of what is transpiring with a certain application. When I was listening to your story about the bug documentation process, is that, from that perspective, it put me in mind of the aphorism about to bake an apple pie, first you must construct the universe.

Bill Sundwall (00:45:19): Those of us that are used to thinking of things from that side, we've got the actions taken very atomized. It does what it does and it doesn't do anything else. I think most of the fine-grained issues come about because end users of whatever stripe assign more meaning to things than is actually there. And so, how to gently walk them back to this is all that's happening, precisely this and nothing more.

Rob Collie (00:45:48): Sounds like a very close cousin, if not exactly the same phenomenon, as when someone's using software and they get an error message unexpectedly, their first instinct is to think, "What did I do?" Nine times out of 10, dear user, you did nothing. Some software engineer messed up.

Bill Sundwall (00:46:08): Absolutely.

Rob Collie (00:46:09): It could just be a random bug. You're just unlucky. Also, it could just be a poor design, because you off doing a human thing with the software that human beings would just do wasn't anticipated and wasn't thought about properly. If you're on the software engineering side of this equation, it's just lovely that your customers immediately blame themselves no matter what went wrong. You've got to get a couple levels higher up in the IT organization before they start to see the truth, which is the software is bad.

Rob Collie (00:46:40): So you're assigning more meaning and more royalty, essentially. There's a little bit too much reverence of the software itself. It's dumber than you think.

Rob Collie (00:46:53): All right. So let's get on a little bit more of a personal note. So you're at P3 as a principal consultant, which we often call PC. You started doing that in ... You were here in '19, 2019, doing that job. At some point, there's almost like a choose your own adventure, like a fork in the road. Does Bill Sundwall want to continue doing exactly the thing he came to P3 to do, or does he want to try his hand at management within this team?

Rob Collie (00:47:18): We've had a few people over time that ... And I'm not sure if I remember correctly, if you're one of them, expressly stating early on, "Now I don't really ever want to be a manager," and then changing their mind and saying, "Oh, okay. I'll try it." So where are you on that continuum?

Bill Sundwall (00:47:35): That was me. It was nowhere near top of mind for me. We're talking about something that transpired about 13 months ago. I'll remember being myself when Kellan dropped an unannounced meeting with himself and you on my calendar, I had to go change pants for the day, and then talk myself back off the ledge, perhaps with Berg's help.

Rob Collie (00:48:06): Yeah. I've learned something, I think, recently, which is when we're going to do something like that, we don't make the invite list visible to the guest of honor. If we did this today, if we did it properly, Kellan would just schedule a meeting with you. I just want to catch up. I'd get a separate copy of the invite.

Bill Sundwall (00:48:24): Right.

Rob Collie (00:48:27): All right. So I'm sorry that we did that to you.

Bill Sundwall (00:48:31): No, no, it's fine. It worked out. I learned a certain measure of self-confidence of don't expect bad things to necessarily just rain out of the sky. So this was when we were first putting another, shall we say, half level in the architecture of the company.

Bill Sundwall (00:48:47): What was on offer there was to become one of the first three associate directors, which I always thought of myself as someone who would not make a good manager of people. The way the AD role was pitched, which was a lot more just like back office, help keep the organization running smoothly, help us enable scaling. I told you guys at the time this is something I would've been skeptical of, but it's like you tailored this to really appeal to the only parts that would've appealed to me.

Rob Collie (00:49:23): We were so clever.

Bill Sundwall (00:49:24): Yeah, which was great. How does that then lead into, at the end of last year, beginning of this year, rolling into the director role? Well, getting that exposure and stuff and really sitting around and thinking about it, to call back to what you were saying about the way we greet new people, again, unofficially, but with the welcome home thing. Truly, as a working environment, this has been as much home, as much community as I've ever felt. Honestly, more community maybe even than places that I have chosen avocationally.

Rob Collie (00:50:02): Another wordagami.

Bill Sundwall (00:50:03): I promise I'm not doing that on purpose.

Rob Collie (00:50:05): No, you've just got a peak vocabulary. My vocabulary is off-peak. So we have this really cool blend of this operational innovation that comes from Kellan. I would've never conceived of the AD role. Then we blend that with our ethos, which I do have something to do with, the ethos that everyone benefits when everyone benefits.

Bill Sundwall (00:50:30): It turns out.

Rob Collie (00:50:33): So when those two come together, something very interesting happens. It's greater than some of its parts. But that does seem like there's still a bit of a jump. "Oh, Bill looks at the AD role and goes, 'Huh, they think this is a management role, but I'm looking at it going, 'Ah, no, it's not. It's not a management role. It's just a little bit more scope of operational efficiency role.'"

Bill Sundwall (00:50:55): Exactly.

Rob Collie (00:50:56): All right. So which now when you say it that way, it sounds a little bit like a management role. But we managed to pull the wool over your eyes. You were definitely not in a management role. The jump from AD to director, there's no misunderstanding that. That's a management role. Why didn't your immune system reject it at that moment? That was the moment.

Bill Sundwall (00:51:15): Because of some of the parts of the AD role that I didn't realize were going to, let's extend your metaphor, inoculate me to that particular virus, because what do ADs do? Amongst other things, I think if I was going to point to one thing more than anything else ... Well, there are two throughlines here.

Bill Sundwall (00:51:32): One, the ADs are largely responsible for the early exposure of our new hires to the company culture through the onboarding process, just through how does this business operate? How do you go do jumpstarts, all of that stuff? I think what that developed in me was the interest to have some kind of a guiding hand in continuing to steer the evolution of that direction and to see that we stay true to these cultural principles, because we have built that home space.

Bill Sundwall (00:52:05): At a certain point, it doesn't matter whether you think necessarily you would be good at it or not. If you want to see it propagated, you have to take some responsibility for that leadership. Then the other side is a lot of the scheduling work. They do first pass on scheduling things. That exposes you to a lot of the inner machinery.

Bill Sundwall (00:52:28): At a certain point, I just wanted to feel the additional freedom to call shots. It's another flavor of the same thing. How are we going to sustain this culture and also how are we going to grow and develop it?

Rob Collie (00:52:41): From my perspective, near the end of one of the most disappointing movies of all time, Episode One, Palpatine tells young Anakin, "We'll be watching your career with great interest." I feel like that's how I've been watching you. So I get the part about managerial reticence or reluctance. There's just something about it. There's a lot of unknown possible ways that moving into a management role can go wrong.

Bill Sundwall (00:53:09): Here's a hot take.

Rob Collie (00:53:10): Go ahead. I like hot takes.

Bill Sundwall (00:53:12): You're familiar with the ... I don't know if we deigned to call it history if it's more just a Greek fable. I'm thinking of Cincinnatus here. Goes and hangs up his helmet and goes back to farming.

Rob Collie (00:53:23): I thought this was the person that started the Bengals franchise.

Bill Sundwall (00:53:26): I see what you did there.

Rob Collie (00:53:27): Yeah. Okay.

Bill Sundwall (00:53:28): Yeah, sports joke.

Rob Collie (00:53:30): Okay. Cincinnatus. Now have I told you the legend of Darth Cincinnatus?

Bill Sundwall (00:53:34): Right.

Rob Collie (00:53:36): I have not heard it. I've not heard it because I've been studying with the Jedi. Okay. So Cincinnatus goes to war, goes back, hangs up his helmet, goes right back to farming. Okay, what's the lesson?

Bill Sundwall (00:53:45): I'm thinking back in the pre-Julius Caesar days, how someone would essentially be the dictator of Rome for some military engagement and then you become a civilian again and how you handle that transition.

Rob Collie (00:53:59): Oh, good trick.

Bill Sundwall (00:53:59): I mean I guess my outlook always was, like, then why seek it to begin with, unless you are unlikely to encounter something so incredible in your life as to actually require that kind of responsibility of duties of authority? I guess clearly my mindset has changed a little bit, has maybe softened on that. But anyway ...

Rob Collie (00:54:20): We got to say Darth Cincinnatus, which I mean-

Bill Sundwall (00:54:23): There you go.

Rob Collie (00:54:24): ... there's got to be someone who dresses like Vader. In fact, I think I've even seen it, like someone who dresses like Vader, but with the orange and black scheme to the helmet that the Cincinnati Bengals have it. Of course that fan is called Darth Cincinnatus. I mean-

Bill Sundwall (00:54:39): Of course.

Rob Collie (00:54:40): ... I'm going to Google it right now. Darth Cincinnatus is a thing. Oh boy.

Bill Sundwall (00:54:44): So you looked at your phone. That was the mistake.

Rob Collie (00:54:46): No, Darth Cincinnatus. There's a Darth Cincinnati T-shirt. Darth Vader himself will be lording over this Saturday's Cincinnati Roller Girls Star Wars Night.

Bill Sundwall (00:54:57): All right then.

Rob Collie (00:54:59): That's the stuff you're looking for. Even from the early days of getting to know you, I was like, oh, this guy is deeply observant. Observant-

Bill Sundwall (00:55:05): Thank you.

Rob Collie (00:55:06): ... thoughtful, sees the meta through the noise. Very empathetic.

Bill Sundwall (00:55:11): Go on.

Rob Collie (00:55:14): I was like, "How do we get this person to be more important here? How do we get more of his time helping us be better, more leveraged, more central position?" Okay. One second. This is bad.

Bill Sundwall (00:55:27): All good.

Rob Collie (00:55:28): Hello?

Speaker 5 (00:55:30): Hello? We need help.

Rob Collie (00:55:31): What's going on?

Speaker 5 (00:55:33): So Bailey was in the garage. I opened the garage. She ran out and he ran from us.

Rob Collie (00:55:37): Oh, you guys. All right.

Speaker 5 (00:55:37): Can you help us?

Rob Collie (00:55:40): Well, I'm going to come help you, yes. And now we're going to have to finish this later. Bye. Okay. And we're back. Wow. In Pulp Fiction, Bruce Willis tells his girlfriend, "Without a doubt, this has been the craziest day of my life." I'm not sure this one's been the craziest, but it's definitely been ... It's been in the running. See, I told you all backstage that it's not often you get bitten in the neck by a dog, but it almost never happens in the middle of a podcast recording, right?

Bill Sundwall (00:56:08): Pretty much the opposite of burying the lead there.

Rob Collie (00:56:11): What an amazing chain of events that leads to a dog hanging off of my neck. So we're recording, right? We're recording the first session of this, which was supposed to be the only session. We weren't supposed to have this intermission in the middle. My phone's just blowing up and I'm like, "Damn it, family. You all know I'm recording a podcast. I told you all leave me alone." My phone's ringing. I'm texting, "I can't talk right now," the autoreply. Eventually, I just have to pick up the phone.

Rob Collie (00:56:43): I mean I can't imagine what it looked like to you two, Luke and Bill, me just sprinting out of here, just leaving everything on. I didn't even stop the recording, whatever. Apparently, I locked the cat, one of the cats, in the podcast there in my haste.

Bill Sundwall (00:56:57): You did. Luke and I spent more than a couple of minutes making the requisite sounds to try to get the cat to come up on camera. But couldn't quite pull it off.

Luke Pirozzoli (00:57:06): Sadly, I stopped the recording.

Rob Collie (00:57:09): No pictures of the cat.

Luke Pirozzoli (00:57:10): Well, you did rescue the cat. There's not a cat skeleton in there.

Rob Collie (00:57:12): No, no, the cat's fine. The cat's fine. Everybody's fine, it turns out.

Luke Pirozzoli (00:57:15): Are you okay?

Rob Collie (00:57:16): I'm okay. So we have a foster dog, a lab-pit bull mix. This poor fellow was an hour away from getting euthanized when my wife steps in and says, "Well, we'll foster him." So there's some complicated morality around this. The ethics of all of this are very nuanced, shall we say? Every carnivorous animal that you save, you now have to feed, what? Meat.

Bill Sundwall (00:57:42): Yeah, fair enough. Yeah.

Rob Collie (00:57:44): So there's a certain preference for certain species of animal inherent in all of this.

Bill Sundwall (00:57:48): I feel like the Venn diagram, though, of dog food and meat keeps turning more and more into the MasterCard emblem and less and less of the circle.

Rob Collie (00:57:58): Yeah, that's true. I have heard stories of vegetarian dogs that did very well. Less so cats. I don't think cats are-

Bill Sundwall (00:58:06): Yeah, they're not built for it.

Rob Collie (00:58:08): ... really, really not capable of eating a vegetarian diet. We have a bunch of cats, too. All of them are adoptees. So this guy, brand new at the house. He's got separation anxiety. He's only lived with one person in his life, an elderly woman who passed away.

Bill Sundwall (00:58:20): Oh.

Rob Collie (00:58:20): He was found guarding her. So, oh, the heartstrings. He's got all kinds of separation anxiety. And so, he can't really be left alone, especially when my wife leaves, because he's really closely bonded with her-

Bill Sundwall (00:58:32): [inaudible 00:58:32] attachment.

Rob Collie (00:58:32): ... really, really, really quickly. And so, I'm doing a podcast. My wife has an appointment also about adopting animals and saving animals. So she's out of the house. So we have my son come over, my teenage son. He comes over to house sit. But then he realizes he's got to get his act together for his job. He's got to a summer job. He needs some paperwork. He's got to get back over to his mom's house, my ex-wife's house.

Rob Collie (00:58:53): So he shows some initiative. He dives in. He gets a leash and he walks the dog the mile over to my ex's house. He's sitting there with the dog, letting the dog cool down in the garage. The garage is closed. Everything's going well. He's just letting the dog cool down for the walk back.

Rob Collie (00:59:10): Then my daughter, who's at high school, comes home early from high school and opens the garage door with the remote, and all hell breaks loose. The dog's off running. My phone's blowing up and I'm ignoring it for about an hour.

Bill Sundwall (00:59:24): This is turning into one of those highly scripted physical farce episodes of Frasier.

Rob Collie (00:59:31): Yeah. The number of things that had to go wrong consecutively in this chain of custody, it's just crazy. We're out of cars because my daughter's already driving the one car we have that's at the house. She's already driving around looking for him. So I hop on a bike, pump up the tires, all that. I'm often racing around.

Rob Collie (00:59:47): At one point, we get a tip from some people who've seen him. So I turn down the cul-de-sac that they pointed me to. There's no dog there, but I'm like, okay, if he was just here. So I dropped the bike and I run into the backyard of these houses. And guess what? These houses are on a lake. I didn't know that.

Bill Sundwall (01:00:03): Oh no.

Rob Collie (01:00:03): And way across the lake, I see this head. It looks like an otter swimming. I go, "Oh no, that's him." I don't think he's ever seen water. I don't think he's ever been in a lake, but he just took to it. He's also never seen stairs. It was very clear that he'd never seen stairs before.

Bill Sundwall (01:00:19): Half lab. I could see the water as well. [inaudible 01:00:23] natural.

Rob Collie (01:00:23): Oh, yeah. Yeah. So I debate swimming across this lake, because you don't want to lose eye contact. But it is a long way, and I decided against it. We raced around to the other side of the lake. It takes us a while to get there. We pick up the scent again, barely. We finally run him down. He's so scared, he's not even responding to me. He likes me. Only a few days in, but he likes me. I'm riding along next to him on my bike and he won't stop.

Rob Collie (01:00:45): He runs to a woman in her garage, who gets ahold of him. We put him on a leash. Everything's great. We've got him. Ah, it's a miracle. I didn't think we were going to get this dog back. So we walk him over to the Jeep to put him in. And I'm just not thinking. I'm just not thinking about what mode he's in and how little he trusts us and all of that. I pick him up and he grabs me by the neck.

Bill Sundwall (01:01:07): Yikes.

Rob Collie (01:01:09): He didn't have my whole neck in his mouth. It was more of a grazing bite, but he was still attached. I just freak out and throw him into the back of the Jeep, figuring he'll let go. He did, thankfully.

Luke Pirozzoli (01:01:23): Hey, you're lucky you're not down one throat.

Rob Collie (01:01:25): I know. I know, right? Right.

Bill Sundwall (01:01:25): Yeah.

Rob Collie (01:01:27): So, yeah, I went to the ER. I got it rinsed out. I'm fine. I'm fine. We'll put some pictures with a trigger warning, of course. Who doesn't want to see a little dog wound on the neck? Immediately I'm being told by people I know, they're like, "Oh man, I'd euthanized that dog." I said, "No, that's not the move." The Keyser Söze move is you let him live.

Rob Collie (01:01:48): By the way, he's limping around a little bit now. I think he's got a sprained ankle from the, from the dismount. So it's a case of you should see the other guy. I feel bad for him. He's getting better. We've already had him looked at. But, no, you send him back. You let him live and you send him back so he can tell all of his other dog friends this is not a man to be trifled with. You need to send a message to the community.

Luke Pirozzoli (01:02:09): Plus, you guys made up.

Rob Collie (01:02:11): We've already been sitting next to each other on the couch.

Bill Sundwall (01:02:13): It remains to be seen. I just looked it up. Let's all make a note to check in with you on June 14th. That's the next full moon.

Rob Collie (01:02:20): All right. We'll see. Yeah. Let's do that. So where were we before all that happened? Any ideas? I have no idea.

Bill Sundwall (01:02:33): Yeah. I mean after a tale like that, who could say?

Rob Collie (01:02:36): Let's recap. We've heard your life story. Pretty interesting. So, Bill, you've been central, been a pivotal figure in our hiring for a new role, PowerApps developer, just the development in itself of the things we've been talking about on this podcast for a long time now. BI and analytics are merely part of the story. It's not about being informed. It's about improvement. The goal is always improvement.

Rob Collie (01:03:02): So improvement involves action, and PowerApps is one way ... Not the only way, but is one way ... to facilitate action-taking. It's also another way to involve human beings more effectively in the data collection process.

Bill Sundwall (01:03:20): Absolutely.

Rob Collie (01:03:21): There are sometimes where your data needs to actually come from humans. I mean heaven help you when that happens. But having some forms and some applications that enforce logic and that take you to the right places and everything is data quality.

Rob Collie (01:03:34): So both upstream and downstream from Power BI. There is quite a demand for PowerApps. We have found ourselves, again, especially as our clients become more sophisticated and their BI needs become increasingly, believe it or not, at least fractionally met for the first time in their lives, they find themselves zooming back, as do we, and looking at the broader improvement cycle. And so, here we are hiring. Right now what's our target headcount for PowerApps developers at P3?

Bill Sundwall (01:04:05): I'll vague it up even further and say mid-single digits at the moment.

Rob Collie (01:04:11): Mid-single digits. Yes, okay. Yeah.

Bill Sundwall (01:04:13): For a company our size, that's a significant chunk.

Rob Collie (01:04:15): That is a significant chunk. So significant investment. We have very specific hiring philosophy at P3. Do you want to try to characterize the hiring philosophy that I laid down originally, or do you want me to do it?

Bill Sundwall (01:04:30): Obviously, the way we hire has a lot to do with sussing out technical skill, which is difficult on its face, but is especially difficult in what is only beginning to evolve out of a niche platform. When you originated hiring for P3, it all ran through Power Pivot.

Rob Collie (01:04:56): That's right, yeah.

Bill Sundwall (01:04:57): Power BI was this whole new thing. And so, the fact that it's coming into the mainstream changes the character of that. But I don't even think that's the main drive of it.

Bill Sundwall (01:05:01): Beyond that, when I introduce new people to the company and what we do and what I do and what the PCs do, especially whenever I talk about it with my therapist, she always really frames it up in terms of the unicorn people we're looking for, because, yes, the technical skills absolutely have to be there, but the communication skills, the people skills, both emotional and cognitive empathy.

Bill Sundwall (01:05:30): You have to understand the client's pain point. You have to understand what's driving them from an emotional standpoint, but it also benefits you to be able to really very cognitively think from their shoes and say, "Okay, how can we get out in front of this and start fixing problems that they don't know exists yet?"

Bill Sundwall (01:05:50): So that's a minute worth of ramble to say the hiring process is looking for those qualities. Those are very difficult qualities to suss out, and I think it does fairly well all things considered.

Rob Collie (01:06:03): Let me just crisp it up for us.

Bill Sundwall (01:06:06): There you go.

Rob Collie (01:06:06): So I think there's three principles that we stick to. One of them is very high bar. We hold a very, very, very high bar. Our jobs, because we're a high-velocity organization, we work with a lot of clients, we work as quickly as we can for them-

Bill Sundwall (01:06:22): We need the cathletes.

Rob Collie (01:06:24): Yeah. So we definitely need the cathletes, but that's coming up next. That's coming up next.

Bill Sundwall (01:06:27): Oh, sorry to steal your metaphor.

Rob Collie (01:06:28): So the thing I learned at Microsoft is that you have to set the bar high enough that you're almost certainly rejecting some people that would've been good. That's the part about all this that I hate. It's so emotionally distasteful. But it's very important that every person who walks in here is highly, highly, highly capable.

Rob Collie (01:06:50): The damage that's caused to an organization by bringing someone in that's not a good fit for the job is so high. It's bad for everything. It's bad for the clients. So you can't have that. Not having that happen needs to be an overwhelming priority. So super high bar.

Rob Collie (01:07:06): In order to have a super high bar and still find people, you have to have a very wide funnel. So you have to have a lot of at bats to sustain that. Obviously those two things go hand-in-hand. That's two things.

Rob Collie (01:07:17): The last thing is you've got to simulate the actual job. Don't be clever. Don't be so clever that you're trying to find indicators, these secondary indicators, of asking them if you were an animal, what animal species would you be, or what does this Rorschach inkblot tell you, all those sorts of things.

Rob Collie (01:07:36): Or even the thing that Microsoft did overwhelmingly when I first got there was solve these brain teasers. It turned out the job I was doing wasn't solving brain teasers. So using brain teasers-

Bill Sundwall (01:07:49): Brain tease [inaudible 01:07:50].

Rob Collie (01:07:49): ... as a screening mechanism was just dumb. That proved out in real life the people who could solve those didn't necessarily excel at the job.

Rob Collie (01:07:59): Something you touched on that's crucial is that a huge part of our job is human. It's not technical. It's the how does all this stuff fit together in the human environment? How do you work with clients? And so, you can't leave that part out and have it be just a technical exercise, can you?

Rob Collie (01:08:18): All right. So the human component's a big, big, big part of simulating the job. So I have had my hands on the creation of the Power BI related test for a long time. But then we were faced with an interesting challenge, weren't we? We need to maintain that same philosophy, but we need to do that with something that I don't know at all. I don't know how to do PowerApps in the least. So we just said, "Hey, Bill, why don't you figure this out for us? Won't that be fun?"

Bill Sundwall (01:08:43): It's true.

Rob Collie (01:08:44): Has it been fun, Bill?

Bill Sundwall (01:08:45): On the net, yeah. Occasionally frustrating, but yes.

Rob Collie (01:08:51): Without giving too much away, how have you approached it?

Bill Sundwall (01:08:53): So it was an interesting problem. So first off, I don't have a super deep PowerApps bench, shall we say, from a project standpoint. I've messed around with it, used it in a couple of client situations before.

Bill Sundwall (01:09:07): To be fair, it's not like we were out hunting for a lot of PowerApps jobs before we brought staff on to cover that. But as far as experience, it was informed by the ERP system that I worked on at the manufacturer I was at just prior to P3 that we stood up had a fairly end-user capable ... I mean super user, definitely. But end-user capable form creation and editing system. And so, the first time I got into PowerApps, I was like okay. So they agnosticized a system for doing the thing we were doing the last time.

Bill Sundwall (01:09:41): So the same consideration, same principles come into play there, which is where we begin to unlock the first difference here. for all that Power BI is wonderful, and Power BI is definitely wonderful, we're dealing with a fundamentally different problem. The solution domain that Power BI applies to is considerably narrower than what PowerApps has as a solution domain.

Bill Sundwall (01:10:04): Fundamentally, Power BI is answering questions. Granted to everything you're saying, we want to drive that in a more verb-like direction. So answering those questions drives actions, but Power BI doesn't support the actions.

Bill Sundwall (01:10:19): So what that turns into, basically we went through, drafted the test, ran somebody through it, and immediately had to scale it back by half because being as PowerApps solves a larger solution domain, creating anything takes longer because, in the parts to the Porsche metaphor, you're at a lower level of parts in PowerApps. So first challenging thing.

Bill Sundwall (01:10:43): Second challenging thing, I think I can fairly say without giving too much away, there are a lot of things about the Power BI exam at P3, in all of its various forms, that are puzzles. Even though we're avoiding brain teasers for the sake of brain teasers, there are some stuff in there, they are outright gotchas. It's purely a matter of how are you calculating the metrics?

Rob Collie (01:11:11): Realistic puzzles. These are the kinds of things that do ... The real world is a sneaky, sneaky conniving thing, is a conniving being that is just ... Oh, it's evil.

Bill Sundwall (01:11:23): One of my favorite points to speak to in Foundations with Power BI class, is that have you ever been in the meeting where sales brings in their revenue number and marketing brings in their revenue number, and everything devolves into your spreadsheet can beat up my spreadsheet? Yeah, metrics are slippery like that.

Bill Sundwall (01:11:39): But, again, approaching the different solution domain, the slipperiness, it's less cut at dried. At of the day, it's the same loop of pressing back on a client and saying, "What do you mean by X?"

Bill Sundwall (01:11:54): I try not to go into it too much because you don't want to get all linguistics nerd on a class of Power BI students. But to me, there's something really cool about the fact that what we call the data model deep in the literature is a semantic model. We are creating meaning. We are taking data and creating meaning, but that exercise still applies in PowerApps.

Bill Sundwall (01:12:21): The ways to do that are more subtle and more sneaky, it turns out, if the first run of people we have put through this test, because sometimes I like to think the clues are out there in the open as to the questions that you should be asking and the traps that we've baked in there. But we have a lot of people who turn in a draft without going anywhere near those questions.

Rob Collie (01:12:44): Yeah, definitely. If you're listening to this and you're going to apply for a job with us, you should ask questions. This is the bonus of listening to this podcast.

Bill Sundwall (01:12:54): It's true.

Rob Collie (01:12:56): It's like when you apply for a job, use the reference code question for a plus 10% chance at passing. Yeah, so we've got a pretty solid interview now. Something you mentioned earlier that I really like is that we recognize that unlike the Power BI test, where if you're the best in the world at being a Power BI consultant, somehow factually provable to be the best in the world, you're going to breeze through it. You're just going to breeze through it. It's not going to take you very long. Whereas even if you're amazing at PowerApps, this is still going to take a bit of your time. You can't just blast through it.

Bill Sundwall (01:13:35): Doing anything that's tangible and worthwhile in PowerApps just takes more heads down time.

Rob Collie (01:13:40): That's right. That's right. So we have the radical notion that we pay people to take the interview. If we're going to have that high bar, it's only fair in return to compensate you for your time, if you're going to be taking that test. That's not true for the Power BI test, because it is a bit more of a puzzler. It's actually fun. Most people enjoy it even as they're struggling with it. Writing code is usually ... It depends upon what it is now, right?

Bill Sundwall (01:14:09): You called it a puzzler, and my NPR brain is just going to, "If you want to apply for a job at P3, write your name and address on the back of a $20 bill and send it to 1 Car Talk Plaza."

Rob Collie (01:14:19): Yeah. So that's a neat dynamic. We've done that for a few things. When we were auditioning web developers, we had the same plan. I think we also did it with designers. But it's not just about being fair. It's also about making sure people put their best foot forward.

Rob Collie (01:14:33): If you don't know what kind of quality bar is going to get you a job offer with us and you're going to put a bunch of time into it, you as a job applicant are also treating things as a funnel. If you're doing it right, you're approaching it with a wide funnel and a high bar. It's the marketplace. So why would you put a lot of time into taking a test with us if maybe the minimal job might be good enough? You've got to respect your own time.

Rob Collie (01:15:02): That makes sure that we can ask people to really treat it with professionalism. I think that works in our favor in the end. It's not just about being good humans. It's good business. It's good recruiting. So when those two things intersect, ah, that's the best.

Bill Sundwall (01:15:18): Isn't that the magic?

Rob Collie (01:15:20): Being a good human and good at business at the same time is, man, that's my fave. So, Bill, let's say there's someone out there listening right now. They've heard that it's good to ask questions. There's someone out there who's been delving in, dabbling with PowerApps, and it's really speaking to them. What would you say to them to maybe nudge them in the direction of applying and taking the challenge that you've devised?

Bill Sundwall (01:15:46): If there's one leading indicator that I think defines people that end up happy at P3, regardless of whether we're talking Power BI, PowerApps, the guys coming in on the data engineering side, all of that, it is passion for the tool set. If it completely jazzes you up that you are able to solve problems that were functionally unsolvable before, that is something that should push you in our direction.

Bill Sundwall (01:16:15): If it's enough so that investing a few hours tinkering around with a, in the end, strictly arbitrary project as part of this interview process, if the tech is fun, that's great. If you like interacting with it, that's great. But if that little investment of your time is worth the potential of unlocking going and scratching that problem solving itch as full-time job, that's the sort of people we want to see.

Rob Collie (01:16:40): All right. Well, Bill, thank you very much for everything, for being such a great person, being a great teammate, for being such a wonderful podcast guest that's willing to start and re-stop. We made a dog bite sandwich out of this podcast. There's another band name, dog bite sandwich. Backstage, front stage, we've been talking about sentences that accidentally make band names a lot in the last several hours. Well, Bill, thanks again. I'm really glad we were able to do this. I'll see you back on the ranch.

Bill Sundwall (01:17:17): See you back in real life, yeah. I will throw this out there quickly. One of my other favorite podcasts, I Only Listen to the Mountain Goats, proceeds from the idea that there's the aphorism about one should never meet their heroes. Two years in so far, I would say, at least in the case of P3, one, I just keep finding more heroes.

Rob Collie (01:17:37): That's so cool.

Bill Sundwall (01:17:38): But, yeah, this is definitely the opposite of that, which makes it the same as the podcast, despite being different than the tagline. I don't know.

Rob Collie (01:17:45): Well, I certainly appreciate it. Yeah. The best versions of these stories, where you come and you meet some high priest of old. That's what I was for a while. You come and you meet them, but then the "high" priest says, "No, no, no, no, no. I am not the high priest. I've met the high priest and they're you."

Rob Collie (01:18:04): The best version of meeting your "heroes" is discovering that actually you're the hero. For me, over and over again, that's been the best part ... One of the best parts anyway ... about this company is if people are showing up going, "Oh, I've come to the summit, the peak of it." I'm like, "Yeah, it's the peak now, now that you're here."

Bill Sundwall (01:18:28): I mean to really cap it off, we've been short on Star Wars references. This is the negative reflection. I mean negative in the photographic sense of the student has become the master.

Rob Collie (01:18:39): Yeah, yeah. That's right. That's right. Exactly right.

Bill Sundwall (01:18:41): This is one of the [inaudible 01:18:42].

Rob Collie (01:18:41): The master has become the student.

Bill Sundwall (01:18:44): Yeah.

Rob Collie (01:18:44): That's exactly right. Yeah. That's the really, really, really cool part of this thing that we've all been building. So, Bill, thanks again. Great chat.

Bill Sundwall (01:18:52): Cool.

Speaker 3 (01:18:53): Thanks for listening to the Raw Data by P3 Adaptive Podcast. Let the experts at P3 Adaptive help your business. Just go to p3adaptive.com. Have a data day.