Derek Knudsen - CTO at Alteryx

(Joel Beasley at 00:00:00) Hello, my friends. Today we are talking to Derek, the CTO of Alteryx, and we discuss why the organizational change journey is a huge withdrawal from the organizational trust bank, the importance of over-communicating when you're going through something new, and how their flagship product, Alteryx Designer, automates the process of data prep and analytics. All of this right here, right now on the Modern CTO Podcast. Here we go. This is the Modern CTO podcast.

(Joel Beasley at 00:00:39) What's last year been like for you?

(Derek at 00:00:42) You know, obviously everybody's had their challenges on the personal front, just trying to adjust to the new norm. Company-wise, it's been a pretty good adjustment for us. It hasn't been extremely difficult. We had a bias towards physical location proximity, and I think that's had to evolve like everybody else as we had to adjust that expectation. So I think work-wise, it's been really interesting. But one thing that I really appreciate about our CEO — since you and I last talked, our founder CEO retired back in October. We got a new CEO in that same time frame. He was super big on basically empowering displaced workers during COVID. And so we launched this program called Adapt Alteryx, which basically means if you're a displaced worker, you can get access to free training and this offer for free to kind of reskill. So one thing I loved about it is we didn't suffer ourselves in the process.

(Derek at 00:01:34) We tried to do much more around engaging with community to do something more meaningful, and that was something that he led. And our new CEO, Mark Anderson, has continued to kind of press forward in terms of just — we talk about the democratization of machine learning, really big on continual education, upskilling the knowledge worker, all that stuff. And a lot of folks have extra time right now, or they had extra time especially during the heart of COVID, to spend time kind of ramping up their skill sets. Some of them forced to because they had been displaced. So professionally, it's been great, a great learning experience, I think, for everybody on how to adjust when things change.

(Derek at 00:02:09) Personally, it's been the roller coaster I'm sure everybody else has been on.

(Joel Beasley at 00:02:13) Right. But you were really strong. I was reviewing the notes from our last conversation, and you do a great job too on LinkedIn of pushing out content. You know, I'll scroll through the feed, and I'll see something that you've put out there, and it just reminds me. It's like, oh, Derek's awesome.

(Joel Beasley at 00:02:29) I'll screenshot that, sent it to Adam, and I was like, we need to talk to Derek again.

(Derek at 00:02:33) Well, I'm happy to be talking to you. I'm trying to keep up with how much content you're putting out, man. Your content rhythm is amazing. I don't have the luxury of commuting to work anymore either. I always used to listen to a couple of them on the way to work, so I'm having to find time outside of that to connect.

(Derek at 00:02:47) But, man, you've interviewed a ton of people, so kudos to you.

(Joel Beasley at 00:02:52) Awesome. Awesome. Did Jake and them reach out to you and tell you that we wanted to talk about change, like organizational change management?

(Derek at 00:02:59) Yeah. Because I think I had suggested a topic a little bit too because we are going through a ton of it. You probably could guess that going from an install-based software company to trying to modernize that and how it changes the go-to-market, the whole organization. So tons of organizational change stuff we've been going through and learning lots of valuable lessons. So perfect.

(Joel Beasley at 00:03:19) You have a background in that too, right? That's what you first started out early in your career at, I think, Accenture?

(Derek at 00:03:25) Everything we did at Accenture back in the day had a change element to it, typically, because we weren't doing small things — maybe multimillion-dollar transformations, and so the change element was critical in that. So I think I learned through the school of hard knocks on how to do it, not to do it. We had a bunch of failings when I was at Accenture on not appreciating the end users enough during the change process of a new system introduction that I think I learned some valuable lessons. That really wasn't my charter, but we created what we thought was really amazing tech. You know, we satisfied all the user requirements, we were really architecturally sound, we rolled the solution out to end users, and they hated it because they weren't engaged correctly in the change process. They didn't understand.

(Derek at 00:04:09) They said the business didn't understand their actual use cases. And so we had to revolt against this amazing technical system we created because it was solving the wrong problems. And I saw a few of those during my time at Accenture that I was like, well, you better get smart on this because you can deny the fact that that's not my area. I'm the tech guy. Like, just tell me what to build and I'll build it. That's up to you all. But everybody wants to create product that their customers love. And when you're seeing that we're failing on understanding the user, understanding the adoption curve, and so the product's not getting used, the system's not getting used, that's hard. And so I started getting super engaged in all the John Kotter books on leading change and trying to really understand the change journey, so I could make sure when I was making technical decisions, I was asking those questions of my product counterparts on, now, what have you talked about in terms of user scenarios? What problems are they trying to solve versus just saying, hey.

(Derek at 00:05:02) I'm going to come up with my own sets of requirements, and then hoping when you throw it up against the wall, the system sticks. Just seeing that fail too much.

(Joel Beasley at 00:05:09) Were there tools — like, the reason why I'm interested in this is it sounded super ambiguous, like organizational change management. You know, these categories, I always try to pluck them apart and figure out what are they actually doing. You know, often you go to websites for these tools or these topics, and it's just not super clear for a person who's outside the industry. Right? And then I was talking with Nick about it who's like a strategy officer at this company called Biz Design.

(Joel Beasley at 00:05:35) And they make tools, but they have part consultancy, but they have part tools for mapping these processes and understanding the organizations. Did these type of tools exist back in the day, or was it too early for them when you were at Accenture?

(Derek at 00:05:49) Oh, yeah. They didn't exist. It was, you know, Visio diagrams and Word documents and a bunch of disparate data sources trying to manage the change journey. It's gotten better. You know, I did take a look at that software because I saw it come up. Interesting software. Boy, it solves a — it's a very broad swath of what that software is supposedly able to do in terms of the enterprise architectural components, the technical change management, process change management. It's got the ambitions of doing a ton. That stuff didn't exist back in the day at all. I mean, even architectural governance was on paper.

(Derek at 00:06:24) Because I grew up in that enterprise architecture space and did tons of advisory there, and that was a really difficult spot to drive change into because it was an underappreciated need. So on the technical side, I think getting some more tools around this is super good.

(Joel Beasley at 00:06:40) At what point — because you know this area, and I really don't. So I'm a founder. My teams and my companies have been less than 100 people. Right? Like, before I've sold them or exited or not been a part of them anymore.

(Joel Beasley at 00:06:53) At what point does this come up? Because I know how I structure teams and I, you know, move things around to other CTOs. They'll have, you know, maybe a couple hundred engineers and they'll structure teams. At what point does it extract into enterprise architecture?

(Derek at 00:07:09) Well, I mean, most folks are asking for enterprise architecture after the fact. Right? Where they recognize they've got architectural drift, and they're trying to find a way of bringing that stuff back together. You know, back in the day, you've been around this industry long enough, there was TOGAF and Zachman and all these different models. I got my TOGAF certification, thought it was going to be this amazing thing. We struggled with adoption of just an appreciation for why enterprise architecture was important. I think to a lot of our customers at the time, it felt like an academic exercise. It stuck really well in the Fed space or the highly regulated space. Tons of traction with Fed equivalents of TOGAF and such like that. We struggled to make it stick really in the commercial side of things, despite numerous attempts, me specifically.

(Derek at 00:07:55) I think we struggled to get people to appreciate the value of thinking about that technical change journey more holistically. I'm not sure why, to be honest with you. You know, because I've learned those lessons at Alteryx, I'm highly conscious of it, or before Alteryx. I'm really conscious of trying my best to communicate the change journey around the technology and not just being a product-led thing to make sure that we're clear on when we make technical decisions, why we make them, what the technical trade-offs are, what it'll enable and disable, kind of coming in eyes wide open with technical debt you're going to incur with those decisions. All that stuff becomes really important, especially in the model that we're in right now, which is in this transformational stage of going from a set of install-based products that our customers absolutely love to trying to modernize and be cloud-first and more SaaS-oriented.

(Derek at 00:08:42) That's a huge journey, huge transformation. So you talk about organizational change management, think about that. I'm a developer who ships software to customers, touched once a quarter. Right? So I have a quarter to develop it. So sense of urgency was low. Kind of organizational commitment to lean cycle times, having really quick time to value on getting stuff into the product, it suffered. But why would I do that if I'm not shipping it for three months? Like, why do I need to have this massive sense of urgency on things like feature cycle times, et cetera? So the tech's changing, but for us, it's a huge cultural shift on you've got to think like a SaaS company.

(Derek at 00:09:20) You have to think like, hey, if I build it, I own it. Where when you install a piece of software, you're like, you know, when I build it, the customer owns it. I don't have to really think about that anymore. Well, that dynamic's changing. So organizationally, for us, it's been a pretty huge shift.

(Joel Beasley at 00:09:34) It's amazing how software as a whole has shifted so much that back in the day, you would ship the disc or the CDs. And, man, getting data back on what's happening with that would be — I can't even imagine it because I didn't really grow up in that world. Like, I grew up using those tools, but by the time I was owning businesses and such, I grew up in a SaaS-first world. So, like, for Adobe making those pivots.

(Joel Beasley at 00:09:59) And I got to talk to Marty Kagan, like, three or four years ago about the billion-dollar Adobe pivot when they switched to a SaaS model. And I can only imagine what it would have been like to ship the physical disk out and wonder what's going on because you don't have a constant internet connection.

(Derek at 00:10:17) Yeah. Well, our new CTO is from Adobe. So he brings all of those — Suresh Patel. He's amazing, amazing leader. He brings all that to Alteryx, all those lessons learned, good and bad, from his time at Adobe.

(Derek at 00:10:30) And he's taken a ton of products either from on-premise into the cloud, from on-premise to a managed service offering, left it on-premise, been cloud-first. So he's seen it all. He brings a lot of that knowledge and experience. Adobe's done a great job on that. I've talked to a bunch of my peers in industry around this transformational journey and how to embark in it. And the one that everybody continues to point to as being kind of a beacon of how to do it right is Adobe. I think others have struggled with that transformation. A ton of them have. But Adobe is one that seems to have done it right. So we're really fortunate to have somebody like Suresh here now who brings all those learnings with him.

(Joel Beasley at 00:11:06) Smart move. When you found him or recruiting him in that whole process, were you directly involved with that or is that more of the CEO? How did you figure out that you all play well together?

(Derek at 00:11:16) Yeah. I was involved for sure. I mean, obviously as much as the CEO, it was a really, really — maybe you could argue the most important hire you're going to make given what we're going through. Right? I was involved in it, interviewed with him multiple times. I think both of us were very eager to make sure there was comfort there in the working relationship between us. You know, he quickly rested any concerns that I had, just given what he comes from and just his engagement style is quite amazing. I've learned a ton from him on a daily basis. So it's great to have a leader that you can just learn on a daily basis how to think about problems differently, kind of learn from the things that he's experienced in the past. And then what was great about Suresh is, you know, he joined probably a month ago. You know, two or three weeks ahead of that, we set a learning agenda for him that we met on a daily basis, and we spent an hour just spinning him up on stuff. So when he landed, his feet were already in motion. That gave him a ton of opportunity to make an impact quickly, which given the transformation time we're going through, we needed a leader that can really drive velocity into the transformation, not be, hey.

(Derek at 00:12:16) Talk to me in six months, and I'll have a point of view. Like, we were like, no. We need to continue to move fast. We don't have six months to let you find the bathroom, for lack of a better term, and then decide what you want to do. Like, we've got to go. He's been just fantastic on that. He's been here a month, and his knowledge in a month is pretty staggering. So good thing for us.

(Joel Beasley at 00:12:35) Well, I've been watching your growth, and it's been phenomenal. It's been insane. You guys are just straight up.

(Derek at 00:12:41) Yeah. I think — I mean, COVID was interesting for us. I think, you know, Alteryx's go-to-market was always structured around selling into the business. Take — you know, we go sell to a knowledge worker in the business. Our Alteryx Designer has got amazing customer sat. You know, our NPS for the product is in the high fifties. That's insane for software. Most software's in the teens, if it's lucky. So customers love the product. You give the product to a customer, they start using it, get a ton of value from it, they tell their friends across the cube from them, they start using it.

(Derek at 00:13:13) Like, it was very much a department-led, kind of the traditional Tableau go-to-market motion. Well, when you get into COVID, right, that don't happen anymore in COVID. Right? Everything goes through central procurement, you know, budgets are tightening up. So we've had to make a pretty hard pivot on to recognizing how to sell into a much more procurement-led process than selling into the individual departments.

(Derek at 00:13:36) Really important pivot for us to make, and COVID was definitely a forcing function for us in that. But then you take into account some of the tech transformation stuff we're going for. I think it accelerated the need to get the tech transformation done sooner as well.

(Joel Beasley at 00:13:50) Yeah. That makes complete sense because it's a restructuring and reeducation of your sales team on how to reach out, what the sales motions look like, the timelines, all of that. Yeah. That's — you guys handled it well, though. You adapted pretty quickly.

(Derek at 00:14:06) Yeah. Maybe — I think we've learned lessons along the way, for sure. I think learning how to sell into IT — I mean, I was a CIO at one point. I know what it's like to sell into an IT organization. I think having more appreciation for the persona of an IT decision maker, a technical decision maker, we've had to learn how to do that better.

(Derek at 00:14:25) You know, how do you sell via the InfoSec office, office of the CSO, when they have, you know, higher concerns? The bar is higher into those organizations than it would be selling into a department where they can make business justifications for software procurement. You know, that dynamic changes. So we spent a ton of time investing in the quality of our software, the security around our software to make that friction less for our go-to-market. So that's been a really valuable thing for us as well.

(Joel Beasley at 00:14:52) I want to talk about security in a minute, but first, let's take a moment. I want to plug the software. What does it do? Why do people buy it?

(Derek at 00:14:59) Yeah. So, you know, we have a variety of parts of our business. Our flagship product is Alteryx Designer. It basically takes that analytic lifecycle around acquiring data, cleansing the data, preparing it, cleansing it for an analytical outcome, and then generating that analytical outcome. It does an amazing job of that. I mean, you don't get a distribution like ours in customer sat or NPS in the high fifties if it doesn't do a really, really good job of that.

(Derek at 00:15:29) What you'll see when you read about Alteryx Designer use cases is time to value. So what used to take a person working at a customer a hundred hours in a given month to go do now takes them minutes to go do. And so that time to value means they get back a huge percentage of their time to do higher value activities. So we've seen a lot of really amazing stories around people who use our products getting, you know, kind of career acceleration because of Alteryx. So I was an IC working on a discrete problem. I found Alteryx. Now I'm solving all these amazing problems in a very quick fashion. I'm getting career acceleration because they're seeing how much value I'm bringing to the organization. That's why you get this almost fervent, dedicated persona around designers because it's changed lives, for lack of a better term. That's what we love about it, the democratization of that analytics journey.

(Derek at 00:16:22) So it takes it out of the hands of the advanced analyst or the statistician to the everyday information worker, and they can create amazing outcomes, whether that be a, you know, last mile visualization or a predictive analytics model or just generate an insight or automate a data-driven process. Like, the tool is now in their hands and they can go do it. And the reason it's successful is because it's easy to use. It's easy to understand. That was what drives the NPS. That has been a huge anchor for our organization. It gives us the—it buys us the ability to make investments to transform the tech because we have such a strong foundational product that customers love. That buys us kind of that collateral with customers to say, we love your products, so we're willing to watch you all go through this transformation. Very thankful we have the Designer product kind of as our flagship piece of software. And the fit in the industry right now is really relevant. I mean, there's so much, you know, data analytics happening. It's really one of the hot parts of the technology industry. You know, the timing for our software is really, really good.

(Joel Beasley at 00:17:25) So I don't want to put you on the spot, but so for us being a small company, like, 15 people, like, would there be a use case for us to use your tool? Or are we too small, or at what point would a company be using it?

(Derek at 00:17:39) Well, we have a ton of our customers are super small, super small deployments. So you got long tail of curve. We got a bunch of customers that are, you know, in that one to 10 seats range. So you don't have to be a massive enterprise to get value out of it. And that could be, you know, a small footprint within a large enterprise. It could be a small footprint within a smaller enterprise. If you're ingesting a lot of data and needing to drive insights, it's a great tool for that. If you're ingesting a lot of data and having to, you know, automate process around that data, it's a fantastic tool for that. If you're data limited, you don't get a lot of data, then, yeah, I mean, it's a data tool, so that probably isn't a great fit. But I'm thinking of your, you know, your podcast or your media-centric space. You could ingest social media feeds and drive, you know, all sorts of different NLP type outcomes from that, customer sentiment outcomes from that. That could power some content creation around that. Software does an amazing job with that. So tons of use cases for you all. So we should at least have you and I look at that offline. I'd love to see you guys get your hands on it and take a look at it.

(Joel Beasley at 00:18:42) Well, we did something really interesting. So the people who were, you know, paying, they wanted to see metrics on it, but we've got listeners on YouTube, people watching the episodes on LinkedIn, from the actual podcast, you know, network, all of these different sources. And so what we did was we built a custom analytics tool and API with all the different vendors, and then we put the links. So we make a customer analytics screen, and we say, okay. Here's when we posted these three clips. Here's the newsletter posting, and we suck in all of the data. So it's like one aggregate reporting function. But because of that, I mean, we know all the people that like the stuff, all the people that—we have so much data on that side. And then on the other side, we have, you know, here's who we're reaching out to. We have all of this data. We do nothing with it, and we're reaching out to, like, you know, thousands of people a week and setting meetings. And I'm sure there's insights in all of this sales data and all of our CRM, things like that about maybe locations that do better, if we plugged in, like, our sales closed data and, you know, maybe help us figure out what areas to focus on or what—there's all of this stuff.

(Derek at 00:19:54) Yeah. I get it. I get it. That's a lot of our DNA is around, if you're a Walmart of the world, you use us to identify where to put a store. Right? But we've done a lot more on the digital front as well, taking these digital inputs in and understanding where to do ad placements and such like that. We've got a lot of predictive analytics tools that are pretty accessible in the tool as well. So I think it'd be a really interesting thing at least to get you guys a copy of the software, have you take a look at it, help you run through some use cases and see if it would make some sense for you. I have a hard time believing it wouldn't given what you told me around the amount of data and types of data you ingest and what you're trying to accomplish with that data. It feels like it would be a really great fit for you all.

(Joel Beasley at 00:20:37) Yeah. And I know, like, I typically advise people, like, have your outcome first. Don't just play with tools. But at the same time, I get paid to play with tools. Yeah. You know? Yeah. I need to know what's going on out there because it gives me exciting stuff to talk about, and then I can connect people because people share all the time. You know? We're doing hundreds of these a year, and people will tell me that they're having problems in this area or that area, whether it's on the podcast or off the podcast. And me being able to connect people just adds value to them. Right? So yeah. I want to talk about security. Right? So I did this interview with a guy named Jason who's the CISO at HITRUST. I didn't realize how big of a company HITRUST was because so many people have since, like, mentioned HITRUST to me. And I was like, oh, okay. So it's very large company in that area. But we were talking about security. We were talking about, you know, organizational strategy. Then you earlier were asking me about, or you were bringing up. You know, we had COVID change. That made a lot of changes. That's changed a lot of people's security. You've got, you know, this new chief product officer. You've got this awesome team over there growing and constantly increasing the size of the company. How do you guys handle security? Do you have a special person for this? Do you have a CISO? How do you think about security?

(Derek at 00:21:55) Yeah. We, when I first got to Alteryx, I was handed too much. It was—I own security. I own IT. I own engineering. I own licensing at one point. I own the EPMO function for kind of cross-organizational initiatives. One area that I said, hey. I don't have a ton of deep pedigree in. That is a very bespoke area of security. Well, I made a hard pitch back then to say we need to bring in—given we're an analytics company, we look at data. We need to bring in somebody who's very specific to a CISO type role that really drives that functionality. So, when was it? A year ago or so? We brought in Billy Spears. He's our CISO. Fantastic. Tons of industry pedigree around how to do that role in our specific industry space, in our kind of technology domain. He's been fantastic. We've had to work on the relationship between—you know, he obviously will own the corporate security governance, BCP side of things. Right? But then there's the injection of that into the product security and how do we think about product security much more proactively. We talk about moving things like quality left in the process. Right? How do I address quality concerns early in the process? Because as you know, being in software, the later you deal with the quality issue, the more expensive it is to deal with. Order of magnitude worse for security issues. So our intent was, how do we work with our CSO now to move more of those security concerns farther left? We've done a good job of getting kind of static code analysis, dynamic code analysis, all that type of stuff into our code base. He's been driving a ton of that effort to make sure that security is a first-class citizen in how we create software. A lot of times people see security in the product space as being something that slows you down. The one thing I really appreciate about Billy as a partner is he doesn't want to be that person that's saying, hey. Hey. We can't get product to market fast enough because security is this wall between ourselves and our customer. He is highly conscious of making sure that he's not an inhibitor to the process. And that has been a great partnership to have. So now when we talk to customers, he just wrote a white paper, I think, we published around our stance with security. But now we're leading more and more with our stance on security, how we test. We are going after some FedRAMP certification for one of our products as well. We're beginning to lead with that as a differentiator, especially as we move our technology to the cloud. We know that that's an important decision criteria for a lot of technical decision makers is how security first are you when you create your software, when you manage and operate your software. You gotta do a much better job of that, especially going forward. Now that he's in seat, we can do that. When it was me in seat with that, I was doing my best to do it, but I didn't bring his pedigree into the role. And that's just an acknowledgment that, hey. This is not an area of expertise for me. I need to bring in somebody who's an amazing leader in this space to really drive this function, and that's what he's done, and he's been a really good partner to work with. Maybe that's an exception. I'm sure a lot of people don't have that relationship with our CISO. Maybe it is more adversarial. In my case, I've got a great relationship with our CSO, and it's mutually beneficial.

(Joel Beasley at 00:25:02) Yeah. I'm spoiled because I get to talk to all these awesome leaders, and then I get to pick up on the trends and the great CISOs. I don't know a ton about them, but I know that one of the brilliant talking points in that space are the CISOs that are enabling the organization and not being a barrier. And so, uh, that sounds like you've got a really awesome person then.

(Derek at 00:25:27) Yeah. For sure. And I think that was something that I think CIOs face too. If you remember even ten, fifteen years ago, CIOs always had the reputation for, you know, being an inhibitor to business velocity. Their charter changed. They became very focused on, you know, being business enabling and value accretive versus just being a cost center. I'm guessing a lot of CISOs have taken that same mentality when it comes to their role. Nobody wants to be an organizational inhibitor and especially the pace of business that we have today and in tech especially. I'm sure that they've kind of taken that charter forward as well in their roles, and that's good. Because if we had CISOs or, you know, these kind of ancillary organizations, legal, data privacy and protection, and they were inhibitors, like, that's not good for our customers. And in the end, it's not good for our business either, and I think it's been good that these folks are taking a much more enabling, you know, perspective on their roles.

(Joel Beasley at 00:26:18) Yeah. I agree. We all kind of snapped to it too because what we saw was all of these businesses that were household names, like, go out of business because they didn't move fast and because they didn't, you know, they weren't nimble and didn't watch the market. And so that kind of woke everybody up to instead of being the command and control version of the, you know, seventies, eighties management, you know, everyone turned into, let's be a team player and just put the best ideas forward and bring value to the marketplace, and that becoming a trend is an awesome thing.

(Derek at 00:26:53) Yeah. Everybody benefits from that.

(Joel Beasley at 00:26:56) So what are you—I want to know what you're learning, like, right now as a leader.

(Derek at 00:27:01) We talked about the tech transformation we're going through. You know, one thing I love about this is I'm getting to, I will call it, further modernize my own personal skill set. I love that. So I get to learn much more around new tech that I wasn't as familiar with, getting, you know, AWS certified, doing that type of stuff, even as a CTO, because I want to understand where we're going at a much more intimate level than, you know, a lot of times, a lot of large organization CTOs, which I'm fortunate enough to be, you don't get time to spend in the deep technical detail. But this transformation is requiring me to. So I'm benefiting from this transformation. So learning more around those technologies has been great. Taking an organization through the change has been great. We restructured how we approach quality in our software. One way we used to do that legacy-wise is we were very manual testing oriented. You gotta remember, Alteryx has been around a long time. Our original version of Designer hit the market early 2000s, so it's been in the market for, like, fifteen years. Back then, manual testing was kind of the norm or much more the norm. And so the institutional way of thinking around the value of automation, et cetera, it wasn't as acute because our deployment base was small. We were a small organization. So manual testing wasn't that heavy of a lift. Well, you know, now we're in excess of 400-ish engineers. The organization's really large. That surface area of testing is gigantic. If you're still doing manual regression testing, that—what does that do to your cycle time? It kills it. Right? You can't cycle stuff out. Your quality suffers. When I got here, I was trying to institute change towards more automation. You know, what I ended up doing was, you know, serving as a forcing function because it wasn't going fast enough. It wasn't preparing us for that next stage of growth. Was really taking us through what I call a burning the boats moment. Right? We kind of burned the boats on being able to go back and rely on manual testers. You know, we had to make some hard people choices around either reskilling folks or moving folks out around that decision. But it had to be a thing to say, hey. There's nobody to turn to to do your testing for you anymore. It's you. And you're a developer, and so you're going to lead with solving things through development, which is through automation.

(Derek at 00:29:10) So we had to work to take the organization through that change. A lot of my time is really focused on executing the change and having all sorts of successes and failures and trying to do that in different aspects of the business, but learning about how to do it better. Because I have a lot of history in the change space. You know, history doesn't equate to success, right?

(Derek at 00:29:29) I think everybody says, "Hey, I've done it before. I kind of know how to do it." Every organization's different. Every culture's different.

(Derek at 00:29:34) And understanding that you're gonna make a ton of mistakes in the change process and learning from those and refining your style to fit the culture dynamic that you have at this company has been a huge learning experience for me. Because doing a change in some organizations, like when we did some of the big stuff at the Gates Foundation many years ago, very different process than doing it at Alteryx, right? Very different business, very different culture. One organizational change process is not working for both. You gotta refine that process. I've had to go through some hard knocks on the Alteryx side around saying, "Hey, that doesn't work here. You gotta rethink how you approach that change process." And that's been really fun for me. I love that dynamic of it.

(Derek at 00:30:14) The tech's great. Love the tech, learning a ton. Spend all my mornings learning tech. That's fantastic. Where I really get my juices is taking an organization through a transformation. This is a great opportunity to do that. And this specific opportunity has been a huge lesson for me in terms of how to think about these harder changes, and I've had all sorts of goods and bads that have come from that.

(Joel Beasley at 00:30:34) Yeah. What sort of people insights did you get from executing this change from no manual testing?

(Derek at 00:30:43) Well, one thing because of the people dynamic is we weren't able to really bring in a lot of our leaders to the conversation because there was a difficult people dynamic to think through. And we didn't want the knowledge of some of those harder decisions to be more distributed. So we had to leave some leaders out. So some folks felt that the change was done to them versus with them. So I think we'd rethink how to maybe do that in the future in the right way.

(Derek at 00:31:07) That was something that I learned especially in the process. Didn't identify as many quick wins that we probably needed to, at least quick wins that were really tangible to the person that was impacted. They felt tangible to us, but that person that was impacted, the developer now who has to completely own their testing, they were kind of left a little bit holding the bag on understanding how to get to success really quickly. So we had to pivot and really think about creating more explicit test strategies for specific domains, products, and providing interim milestones that felt like success that were very achievable to give them very distinct ideas of, "Hey, if you can do X by Y, that's success. You don't need to get to the end state in the next X amount of months. Let's incrementally work our way to where we want to be." We need to be much more intentional around those "What's in it for me?" type moments for the developers, and I don't think we did as good of a job at that as we probably could have.

(Joel Beasley at 00:32:03) But you learned.

(Derek at 00:32:04) That's the beauty of it, right? You always learn. You always hope that you've earned enough organizational trust before those events. You know you're gonna take a withdrawal, right? Any change journey is a withdrawal. And the hope is you've made enough deposits that you're not emptying the bank account by doing it. We felt like maybe we were close enough that we could take that withdrawal and have a learning. We had thoroughly admitted this isn't a fully baked process, but if we don't burn the boats, we may never change. We have to burn the boats. So we burned the boats, and it hurt a lot more than we had liked it to. But sometimes you have to really force the change. I'm hugely confident that by kind of September of this year—it'll be roughly 10 or so months—people will be like, "Why did we ever work that way? That way made no sense." But that doesn't mean that the 10 months before that were easy. And I think just having an appreciation and acknowledgment for "This isn't easy. It's gonna suck at times. It's okay. We're gonna get through it, and we're gonna get to a really good end state" and being consistent with that message. I can always do a better job with that. I try to be really consistent with that. "Let's go, let's go. It's all good. We're gonna get there. Have faith" message. But people have to have faith still, and you hope that, again, your political capital is high enough or your relationship capital that people will continue to believe it.

(Joel Beasley at 00:33:20) And you can feel it too when you go into organizations, whether you're a guest there or it's your own organization. You can kind of feel the sentiment and how much—I love that you use that. What did you call it, organizational capital? Or how did you—

(Derek at 00:33:35) Yeah, I think it's relationship capital, a lot of it, right? You gotta have that bank account. And if you don't have the bank account, you can't ask for withdrawals. And I tried to work to put enough stuff in the bank to make a withdrawal ask. I'm sure some folks felt it was a very small withdrawal. Some folks felt it was a huge impact. Probably I took a huge personal kind of withdrawal from that bank account. But you have to make the deposits before you can do these types of things, and that's something that I hope that we did a good enough job on this one. But we'll continue to learn over the next handful of months as we continue the change journey.

(Joel Beasley at 00:34:09) Over-communicating—do you find that important, like constantly reiterating and discussing the change and why you're doing it and where we wanna go?

(Derek at 00:34:18) Yeah. This is an area that I was very clear on when I announced it. It was my desire to drive it. My leaders were very much on board with wanting to do it. So at the end of the day, what I love about the leadership team I'm fortunate enough to lead is they raised their hands that this is our decision. This isn't Derek's. It's ours. We need to change, which I love. And then we've been trying to be really consistent. We took one of my leaders and we said, "Hey, this is a smaller thing to focus on. You need 100% focus on this one activity because it's that important." And a big part of his communication is continually talking about where we are on the change journey. We were on a call this morning with our delivery leadership chapter talking about where we're at in that journey. He's really, really consistent with this communication. I think that helps. It helps to hear and see that. We're in the kind of trough of disillusionment in terms of where we're at with the change. But it feels like we're on the upslope coming out of it. But you're gonna have that trough as you're going through a change. You just have to be okay and recognize that, no, there's upside coming out of that. He's helped mitigate some of the depth of the trough, let's say.

(Joel Beasley at 00:35:24) So you've got this new CPO. You've got an awesome CISO. The team is growing. You guys are building amazing product. You're switching from—or you switched. You burned the boats from manual testing to now you're doing automation-based testing, which was a big change, you know, for me when I first saw other developers. Because I was self-taught, when I saw some of the best developers, I started buying their books and watching their videos. I saw that they were all testing, and me making that switch was personally difficult. And it didn't happen the first time. I tried to switch to a test-driven-first development process probably about three or four times before I got frustrated enough to where one day I said, "I'm just gonna do it," right? It's like working out. You try so many times and you're just like, "I'm gonna do it. I'm gonna commit." And but after you do that, it's just like, "Why did I ever?" And now every time I see somebody that's writing code that's not test-driven, I wanna sell them on the life change that they'll experience if they do TDD.

(Derek at 00:36:23) Well, we've talked about that internally around, "Hey, if you wanna refactor, what do you do?" If you don't have tests around it, you can't refactor. You don't know if—we wanted to do some large transformational stuff around our engine. Well, that's a core piece of IP and it's got 20 years of legacy in it. If it doesn't have good sets of automation test cases around it and you start entering the refactoring exercise, how do you know you're creating the same outcomes? With the testing surface being enormous, you can't, if it's manual.

(Joel Beasley at 00:36:53) Yeah. The answer is you don't.

(Derek at 00:36:56) Yeah, exactly.

(Joel Beasley at 00:36:56) And that's a horrible answer.

(Derek at 00:36:58) And I think that's what we did is we had a couple of folks have really good success saying, "I did automation, and I had to go through a refactoring for whatever reason. And I had something to fall back on that I knew that if my test passed, it would satisfy the same outcome expectations that it needed to solve—it solved yesterday, will be solved tomorrow." And then people are like, "Oh, I get it now. It's not just because I'm checking a quality box. It's because I, as a developer, get value in terms of freedom to do that type of refactoring when I need to, knowing that I've got this backstop of automation to turn to."

(Joel Beasley at 00:37:30) Oh, man. Dude, whenever I meet really interesting people like you, I'm always curious, what was their childhood like?

(Derek at 00:37:39) Well, that's a tough one. I'm trying to relate it to our conversation.

(Joel Beasley at 00:37:44) Oh, you don't have to. You are the conversation. What was life like as—what's eight-year-old Derek like?

(Derek at 00:37:50) Well, I was mother-led, let's put it that way. Kind of got a higher bar, higher expectation. I think that shaped a lot of who—the way I think about the problems I solve, for sure. Yeah. You know, parents didn't have a ton when I grew up, so I think you—they had a high expectation of us despite not having a lot, and I think that really resonated with me. You have a huge appreciation for that kind of post-process with your parents when you're out of school and you're kind of on your own life and you look back and say, "I wouldn't be—I'm sure everybody feels like this who've had really positive influences as parents—I wouldn't be who I am if it wasn't for kind of the mindset you put me in as I was growing up, right? Where you made sacrifices so we would have an opportunity, and I saw that. And so I tried to take advantage of those opportunities because you were making those sacrifices." I think that kind of created this ever-hungry element that I think I'd like to think I continue to have, where I'm always eager to learn and learn from bad experiences and progress my own skill set and do more, experience new things. I think that's a huge part of it. You got an opportunity to do that as a kid because your parents made those sacrifices. Now I gotta go and take advantage of that.

(Joel Beasley at 00:39:01) Dude, I love that. I've never heard that term. Did you say mother-led?

(Derek at 00:39:05) Yeah.

(Joel Beasley at 00:39:07) It hits home perfectly. So we were poor, but we shortly got out of that. That was early on in my years, and then my dad built some technology. And so we definitely got to middle class, right, which was a huge change for us. Looking back as a kid, you're going through it. You really don't know or care. You're just like, "This is life, and I go to school." You're a kid. But the one trait that stood out to me when you said mother-led was, while we were like that and the rest of our family was like that as well—we were all pretty poor, the uncles and aunts and everything—my mom was always putting us up, always had just this huge expectation and never wanted help, right? She wanted to be seen that we were doing well and everything like that. And she always looked forward like that. And so when you said mother-led, I was like, "Oh, this must be a trend. Other moms must have this personality where they wanna push things forward."

(Derek at 00:40:11) Yeah. You know, whether it's mother-led or father-led, just having somebody you can look to that role models that expectation—I joke with my wife all the time. She'll say, "I'll never be your mom. I'll never be your mom," right? Because my mom—a lot of folks put their mother on a pedestal or your father. Whatever association you had, you put them on a pedestal in terms of what they could do in a day. The raw horsepower they had, it was amazing. My mom was off the charts on horsepower. She'd wake up at 5 a.m. and she'd be the last to sleep at 1 a.m., and she wouldn't sleep much, and she would grind. There was never sitting around on the couch doing nothing, reading a book. It was go, go, go, go. That wired a ton of who I am. I think that's a frustration that I think the people I work with sometimes, and I know my family have, is like, "You never stop." And that's not a good thing, I would say, at times. You need to learn to take peace and take a breath and smell the roses a little bit. But that's definitely from my mother because that's the way she was wired. It was always, "What's next? What's next? What's next?"

(Joel Beasley at 00:41:09) Well, dude, it's in your DNA.

(Derek at 00:41:11) Yeah. I would say it's sometimes unhealthily in my DNA. We maybe overemphasize that, but at least I self-recognize maybe. Trying to get better about that.

(Joel Beasley at 00:41:20) I can't remember. Do you have kids?

(Derek at 00:41:22) I do. I've got two kids, one 11 and one 13.

(Joel Beasley at 00:41:25) How are they doing?

(Derek at 00:41:26) They're doing fantastic. They're back at school full-time, which I think a lot of—sounds like a lot of states are getting back in school. We live in Colorado, so normality is returning, I think, to some extent. We live next to a speedway. I drove by it the other day, and they actually had a bunch of people in the stands, which was an anomaly to see people attending an event. You know, sports games are opening up, attended well. But my kids are doing great. Kids are resilient in stuff like this. They learned to adjust from working from home. My daughter was extremely proficient. My son drifted at times—he'd try and sneak in TV or goof off and didn't really focus on his studies. But I think they all learned how to be successful, just like we all did. They kind of learned how to adjust because we're humans and that's what we do. They learned how to adjust to revised expectation, and they both thrived with it. I do think they're both excited to be back at school full-time. It's hard as a kid not having that social interaction. I can only imagine if you're a senior in high school or something and it's your last year and you don't get to really soak in your senior moment. That would be terrible. Luckily, my kids are 11 and 13, and so that wasn't what we faced. But I can only imagine how hard that would be for people that are in these inflection points in their life, like graduating high school and not getting a chance to enjoy that, or graduating college and not getting a chance to enjoy that last year in person. That had to be a tough one.

(Joel Beasley at 00:42:47) Yeah. But I think it's gonna create some stronger people, though.

(Derek at 00:42:51) Oh, for sure.

(Joel Beasley at 00:42:52) Yeah.

(Derek at 00:42:52) I wouldn't doubt that at all. I think the dynamic of education, I think with COVID especially, I'm curious to see how that plays out over the next handful of years. You know, virtual learning was a forcing function now because of COVID. So now what's it gonna look like? What's the traditional institutional education process gonna look like?

(Derek at 00:43:09) It's gonna be... it's already gotten pressure. COVID probably provided more pressure to it. So how do the institutions adjust to it? Not sure what that eventually is gonna look like, but you have to know that they're gonna have to make some adjustments.

(Joel Beasley at 00:43:22) I can't remember exactly who it is, but I saw a LinkedIn post. Maybe the CTO of Ohio State University or just some big university, right, where they had like 20,000 plus students. And I saw that within like two or three weeks of COVID, they had to... their entire tech department had to put the entire university from in-person classes to online classes. And he wrote this blog article about it, but at the time he was getting so much... like it was so chaotic in his life that in the moment of the incident, it's like when everything takes off, that's not when Harry from Zoom is gonna come on. He's gonna come on like four months later and talk about the whole process because everyone was in this all-hands exercise.

(Joel Beasley at 00:44:03) So I'm still waiting on that interview, but I was curious to see how they turned and transitioned within like two or three weeks an entire university system from in-person to online because I think that'd be a fascinating story.

(Derek at 00:44:11) Oh.

(Joel Beasley at 00:44:11) System from in-person to online because I think that'd be a fascinating story.

(Derek at 00:44:16) Yeah. I would make sure I listen to that one. That I would love to hear that story. It wasn't that hard for, well, like a lot of traditional corporate corporations who had at least some footprint in virtual meeting, especially if you're a global company for sure. But I don't think it was probably more the exception, not the norm, to not have it. But for an educational, you know, kind of an institutional education organization where maybe a small portion of your classes are a little more virtual or you can attend virtually, now you're going full virtual to maintain some level of momentum around your institution, that had to be a super hard problem to solve and an expensive one.

(Derek at 00:44:52) Right? You don't plan for that in your budget. Could you imagine the Zoom or Teams or what have you cost of that endeavor? That would be substantial.

(Joel Beasley at 00:45:02) Yeah. It's unfactored in their business model. Right? And they're still having their cost from their buildings and all of that. Their in-person business model. Right? And then on top of that, one of the big ones that was pretty transparent to me, my wife had finished her last class. Like she had three or four weeks of her last credit or class right when COVID had started. And I remember her teacher couldn't figure out how to use Zoom because we're professionals in meetings. We've done a Zoom meeting before. Even before the pandemic, at some point people have done some form of video meeting.

(Joel Beasley at 00:45:34) You have a general idea of how it goes, but these teachers have literally never done one before. And so they were trying to figure out, do we live stream on YouTube? Do we do a Zoom call? How does the Zoom features work? How do I engage with my class like this?

(Joel Beasley at 00:45:49) So it was a hard education thing to teach the teachers how to do this and then, yeah, the business model. So I'm super interested to figure out how they did that at a major university. One thing I wanted to touch with you on though is you personally. Like what are you really excited about in the future? Like AI, self-flying cars, self-driving cars? Like what are you geeking out about?

(Derek at 00:46:15) That's a good question. On the tech side, you know, given the space we're in, I spend a lot more time thinking about the space we're in. The explainable AI stuff is really interesting. We talk a lot about that internally. Trying to get quicker, especially as you democratize data science, how do you generate quicker time to insight?

(Derek at 00:46:35) You know, part of that is access of the data. Right? And companies like Snowflake are trying to solve that. Right? Hopefully that is a more solvable problem in the near term.

(Derek at 00:46:45) Then there's the tooling. Alteryx plays really well in the tooling side of it. And there's education to that where we're spending a lot of our time on elevating the capability of the knowledge worker. But then there's the time to insights piece. It's, yeah, okay, I know how to manually go through the 15 steps to generate an insight. I don't have time to do 15 steps. So how do I get to an insight quicker? So we are seeing in our industry an emergence of all these auto solutions. Right?

(Derek at 00:47:11) You know, whether it's AutoML, whether it's Auto Insights, or I'm presenting data into a machine that's giving me prescriptive analytic outcomes. Well, based on this data, here's what you should go do. You're seeing a lot more squishing of the time to value happening. But then what comes out of that is, well, that's great, but then how do I trust the data? It's one thing when something says, hey, here's a Netflix movie recommendation. That's kind of a low-risk thing to engage in if it's wrong. If it's an underwriting activity or insurability decision, you can't just fire and forget and be like, well, model told me that's the answer. So we either shouldn't give this person a loan or we shouldn't. Like that's not acceptable right now.

(Derek at 00:47:51) And so watching organizations trying to understand how to deal with the dimensions of explainability, I think is one that's really interesting. And then watching the race to get there the quickest, that's where the TAM in our industry is at. Right? The total addressable market, it's a huge TAM. You know, it's, I think, $15 billion or something.

(Derek at 00:48:10) But I would tell you 80% of that is around taking the knowledge worker and giving them kind of these auto insight capabilities. But at the end of the day, if I know anything about ML and I'm trying and I use one of these tools and I generate a decision and that decision is an impactful decision like a loan or underwriting or insurability or something that is impactful, but I don't know it and I wanna go to my boss and get an approval on a decision, but I don't understand how the decision was made, that's a big deal. Like as a leader, how am I gonna trust that that's a good decision if you can't explain it? So the explainability side of the AI space is fascinating. You know, there's a ton of dimensions to that space in terms of how to approach that problem, whether it's explainability or interpretability.

(Derek at 00:48:55) All that stuff is really important. So we're watching folks like ourselves that are trying to figure out how do I execute this path much more quickly. We have kind of forked our investment. One is making the tools more accessible, increasing the knowledge, so folks can understand the process of creating analytical outcomes. But there's also a tool-oriented thing, kind of an AI for AI.

(Derek at 00:49:17) I know I created a set of ensemble models that produce some analytical outcome, then I run an AI model over the top of that that says, is there bias in this? Like, am I biasing against race, gender, et cetera, in the outcomes? And so your AI model then looks at your other AI models and tries to understand is there bias? You're seeing a lot more investment on that. So you're seeing AI stacked on AI to drive things like explainability, interpretability, et cetera.

(Derek at 00:49:44) And I think that's a super fascinating field for us in our space.

(Joel Beasley at 00:49:48) Yeah. Because when you're talking, I'm thinking about all of these different use cases that you're, you know, loan origination, maybe something in like the legal field. Right? Do you get into any of like partner... I don't know how to say this. Like maybe a partner network or maybe where you've got people who are in the law category or in the underwriting category, helping you shape the insights for when it's being used for underwriting?

(Derek at 00:50:15) I do think what you're seeing now is a verticalization. Right? So you're seeing verticals come up with maybe more curated ways of answering underwriting questions where, you know, not everybody's having to figure out explainability or fairness or interpretability on every single model, but there's a handful of kind of bespoke but highly curated models on a per-industry basis that, you know, that work has been done for you. So you're not having to go do all that work yourself. So if it's a deep neural network or something that, you know, I've done my ML education, you know, even a lot of folks I know that are strong in this space don't understand how to look at neural networks because it's really sophisticated math.

(Derek at 00:50:56) So how do we create a common way for folks to understand, hey, these X amount of models are very consistent in the underwriting industry. They generate really strong outputs in terms of, you know, high correlation between the inputs and creating really tangible outputs. How do I create explainability, interpretability, you know, fairness explanations around that?

(Derek at 00:51:15) I think you're seeing a trend towards that. Because trying to generalize that across every scenario for every model is an extremely difficult task right now because the tools don't exist for it. I can't take a set of AI outcomes and hand it to a tool and say, you know, generate explainability, interpretability, et cetera for me because I don't know how to do that. That tool doesn't exist today. So the only way you solve it in the near term is by being very specific to say I'm answering an underwriting question.

(Derek at 00:51:41) Here are the X amount of models I'm going to use because these dominate kind of the predictive analytics outcomes. And I'm gonna make sure that for these X amount of models, I can answer all those questions. That may be more manual than it is automated right now. But if you can do at least that, you're kind of meeting in the middle in terms of explainability.

(Joel Beasley at 00:51:59) Well, you created a whole new class of jobs. It's like you created this tool that now needs... it's got these insights, but now it needs these vertical-specific insights about the insights and it kind of layers on top of each other. Yeah. It would be beautiful if there was this AI that you could just say, I mean, the equivalent of go make me money or go generate insights and explain this other AI. But it's just gonna take stacking of human knowledge workers learning how to do this because that will become its own craft.

(Joel Beasley at 00:52:29) Like designing the systems for those insights or the messages that you send on how you explain and interpret this stuff will become its own craft. Eventually there's gonna be a book on it or something of that nature.

(Derek at 00:52:41) Yeah. There... I think if you go to, you know, Google's website or, you know, DataRobot, H2O, they all have now, you know, explicit blurbs on explainability, how they're approaching. I think some folks have written books on how to approach this type of problem. Not to say, hey, I've got the magic, you know, the magic button. You get this button on a model and it gives it all to you. So I think there is, to your point, there's some knowledge stacking that's gonna have to occur to get us there, especially as you continue to democratize, you know, this tooling to more and more people, especially the information worker who has no statistical background, no linear algebra background, doesn't really understand what's being created. Like you have to have that. If you wanna really maximize access in that TAM, you have to... and that's that democratized user base, you have to make it explainable. Otherwise, they can't engage in it because I won't trust that decision because you don't know how it was made. Right?

(Derek at 00:53:33) I'm not gonna risk my business on a decision that was generated by some machine if you can't explain how it was made. So I think that's really the gold... the holy grail of that whole space, that AutoML space is that explainability aspect of it, and we're just not there today. We've made so much progress in this space over the last three years. You gotta believe in the next three years we'll have something out there that does a, you know, at least a decent job of this and is much more automatable.

(Joel Beasley at 00:53:59) Well, hopefully, or maybe one day you'll crack the consciousness code and all of a sudden the computer will start talking to you and...

(Derek at 00:54:05) Oh, man. We'll get some WarGames stuff. I'm not sure I wanna go there. Right?

(Joel Beasley at 00:54:10) It depends. I'll ask like, what can you do for me? Right?

(Derek at 00:54:14) Yeah. Yeah. That's the scary stuff right there. It's the efficacy of automation. Oh, dude.

(Joel Beasley at 00:54:19) That'll be the topic of our next conversation, man. Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected].

(Joel Beasley at 00:54:43) Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.