Free Trial

Alter Everything Podcast

A podcast about data science and analytics culture.
Podcast Guide

For a full list of episodes, guests, and topics, check out our episode guide.

Go to Guide
AlteryxMatt
Moderator
Moderator

Join us in a chat with Collin Graves, CEO of North Labs, about building trust in data and accelerating cloud maturity. Learn how Collin's journey from the US Air Force to data analytics shaped his approach to helping organizations in manufacturing, healthcare, and education. Discover tips on starting small, creating solid data foundations, and tackling challenges like disparate systems and AI adoption. Perfect for leaders looking to enhance data literacy and drive long-term success.

 

 

 

 


Panelists


Topics

 

Ep 175 (YT thumb)-2.png

 

Transcript

Episode Transcription

Ep 176 Navigating Data Maturity 


[00:00:00] Megan Bowers: Welcome to Alter Everything, a podcast about data science and analytics culture. I'm Megan Bowers, and today I am talking with Collin Graves, founder and CEO of North Labs. In this episode, we chat about all things data maturity, developing trust and data. Getting started with building data maturity strategies for success and challenges you might face. 


Let's get started. 


Hey Collin, it's great to have you on our podcast today. Could you give a quick introduction to yourself for our listeners?  


[00:00:34] Collin Graves: Yeah. Hey Megan, Collin Graves, founder and CEO of North Labs. I've been in the cloud data analytics space now for 17 years, which is longer than I care to admit. And prior to that, I was serving with the US Air Force and NATO as an air crew member flying around the world doing. 


Non-data things, but I was really ushered into the cloud space with the advent of AWS back in 2007.  


[00:01:06] Megan Bowers: Very cool. Well, you're our second, or I guess third veteran we've had on the podcast this year, so that's fun. But yeah, I'm really excited to chat with you here from all your expertise on this topic, but can you tell us a little more about your company, north Labs, and what you guys do? 


[00:01:22] Collin Graves: Sure. Our job really is to help accelerate a customer's journey into their cloud maturity. We do that with really a technology assisted services approach to things. So we don't sell SaaS product and we're not pure bespoke engagement, although we can certainly take that kind of work on. We bring a prebuilt environment to the table. 


We have prebuilt data models that we leverage. For customers, primarily across manufacturing and industrial healthcare and life sciences and higher ed are three core verticals. So my fundamental thesis is that if you can build efficiencies for customers to get value faster, it makes you a better partner for them. 


And so that's really what we work on a lot behind the scenes. But ultimately it's about how these organizations can leverage their data to drive insights. To drive operational excellence and ultimately drive. Profitability within their organizations and we just crossed our thousandth project mark with our customers. 


We haven't worked with a thousand customers, but our 1000th project we've delivered for our customers. And yeah, it's still early days and we're have no intention of of. Yeah.  


[00:02:39] Megan Bowers: Well that's an exciting place to be. Congrats on that milestone. For sure. Thank you. Today we're going to talk a lot about data maturity, so I'd love to hear from you. 


What data maturity means to you and why it's important?  


[00:02:53] Collin Graves: I think the most fundamental aspect of data maturity, and this is sound overly simplistic, but I promise it's not, is how quickly or how efficiently can we develop trust of our data within our organization Today, more than ever, we have more data. We have more systems that create data. 


We have more desire than ever before to leverage that data for decision making or automation or operational excellence like I alluded to. But the vast majority of organizations don't know how to get started on that journey. And oftentimes what I see is that trust gets eroded. From their employee's perception of data. 


Because it's unclean, it's not assembled correctly, and it doesn't actually provide a ton of business value. So to me, the underlying current of developing more sophisticated capabilities is really like how do you continue to up the ante from a trust perspective so that people within your organization know, Hey, when I tap into this, it's actually gonna gimme a proper definition of what I'm looking for. 


It's not gonna lead me astray. So I make poor decisions based off of poor foundational data that really is in, in its essence what we seek to solve with customers along their journeys.  


[00:04:20] Megan Bowers: Yeah, and that takes me back to, like before this, when I was a data analyst, entering into maybe showing a proof of concept of a dashboard. 


Immediately someone's like, that data looks wrong. Like that's not right. And it just really tanks that conversation. Right. And that project starting off on the wrong foot can be so challenging if you don't have that trust upfront is what I've  


[00:04:43] Collin Graves: seen. Exactly. And now more than ever, and this is just where the world is going culturally, it people are a lot slower to trust initially. 


They're a lot quicker to brush you off if the initial aspect of things is not right. So what I always tell people is that's why it's worth starting small and really focusing in on one key problem statement out of the gate because you have a better likelihood of getting close from an accuracy standpoint, and you can expand on that over time. 


But if you take the. Proverbial boiling of the ocean approach, and even one department within an organization. It's like, I don't trust this. I'm gonna keep using the Excel spreadsheet that I keep locally on my computer to drive my decisions. You've lost that. You've tarnished a bit of your reputation as a data team, and it takes a long time to earn that trust back, and we see that over and over again. 


So. It really is about like how do you eat an elephant, right? One bite at a time. The same should be applied to building data maturity. You slow down in order to move fast because velocity doesn't matter at all if trust doesn't accompany that velocity in a data program.  


[00:06:08] Megan Bowers: Yeah, totally agree. So that sounds like one of your strategies you guys use for data maturity starting small. 


What are some other strategies that you use at North Labs for I. Increasing data maturity in companies.  


[00:06:22] Collin Graves: Yeah, so we always start engagements. I mean the, we work in quarterly cycles at North Labs. I think that's a great balancing point between a couple weeks of effort, which doesn't let you get very far and say a multi-year effort where things can become super ambiguous and muddy very quickly. 


We find that operating in quarterly turns of the wheel where you're saying every quarter there's gonna be something tangible delivered to stakeholders. Either to move into production or at least begin to test within their own worldview and make sure that things are as we think they are. The other thing for us is, and you know this from your past life as an analyst, requirements change and people always want to throw in edge cases at the 11th hour, which can completely derail progress. 


And so our approach really is how can we bring prebuilt data models to the table? Then focus on the last mile aspect of things much earlier in the conversation. So instead of having to start from scratch and say, as a group like mine, we build similar use cases over and over again for customers. For example, manufacturing customers care about on time, in full and on time deliveries of their product. 


They care about their supply chain. They care about their production throughput. They care about their scrap rate. They care about their predictive maintenance. So there are similar themes throughout every customer we talk to, but historically, even if we've delivered that solution two dozen times in the past year for customers, we almost need to start over from scratch every single time to rebuild those data models. 


But the way we're building it now with more advanced data modeling techniques is we almost have like a Martha Stewart recipe book for these functional. Areas that we can drop into place and then focus on tweaking. 'cause tweaking is where the gold is. 'cause you're making that data model align perfectly to your organization's workflows and really making it work for you and how you define your world. 


And so that all that really serves to do is shorten and compress feedback loops that you have with the customer. So those edge cases are known earlier in the conversation, earlier in that that quarterly turn of the wheel that I was talking about and can help you really get your arms around those different things that people are looking for and make the model work better. 


So that's another thing that we do that, that we find works really well. Not every organization has that capability, but really it's about how can you tighten up those feedback loops. As much as possible in an organization so you're not like disappearing into a dark corner of the business for six months and coming out and hearing your stakeholders say that data doesn't look right. 


How can you get those loops happening more quickly throughout the process?  


[00:09:29] Megan Bowers: Definitely. I think data feedback loops are super important in any sort of project, but. Especially a big data modeling effort. Exactly. For sure. I'm curious if you run into, especially in manufacturing companies where it's like a very old company or maybe they've acquired a ton of companies, how do you handle tons and tons of data systems or like just a massive amount of. 


Variability.  


[00:09:57] Collin Graves: There's a lot of that, and you, you're spot on. I mean, most manufacturing organizations have been running off of gut instinct for a really long time, and the number one growth driver in manufacturing is mergers and acquisitions. So you're not typically saying, oh, we're just gonna stand up a new warehouse and new lines with new equipment and start creating this new thing. 


You're gonna buy a group that already does that because it's, it's gonna allow you to skip over many chapters of that book. Yeah, you're totally right. The number of disparate systems within manufacturing is overwhelming. I mean, we have customers in manufacturing who have 30 ERP systems, and that's like an improvement from the 44 that they had five years ago, and they eventually want to get down to 10, 10 ERP systems. 


That's nuts. That's where I keep going back to from that frequent feedback loop perspective, but really understanding what it is you're trying to solve. Within the business. How many times have you heard in your career, oh, we just want to do more with our data.  


[00:10:59] Megan Bowers: Right? You're like, do more what? Yeah, exactly. 


[00:11:02] Collin Graves: More of what and what, what, what is it supposed to do mechanistically for the business? How will it actually bring measurable impact to the business? Because if you're just gonna say, do more with data, that's a massively expensive and complex r and d experiment. But if you can say. Hey, our scrap rate as a manufacturer is currently 20%, and our industry average is 14. 


And if we can figure out where scrap is being produced and even cut that difference in half, and it's gonna equate to $12 million a year in gross profit retention. Now you have a very poignant story that you can run after that sort of North Star that we call it at North Labs. We know what we're moving toward from an objective perspective and a a, a sort of mission perspective. 


And it helps keep the folks corralled, if you will, from a conversational aspect, but it also helps reduce the scope of what sources you need to worry about. If we're worried about scrap, we probably need one ERP one of those 20 ERP systems. We need our manufacturing execution system and we need our historian database. 


That's capturing sensor data as it flows throughout the line. So if we start with those three, now all of a sudden you have a much more manageable use case versus, Hey, yeah, we have 110 total systems in our group, and let's go ahead and get everything ingested and everything modeled and ready to go, and then perform analytics. 


That's a multi-year effort to arrive at the same place.  


[00:12:42] Megan Bowers: Wow,  


[00:12:42] Collin Graves: this  


[00:12:43] Megan Bowers: is bringing back all the memories of positions and everything. No, it was just, I was at a company that had like a almost a hundred, I think it was er P systems and there was, I feel like times when we did do that approach of chipping away at, here's this one project, here's these 10 tables we need cleaned up and working with it to get each table one by one kind of cleaned up. 


But yeah, it can be an intimidating effort, but I like your advice of really honing in on the problem, and I think that's been a common theme on this podcast across different guests, different topics, especially when we're talking about AI as well, like the importance of really defining the goal and the problem statement well, from the beginning so that you can. 


Know when you've reached that point as opposed to, we wanna use our data better, or we want ai, those types of projects or statements from the top can get really messy fast. So I think that's great advice to really start with a very clear problem, like the example that you gave for sure. And it's  


[00:13:49] Collin Graves: taken us a long time to get to this point, and I'm so proud of our industry that we have finally, right? 


Like really having those business-centric conversations. To understand what needs to happen in order to enable change in the organization or enable efficiency in the organization because it did not use to be that way. Data people were, they took orders from folks and just went, yeah, we can do that. I mean, it's the software engineering parable over and over again. 


Sure, I can build that, but why are we building it and what will it do for the business and what are our assumptions to measure against as we roll out these capabilities? Those are the important questions. 'cause at the end of the day, it's the business stakeholders leveraging this data to drive efficiency or gain of some sort in the business. 


So we can't all just powwow together and go, Ooh, this'll be a fun data project and it's gonna be super complex and I can't wait for all the moving pieces if it doesn't produce actionable insights at the end of the day, even if we as data nerds thought it was awesome. It's gonna be seen as underperformance or non-performance by business stakeholders. 


And ultimately the project is dubbed a failure or a disaster depending on how close you got with it. And that's really, that just boils down to that maturity conversation, right? How can we make those stakeholders, those business stakeholders, trust what we're creating for them and trust that it's going to. 


Provide them the insights they need to pull levers within the business to drive that operational excellence.  


[00:15:32] Megan Bowers: Definitely. So what advice do you have for leaders in the company that are looking to advance their data maturity, their organizations?  


[00:15:41] Collin Graves: Pick one key thing that you think needs improving. Just one. 


And it's super difficult because you will probably think of 10 things or 20 things. Absolutely put them all on a whiteboard noodle on all of them. But then ultimately you have to rank order which one is going to drive the most value for the business, and you need to be hyperfocused on solving that one thing. 


Because here's the thing with modern technologies like Alteryx, right? You can build a really solid foundation of your data infrastructure early on in the process. We're no longer in the world of, we need to pre-buy technologies and amortize this on-premise data warehouse over the next seven years. So it better work. 


The cloud is just in time, scalable. Pay as you go, and we ought to lean into that. So if you build the foundation correctly, you can add on additions to that data house over time. You don't need to start with a three story. Mansion, you can start with a little single story ranch knowing that you can build outward and upward over time if that foundation is solid. 


So that sort of ties into my second piece. Like you pick one thing, but don't try and cut corners. I cannot tell you how many organizations over the years have been like, yeah, we know we need a solid foundation. We need a well-structured data lake, data warehouse, ingestion mechanism, transformation mechanism, BI mechanism. 


But we'll get to that. We just need something quick and dirty for now, and we'll address this two years from now. Well, how often does that end up? Just as a flaming dumpster of mess? Pretty much always, but it's still sexy for people to think that way, especially with the advent of ai. We just wanna start using AI, and then we'll round out the structural integrity of the house down the road. 


But that's like building your house on the sod in your backyard and not actually pouring the concrete and laying the rebar and making sure that the structure can hold. And so what ends up happening is you just end up standing on one of your walls, making sure it doesn't collapse, which is not deficient at all. 


We're seeing that over and over again in the space right now because AI is such a hot topic. Really the success of AI is the same as the success of analytics or data integration or machine learning or whatever. It's that underlying foundation that has to support the weight of those workloads, and it's usually where people are tempted to skip ahead a few chapters. 


[00:18:31] Megan Bowers: Do you see. Pressure coming from maybe at the top of the company, or even if the company is public from shareholders, to get AI implemented. Because that's something that I've noticed in the industry.  


[00:18:43] Collin Graves: Oh my gosh. Everyone is AI driven now, right? I mean you, any SaaS company, any, anyone. Like look at us. We are AI enabled, ready to go and Yeah, so I mean, it is like topic du jour for sure. 


What's gonna end up happening is that a lot of these orgs run into limitations with what they can build and support for their companies because they don't have that foundation. You can stand up little piecemeal one-off things and say that you are AI enabled for sure. But five years from now, you had better be fully AI enabled, like across the entire surface area of your business. 


And I think a lot of those organizations are, are in for a bit of heartburn when it comes to doing that. And we may end up seeing a lot of AI silos throughout those businesses over time. I would venture to guess, but some of them will slow down a bit, pour the foundation and be able to build and support a bunch of AI workloads on top of one foundation, which is ultimately gonna make them more valuable in the long run. 


[00:19:53] Megan Bowers: That makes a lot of sense. We've been talking about like at a high level or for leaders, but I'm curious what individual contributors, what the analysts can do to advance data maturity in their organization.  


[00:20:05] Collin Graves: Yeah, I think it, I mean, I'm gonna sound super redundant, but it's how do we build maturity? 


Maturity is leadership all the way down to individual contributors? How can we put ourselves in a position where we're asking why something matters and what it's going to achieve for the business? Businesses don't operate. If they don't put a cloak over you and say, don't ask any questions, just do it. 


If they do, maybe look for an organization that doesn't, but it should be widely appreciate that an individual contributor would ask that in order to help prioritize work efforts. Me as a business leader, I as a business leader, would really appreciate that, okay, what are we trying to solve? What's it going to do for the business and what's the most efficient way to get there? 


So I think just speaking as a reformed individual contributor myself, the tendency is to say, okay, cool. I got a directive. I'm gonna go build it and I'm gonna look like a rock star by delivering something. But if you're off on what you interpret as being what you need to build, then that's risky for you because you're gonna sit down and somebody's gonna say that the data doesn't look right. 


So if you take that extra time to really understand what. Levers in the business we're trying to influence and what that's going to do for the business. Make sure there's clarity there. Um, that to me, I think is, that's that feedback loop that we keep going back to. That's how you tighten it up and make it more effective across the entire organization. 


It can't just be one sided. It has to be reciprocal in nature where we're talking through. What the ultimate success of a, a solution or a, a bit of functionality would look like.  


[00:21:50] Megan Bowers: Yeah, definitely. And we do have a really good blog about this, so I'll make sure to link that in the show notes from one of our Alteryx associates writing about the process she uses for the feedback loop in analytics when she's building out solutions. 


So I'll definitely include that for this one as. If people wanna learn more or see how to practically implement strategies like this, that would be a great next step for them. But yeah, I think a nice place to finish off on would be your take on what the payoff is for organizations investing in things like data maturity and data literacy at their companies. 


[00:22:29] Collin Graves: I mean, we know the statistics are definitely in favor of. Focusing on data maturity, you're much more likely to be profitable. You're much more likely to beat your peers in terms of growth within your industry. Employee retention, they found is higher. There's just so many opportunities to leverage data for better operational excellence. 


We're no longer in in age. Where a spreadsheet or two and your spidey senses is going to be the best possible approach to building a company. You can do that to a certain point, like when you're just starting out, you have to do that. I started North Labs in 2016. We didn't really start leveraging our own data until probably 2018 or 2019, just getting off the ground. 


You're just trying to survive, pay the bills, not get squashed, but as you move into being a real company. And I think a lot of the listing audience works for a, probably a very well established company in today's age. Really, the only choice is to begin leveraging data because business is only becoming more complex. 


It's only becoming more personalized for customers. It's only becoming more real time in nature from a supply and demand perspective, from a marketing perspective, a supply chain perspective, whatever the case may be. You don't really have a choice but to leverage data to help drive that prowess within your business, that maturity. 


So, and again, we talk mostly with organizations who are just getting started doing it, or who have tried and failed on their own already. That's the, I mean, that's probably 70 or 80% of the customers we talk to. It's very rare when a customer comes to us and goes, this is working super well. We would just like some extra capacity in making it work better. 


Most of the time it's, Hey, we've been trying to do this for two or three years. We've spent millions of dollars trying to do it, and we don't have anything more to show for it than we did when we started. That's the most common scenario for organizations out there because it's hard, right? And so I think that's where starting small and over communication within organizations, but trying to solve one. 


Very tangible business problem first is how you need to get started because that's not the way industry started out adopting data maturity. It was definitely a, let's hold our B to the ocean and see if it starts to simmer, which is totally never gonna happen. So now I think there's a lot better philosophy going around of start small, leverage the cloud's ability to. 


Let you start small and let you be flexible and iterative and agile in how you do things, and then build on top of it once it begins working. So yeah, I mean certainly right now we're, I think we're in a transitionary period where you're starting to see a lot of, maybe the early majority of adopters, maybe into late majority of adopters are starting to think about this. 


Like early adopters have. They've been ahead of this for quite a while. Yeah,  


[00:25:46] Megan Bowers: they're already fully in the cloud. Yeah.  


[00:25:49] Collin Graves: We're starting to see that early to late majority shift, but five years from now, seven years from now, 10 years from now, I think that curve is going to be in a very different spot and organizations will understand how important it's to be doing this, and it'll be more about how do we become more sophisticated in how we do it. 


Versus how do we even get started?  


[00:26:11] Megan Bowers: Well, that's encouraging and exciting to think about for people at all levels in the future. There could be just more interesting problems once you've got that foundation set to be able to tackle, like you said, the more sophisticated data problems and challenges. So. 


Hope it moves in that direction as well. But yeah, it's been really great to have you on the show today. Thanks so much for sharing your expertise, and we'll definitely link resources in the show notes about your company as well as maybe some of Alteryx's Cloud offerings too. But it's been a pleasure to have you. 


[00:26:43] Collin Graves: Thanks so much, Megan. It has a lot of fun.  


[00:26:46] Megan Bowers: Thanks for listening. To learn more about topics mentioned in this episode, head over to our show notes on alteryx.com/podcast. And if you like this episode, leave us a review. See you next time.


This episode was produced by Megan Bowers (@MeganBowers), Mike Cusic (@mikecusic), and Matt Rotundo (@AlteryxMatt). Special thanks to @andyuttley for the theme music track, and @mikecusic for our album artwork.