Heather Rhyne-Christensen
Snowflake GTM Leader
Atrium
Many organizations migrate data platforms, but few understand the complexity of reconfiguring legacy processes that connect migrations directly to revenue impact – fast.
Hendrick Automotive Group moved from a fragmented Redshift environment with inconsistent KPIs and limited visibility to a governed Snowflake foundation powering Customer 360, lead prioritization, inventory optimization, and AI-driven insights.
Join Hendrick and Atrium for a live conversation on how modern data architecture became the engine for sales and marketing performance, not just better reporting.
Why Hendrick migrated from Redshift to Snowflake
How they unified sales, service, inventory, finance, and marketing data
How Customer 360 improved lead prioritization and revenue visibility
Lessons learned when aligning data architecture to business growth
How to design your Snowflake environment for AI-ready use cases
everyone. My name is Heather Ryan Christensen, and I lead Snowflake go to market at Atrium. What we’re gonna talk about today is something many organizations are navigating right now, how to use your data to fuel your business. This is something Atrium knows a bit about. As an elite Snowflake partner, we help customers get to business value on Snowflake faster by combining strategy, engineering, and AI to turn data into measurable outcomes. As many of you know, migrating infrastructure alone doesn’t automatically translate into business impact. The real challenge and the real opportunity is connecting modern data architecture to measurable outcomes and the people and processes that get this work done. Today’s session will explore how Hendrik Automotive Group approached that challenge. Like many enterprises, Hendrik was operating in a fragmented Redshift environment where critical business data was siloed and KPIs were often inconsistent or difficult to operationalize across teams. Through their move to Snowflake, they were able to establish a governed data foundation that powers a customer three sixty view. This helped them improve lead prioritization, optimize inventory decisions, and enable AI driven insights enterprise wide. But what makes this story particularly valuable is that it’s not just about technology. It’s about how the data architecture was designed to directly support sales and marketing performance. Okay. Let’s get started. We have a few incredibly talented folks from Hendrik Automotive with us today. Mark, can you go ahead and introduce yourself? Hi. Sure. My name is Mark Pesker. I’ve joined Hendrik back in October of twenty twenty five, so I’ve been around all of six months. And I lead our data and in engineering as well as our analytic efforts. Awesome. Thanks so much, Mark. Dan, can you introduce yourself? Yeah. My name is Dan Meyer. I’ve been here a little bit longer than Mark. Been here about a year and a half. I’ve kind of been the one that stirred up this whole issue, and also part of the resolution. I am the enterprise project director for Hendrik. Thanks so much, Dan. And then also joining me from Atrium is Zofir. If you can introduce yourself. Sure. Hi. My name is Ophir. O like the letter, feel like the emotion. I know it’s not spelled that way, but I promise it’s pronounced that way. I’m a lead Snowflake engineer and architect here at Atrium and had the good pleasure of working with and continue working with the Hendrix folks. Yes. It’s been an absolute pleasure to get to know you, Mark and Dan. Excited to to dive a little bit deeper today. I think, Dan, you just spoke to a little less than than two years here at Hendrix in what we like to call opportunity that you’ve been able to uncover. So can you talk a little bit more about Hendrik Automotive’s group environment today prior to the work that we did together and your decision to move to Snowflake? Yeah. I mean, a little quick background. I mean, my my experiences with management consultant, data management, going through and cleaning up companies. Got the opportunity to start working with Hedrick. It’s a renowned company in the Charlotte area, also the East Coast. They brought me in initially to manage their BI platform, which at the time was Tableau. Came in, started doing some auditing, and then started to find some irregular data issues, within the the dashboard itself. Basically traced it back. At the time, we had Redshift. We had data issues coming from Redshift. So as I started digging down more and more, we started finding some issues not only within the finance department, but also other departments. With within Hendrik, a lot of the data is fragmented just because everybody’s kind of using a different system. So within this, I mean, I took probably about six weeks to do some analysis. And then after that, we decided, okay. Well, Redshift, while it is a great platform, it doesn’t necessarily meet our needs. So with some conversations with the CIO and the c suite, we decided to try we tried to decide Snowflake. Just different aspects of it. We knew that it would meet those needs that we needed. Unification of all the data, reports based off of event trigger versus time. There was a lot of different aspects of making the migration that came into play. One of the biggest pieces that the C suite executives wanted to have was reliability of data. They wanted it timely, and they wanted it reliable, which I mean, anybody in a leadership position, you’re gonna wanna have that. The way that it was right now is you had all the different departments. As I said, everything was fragmented, so you had different people that had different business definitions of what a data piece was. So you have five guys sitting down at the table, and nobody had the same number. With distrust, people were kinda like, well, that’s not right. So we were like, okay. Let’s let’s start over. So we went through. We did the pilot. It was successful. And then within that, we got approval to go ahead and do the full migration. And, of course, Atrium came on, which was it accelerated everything we were doing. I mean, I I can’t speak enough about what Atrium has done. Not not that Mark hasn’t done his piece, which he has. But, yeah, we we had a rough probably about six weeks there. There was just me, and I I know to tech I know the technical aspect of it to some degree. But that’s that’s why we we got Mark. Yeah. But yeah. Well I mean, it’s it’s go ahead. No. I was gonna say, Dan, with that, I know Mark joined more recently and got a a look under the hood at some of these opportunities as well. Mark, do you wanna share a little bit about what you what you stepped into in in partnering with Dan on the the prior environment. Yeah. Absolutely. And I will say it’s my description to Dan of it all was this is exactly what I would expect for a data wrangling one point o effort would look like, because it was simply, let’s take all these different, you know, files that people are passing around as imports and exports to do reporting, and let’s get it into a database. And then let’s use the database for reporting. And while quite simply, that did lift the efforts to a certain degree, as Dan was explaining, it still left lots of gaps for misinterpretation, miscalculation, and, you know, basically, executive meetings that were wasting time arguing numbers rather than making decisions. So as coming on board as a part of the Snowflake acquisition and being able to work alongside Atrium to, as the business put it to me, I don’t care what it takes. Let’s just do it right this time, was been a been a wild ride and been great to be a part of where we have moved from this into our next environment. Yeah. I mean, we see that all the time. Right? The the IT heroics that still end in different different numbers when it comes to executive reporting. So just frustration all around and then, you know, limitations in how you can actually execute on Hendrik’s strategy because you’re sitting and and debating numbers rather than actually moving the needle and thinking about what’s going to be the next innovative action from Hendrik’s viewpoint. So let’s move a little bit into architecture. And, Dan, I know you spoke about evaluation and POC and some of the solutions. What made Snowflake stand out compared to other options? For me, one of the biggest the biggest factors of moving to Snowflake, I mean, it was the versatility. I mean, that’s and and we’re learning that. I mean, every single day, we we have other departments. I mean, Mark and I joke about that now this whole thing has started where, I mean, all these departments that are popping their head up saying, well, I want Snowflake, and I want Snowflake. And the more we dig into it, the more we learn about the versatility of Snowflake. For the fact is if you’re working with AWS, you’re gonna have some limitations on what you can do, whether you’re having to create a glue script to make it work. Whereas with Snowflake, a lot of times you’re gonna have a direct connect. And to us, that’s that’s invaluable. We were even talking about this yesterday. There’s so much low hanging fruit that we can go out and we can get now that we’ve never been able to before. For us, that was just that that was a huge factor, for the ability for us to unify our data, which we’ve never been able to do before, that gives us insight to be able to see so many different facets of our business that haven’t existed. Yeah. I mean, I think being able to move from that the maintenance phase to innovation quickly is something we see quite a bit with Snowflake’s capabilities. So I’m so glad that that you were able to to see that through your decision making as well. Mark, any thoughts, comments there as as you went through evaluation yourself? Well, I’ve been a long time Snowflake raving fan. And just seeing how that platform has grown and accelerated its growth over the last several years to a point now where, you know, you feel like if you haven’t been checking their website in the last week, you’ve missed a feature release. So knowing that, you know, that platform’s commitment to being future proof, AI ready, and AI embedded in everything, as well as, you know, still maintaining their fundamental ease of use and, you know, light lift in terms of having a team be able to build and execute on it been a game changer for us here for sure. My team is very small, and the number of hands that we have having to stand up and run Snowflake is almost zero because we’re so focused on just building the data and executing on delivering that value to the business. I couldn’t imagine working with any other platform that required any any more effort than that. It’s been a great thing for moving things forward here at Hendrick. Amazing. I know, Ophir, I feel like we’ve we’ve kept you back. Can you talk a little bit about this architecture design? Touching a little bit on Snowflake Snowflake platform that that Mark and and Dan spoke to in terms of, capabilities and and accelerating, our time to value, and then maybe a little bit more on the medallion model. Sure. Yeah. So I I think this is a really fun question. And one thing that the image before us can’t really describe is what my first week with Hendrik was like. And I think that kind of sets up the level of success. And, hopefully, Dan and Mark, you don’t mind me just revealing a little bit about how the sausage was made. But, really, my first week was, hey. We went through this process, right, with Redshift, and we wanna do it right this time with Snowflake. So let’s take the time we need to actually make a meaningful change and make a meaningful design. So it wasn’t like, hey. Week one, what are we doing? How are we doing it? Let’s get it done. Right? It was we’ve been burnt. We we thought we, at multiple scales of the business, feel the the the the sense of urgency, but we also know that, like, we need to be meaningful about how we approach it. So my co architect and I, Kyle Bowerman, you know, we we had plenty of conversations really thinking about how do we design a system for work rather than designing a system telling people how to work. So that really meant that when we got involved, we had to understand the business, and we weren’t here to tell them, Dan, you need be working like this. Mark, you need to be working like this. Right? We just really wanted to understand kind of, like, what challenges they faced, where Snowflake could provide ease of friction, and what was the best architecture that we could develop today and have one eye on the present and then one eye on the future for when they’re massive and have multiple teams in it and all these sorts of things and AI and all of that. So kind of from the first week, we really thought a lot and designed a lot and talked a lot about what are some different architectures that made sense for our current situation, but also would help them evolve and adapt, in the future. And I’d say that was really, part of this new architecture design. I was thinking very carefully about those different moments in, right, data growth, data maturity, enterprise, all that kind of fun stuff. So part of it was how do we make a reproducible core that we have a certain set of rules and certain set of operations to make complex processes simple and simple to quote some giants, way ahead of me in this field, simple made easy. So we chose Medallion, and we thought through about naming convention, schema, RBAC. You know, how do we how do we cut the friction as much as possible? Having looked at the previous slide, right, like, there’s so much that had gone on and so much siloing. So we wanted to think about what are some core tools that we can work with the team on. Right? We met the team. We got to know them. And the I think the design decision to stay really focused on things like DBT, Daxter, Terraform. Right? Data sharing from Snowflake. Right? Being very meticulous about, like, our CICD process and all that kind of fun stuff, allowed them to grow without having to have such a hard mental constraint. So, like, the three core philosophies, I think, were keep it simple, composable, and growable. And so the simple part of the tools. Right? Like, let’s have a focused set of tools that your team can own and feel good about, and that allowed them, I think, to accelerate really, really quickly in a matter of six months where folks who had never heard of DBT are now cranking out pipelines like it’s no one’s business. That also afforded them, like, ease of mental models. Right? There are so many fantastic tools out there. Right? There’s Airbyte. There’s OpenFlow. There’s all these things. But each each new infrastructure you bring in brings a new mental model that we wanted to keep it simple because they said they had ten, twelve different processes that everyone was using at different points. So we wanted to make sure it was super focused. And if we introduced a new tool and a new mental model, we would have to see the benefit up front. Right? So we were we tried to keep it conservative, to make sure we’re all up to pacing on a strong foundation. Composable, I told Mark that I would think of a clever way to make a car reference during this call, so you better believe that this is the one. But we thought a lot and had a lot of conversations about at different points in your journey, you may wanna remove some things and switch them out just like car parts. Right? So how do we make sure that as you grow in your maturity, you can switch things out without having to hinder development or refactor or all these things that plague our field in our industry constantly. So we really set up on separation of concerns. This does this, and it does it well, and we’re not gonna be mixing those things. Right? So we have loosely coupled systems, which allowed things to grow. And when we had to adapt, it was super easy. Right? Like, we would have to try things that maybe didn’t work, and we would just switch out those car parts very quickly. And then growability, they’ll have more data sources. They’ll have more things coming in, more speed. How do we have observability on that? Right? How do we grow those pipelines? How do we answer those questions at those activation layers? But there’s also one extra component, which is how do we have an adaptable architecture? And Mark and I have talked quite a bit about that, right, where a lot of the industry was set into these certain architectures, and suddenly, there are legacy systems, and now they need to do a migration. Right? How do we have a system that grows and is designed for work? And, again, not telling you how to do the work. So Cortex became how we’ve grown that. Right? Cortex is adapting the medallion of your data sources, your ELT, right, your extract process, your load process, our transformations, putting that through the bronze, silver, gold, bronze being the raw data, silver being the transformation layer, gold being the data marts or, right, that activation or what you see on dashboards, and then moving that through this kind of intelligent system where now we have these negotiations, right, between a Sigma handshake and a Snowflake handshake on our back or Cortex agent where where now we have to adapt the architecture to grow it into something slightly different. That medallion architecture is really good for processing data, but now we wanna make sure that we’re adjustable. And if they have other future concerns that the architecture can adapt, and that really sets and came from how we set up that foundation. But I know I just did a lot of yapping. I know there’s a very beautiful image in front of us. Mark, please back check me. Right? Like, let’s be honest here. What does this look like, you know, in the trenches on the day to day? I said a lot of very beautiful stuff. I’m sure marketing is very happy, but what’s the what’s the ground truth? Well, the the first fact check I wanna put out there is the casually glossing over. We had a few chats around how we wanted to proceed with the data. I think we spent the first month or so going beautifully slow. And what I mean by that is there were a lot of there was a lot more than just a tool selection at stake with this migration. It was really an entire reimagining about the people and the process and the tools around how we wanna treat data from ingestion to reporting. And the great thing was with having Dan and the finance team, me and my data engineering team, and Atrium by our side, we had all the right players at the table to lay out all the cards we had to really steer how that needed to come together. I made no hesitation to let the group know from day one that I was on the team that I came in highly opinionated about what I wanted to see come out of this and the results of not just the results, but the structures that I wanted to see work through all of this. And the amazing part was that not only was there agreement, but there was already kind of a lot of synergy around that, you know, going in that same direction. You know, when we talk about the medallion architecture that we’re going after, I wanted to make sure it was more than just a let’s do three layers for three layers’ sake because that’s how it’s stated. You know, we’re saying that our bronze layer is really just true in raw ingestion of everything that we’re seeing and store it all, and we’ll figure out what to do with it later. So that’s where we’re including snapshots. That’s where we’re including capturing data change and, taking complex, semi structured data and breaking it down into tables and rows and columns so that it’s easier to handle downstream. When we were talking about the silver layer, that sort of phrase that I’ve coined around that is, well, units of truth, meaning that each silver model speaks to a very specific data as business aspect, you know, from the simple act of if you’ve been to a dealership and they take you for a test drive, we call that a demo. We have an object for that specific thing. When did it happen? Who was there? What was the car? What was the time? And we break it down that to that atomic level across so many aspects of our business that then when we do move to the gold layer of preparing data for reporting and analytics and all the beautiful things that business likes to see It’s a matter of assembling Legos. The basic building blocks are already put together. Have so many back and forth nerd arguments and discussions at the lowest levels of do we use one underscore or two underscores here? Do we put the word them, in fact, at the beginning, at the end, and what’s the name for this? Because we know that all those little discussions can have downstream effects, and they have long time effects. You know, the patterns and the discussions we were having were setting up build patterns to go forward. We would tackle all these complex items down to the most minute details to arrive at a point that much of what we’re doing became sort of blissfully boring. And the fact it was a machine process that we could repeat, scale out, clearly throw Claude at it because I’ve seen my team, once they understood where we were going, be able to produce lots of models to exactly where we needed them to be in a matter of days. And it’s been a just a really reaffirming process that as we’ve taken all that time upfront to invest in getting all the business knowledge and listening to the finance team tell us about the different aspects they look at, all the woes they’ve had over time, turn that into a process that we’re able to easily scale and manage. Obviously, showed the, that we laid all the right foundations and that we had great guidance for doing so along the way. And, Mark, I think that flows really well into, how has this project changed the way the business talks about data? Well, I know what I have heard firsthand from some of our top leadership here is pointing the answer is gonna be customer three sixty. That is a need that has been expressed from the very top levels where where people are getting excited and hopeful about the fact that we can have a single pane of glass that abstracts the information away from the source system and where it might live and what server. Is it cloud? Is it this? Is it that? Into one picture that allows us to see each customer and all of the transactions that that customer has had with Hendrik, the vehicles that they purchased, the service records, to where we can begin from our side, from an analytical perspective, do some interesting things with trending and predictive. But more so from the, customer facing side of Hendrik, we can reach that point where we assure each of our customers that every time you walk in the door at a Hendrik dealership, you are gonna be treated like the Hendrik family and recognized for the value that you have to us because that we don’t treat every customer well and give them top, you know, world class service, but we certainly want to recognize when somebody’s been a long time loyal, Hendrik customer, we don’t want there to be one minute of a gap in recognizing that versus trying to to decipher how much they’ve done with us and wondering if they’re really a first time customer. So that’s very valuable to us. Yeah. It’s I mean, it’s incredible, Mark. I think knowing knowing how challenging it can be to create a differentiated experience for your clients. I mean, that’s foundational. Showing them that you know them and having Snowflake be a a great platform to be able to provide those insights and have the business change the way that they’re thinking about, again, activating a strategy like that and and the art of the possible. I think that is one of the most exciting things coming out of at least the first phase of of our project. I also loved what you spoke about when it comes to the business working with Dan, working with Ophir and the the Atrium team, and then yourself, kind of like that three legged stool that was critical to have everyone in the room slow down to move fast so that all of the building blocks and all of that Legos pieces were in place so that you could move to execution. And I think we can talk a little bit about that from c three sixty as as you spoke to. I know Dan, Mark just Mark gave us a little bit more on what that life cycle and what that engagement could look like. Can you talk a little bit more about the big outcomes that you’ve seen when building out customer three sixty on Snowflake? Mean, the the biggest thing for us is just to understand our customer base. And and there’s like, the two big aspects that we’re gonna get from this is the customer, but it’s also the vehicle. We’re gonna be able to draw so much insight from pulling this data and seeing different aspects of it that we haven’t been able to see before. And what you were kinda speaking about earlier is that there is this level of excitement. There is you’re having, like, CFOs and individuals that that aren’t necessarily used to getting into the weeds with the data. Like, they’re excited about it. It’s it’s like somebody being excited about dish towels. Not not that there’s anything wrong with dish towels, but I’m just saying that it it’s very it’s very interesting because most of the time, like, you come, you start talking about data, people kinda glaze over, and they’re like, okay. But people are approaching us because they wanna talk about it. They wanna talk about what they wanna be able to do. I mean, not only from an internal standpoint, basically, what we can glean from the data, but also the the experience of the customer. And in the end, that’s what it comes down to. We want the individuals to feel like that, yeah, I’m I’m a lifelong customer, and and I see that appreciation coming from Hendrick. To Mark’s point, not that we would we would treat people differently, but you you kinda get what I’m saying. Yeah. Absolutely. Yeah. I mean Okay. That’s that that that’s the biggest thing we want. We we just wanna have that that better interaction with our customers and also know a little bit more about them and and also kind of what their their taste are. I mean, if you go and you purchase a Corvette every three years, we wanna be kind of ahead of that and be like, hey. Well, just to let you know, they have a new model that came out, and it seems like in the past, you purchased x y z because of this. We’re able to kinda get ahead of that and and be more in tune with our customers. A hundred percent. I mean, I think understanding the new new models, new capabilities, having that put in your hands, maybe there’s trade in trade in options. But, yeah, being knowing that Hendrik knows who you are, what you want, and is looking for ways to facilitate that relationship, I think, is is absolutely amazing. I know we’ve talked a lot about Customer three sixty. Vehicle three sixty is also a critical part of how we were thinking about building out this model. Can you talk a little bit more about how the business is gonna utilize that Vehicle three sixty information? Yeah. With the the vehicle three sixty, this will allow us to track every aspect of a vehicle, kind of when we get it in, what needs to be done to it. I mean, I I work in finance, so we wanna kinda know the numbers. If you have a vehicle that comes in, we wanna see how long does it take to be able to do a recondition on the vehicle, from an operational standpoint for us to be able to create KPIs for the different dealerships and say, okay. You have two Honda dealerships. One is performing better and to be able to kind of streamline our overall process. But the biggest piece is also able to understand from a vehicle standpoint, but also a sales standpoint, what our customers actually want. Because if we’re out there and you somebody’s going to auction and they’re buying a nineteen seventy three Pinto and nobody’s buying them, then what’s the point? Like, we wanna be able to provide what our customers want. And that gives us an understanding of basically what’s out there and where our demand is coming from. Is is there a particular dealership that that excels at selling a particular type of car? We we wanna be able to tailor to that. And and it sounds like you just wanna be data driven. Right, Dan? Yeah. That that’s Wouldn’t wouldn’t wouldn’t that be wonderful? Yeah. Oh, Fear, saw you come off mute. Did you have additional thoughts there? Yeah. So just just to kinda tie it back to a little bit of how the sausage is made. Mark, what are your thoughts around, like, how we’ve been talking about this in terms of Cortex and setting up these models and how to act out activate on it at different levels. I’d love just to you know, reveal a little bit of those conversations. I think they’re they’re fun for how to conceptualize all of these different things. Right? Three sixty is typically a marketing thing. Who’s ever heard of a vehicle three sixty? I know. It’s it’s kind of a wild thing to track down, but, you know, in I’ve tried to take any sort of complex thing that I’m looking at and boil it down to its its core units to make it easier to to speak to and react to. And when I think about a car dealership, you know, we’re talking about a a place where there’s a a limitation of space in terms of a lot has a number of parking spots to put a car. And so that means every car that we bring onto a lot is a choice. You know, we had ninety nine other cars we could have put in the spot except for this one. So by having a vehicle three sixty, we can see how is it that we are how are we best managing that resource, both our space and the number of vehicles we have in inventory. We can get a sense of, does it make sense to, you know, sort of what Dan was pointing to earlier, move certain cars from one dealership to another because we’re seeing, for whatever reason, a higher throughput of a particular type of car at one dealership versus another. And when I say particular type, I don’t necessarily mean make and model. It could come down to, for some reason, somebody’s better at selling yellow cars than everybody else. And I don’t necessarily care what the reason why. I’d love to know why because maybe I want two dealerships to sell yellow cars, but I’m sending every yellow car to that dealership. And so having that sort of visibility around when do cars get brought onto the lot, get prepared with our reconditioning process for sale when it’s a used car that’s brought on? How long does that wait for us wait on the lot before it gets sold? Does that car then come back as a trade in later? Is this something that we’re already familiar with? And can we see you know, how often does that repeat life cycle happen to where maybe we need to be ready for a certain set of inventory that we know we’re selling today? We’re probably gonna see it back in a couple of years as our trade ins. Do we best handle that? And is there what’s to be gained by following that around? Also, we can look at things like, you know, the amount of service that’s been done on a vehicle to know how well it was taken care of, you know, so we can get beyond, like, just the CARFAX side of things, which is obviously easily available. We can get to know our work by who handled our cars and be able to treat that at a a much deeper level. That’s incredible. And so let let me let me push this a little further just for fun. So, the that’s wonderful in terms of the servicing the business, letting perhaps executives know. How do you see this benefiting the dealerships themselves since you made mention of improving that experience? I think by having a better sense of the entire vehicle and everything we know about it, it allows us to gain the trust of the customers better. I think by us having a true complete picture of our vehicles, we can portray we can show confidence in what we have on the lot in a way that someone who’s not getting as deep and wide of our vehicle information as we are be able to match. They won’t be able to speak to the the details and the granularity. And we can make all that data available right up front to the sales reps. We can even find ways to put that in front of the customers as they’re doing their research as well. I think it’s gonna be something that can be easily translated into a stronger conversation between a sales rep and a customer to where that customer can have more confidence in the purchase they’re making. Sounds like big impact. Well, I know this is just a a piece of the journey that we’ve had so far and seeing such amazing business impact, dealership impact, operational bottleneck identification, great great ways to save money. I’m sure Dan’s very excited about that. What is what does the future state look like for Mark? And and maybe touch a little bit on we were just chatting about Snowflake as a platform and what’s seems like a a a flavor of the week innovation, that’s that’s coming. Tell us a little bit about how the business is looking at and bringing ideas to you and and how you’re using Snowflake to to accelerate those ideas. Well, I think we’ve reached the right point in this session to bring up that favorite two letter acronym that everybody’s talking about today, AI AI. Which is absolutely glad I could break that glass for us. Absolutely top of mind for our executives from the top down. It’s really exciting to hear, you know, people who have been with this company for thirty plus, you know, nearing forty years or more talk about AI at the table. And, you know, these are people that are they’re not satisfied with status quo. If anybody knows anything about mister Hendrick or the Hendrick found Motorsports or Automotive Group, you know, selling for status quo and not fighting to be number one is not part of the picture. That is not an option. And so knowing the Buzz AI gets like, it’s top of mind in every discussion. The great news is this. I was sharing earlier today that we, I attended the Charlotte Snowflakes user group meeting last night, and one of the topics was, are you and how should you make your data AI ready? And in that topic, there was a slide presented about six different bullet points about how to handle that. And it’s everything from, you know, having, you know, yes, good architecture, but clean labeling, deep descriptions down to the field level and the table level, and understanding that, you know, LMs are looking for that descriptive language. The great news is as I read that slide, I’m sitting there smiling because the guidelines and guardrails that Atrium put around how we proceeded with our data set us up for success in every one of those areas. There was a hundred percent coverage in that regard. So, yes, we understand that, you know, we want to have that, that clawed chatbot or, you know, Snowflake Cortex item out there to where anyone in the business can reach out and just say, like, how are my sales doing? And the system knows who you are, knows where your domain is, can find your numbers accurately, and bring those chats up to where it’s responding in a very, meaningful way to help everybody control their part of this organization. Absolutely. And I think you you did show quite quite a bit of restraint, Mark, of not saying Cortex code as we as we think about some of those AI capabilities in in places even from GL optimization that we were talking about that you’re able to turn around and and respond to business ideas and inquiries now that you’ve built this incredible environment across the enterprise of thinking about art of the possible, thinking about what we can do next, and then being a great partner. I I again, Mark and Dan, working closely with with both of you, it’s, you know, no longer that separation between technology and the business. It’s, hey. I have a thought. Could we be doing something different? Can we be using AI in this situation and having the right partnership in place with the right tooling and foundation to now execute on those? It seems like ideas are really coming into place. And I know, Dan, as you’re thinking about future state and the ways that you’re gonna drive the business with access to this type of information, I’m sure you’re getting inundated with use cases galore. How are you thinking about your road map? I mean, of course, a BI solution, which we’re, like where we’ve signed up with Sigma, so we’re we’re starting to develop that. But with the the AI aspect of it, it changes the speed of which our executives can make decisions. Because a lot of times, data processes are kinda antiquated. You have executives come down. They say, hey. I need to know how many yellow Corvettes we sold in June of twenty twenty four. And the data analyst may run off and try to find the information. By the time they get back to the executives a day or two days later, they’ve already moved on. With what we’re developing, they can ask the questions directly and be able to make those key business decisions in that moment, like waiting thirty seconds to get an answer and then moving forward with whatever opportunity is that that been presented with. And for us, that that’s the biggest thing. Not only, like, just automating with different functionalities of Snowflake, like DBT and different things like that, but the ability for them to have access is is huge. It’s it’s one of the biggest wins of coming out of this is the executive team just having that constant foothold in the data and and not having to get overwhelmed at it at the same time. They don’t have to dig through tables and try to figure out what it is or do the computations. It exists. They just need to ask the question. And to pick up on, Heather, the comment you made about not just AI for the business, but AI for this effort as well, yeah, I would be completely remiss to not bring up and talk about the game changer that that’s been for our team. You know, when you mentioned the GL, just for the rest of the room here, we’ve not had a the the most concise way to describe all of the GL accounts that we have at this company in one particular platform. We’ve had many subject areas that have covered it to some degree, but what we’ve been able to put together now within Snowflake is that list of that master list of GL accounts. And, of course, when we bring this data in as sort of our dimensional data for, accounting, the accounting team needs a way to manage it. And I’m surely not gonna put myself in any sort of a barrier between the accounting team cabin to manage the account numbers and having that surface in the data. So the most sensible answer that we came up with was to put together a Streamlit app, which was done using Cortex code with one very limited part time developer who happens to be on this call and is talking right now, and be able to use Cortex code to produce that, put it in front of the business, say, here’s the first draft. Give me feedback. Here was the great part. The feedback email came with here’s a suit a few suggestions I have. I did nothing. I copy pasted that, put that into Cortex code, said, do this. Get back version two. I send an email later. Look like a hero. And, really, all I’m doing is copy pasting and sending emails, which is what probably a lot of executives do for most of their day. So good. So we saw a great return on that. And, and, you know, that’s another example of the Snowflake tool being the right fit both for the business, for our scalability, but also for the team internally. I mean, that it it’s could bring a tear to my eye thinking through, again, where you guys started, right, and just in maintenance, heroic, stitching pieces together. You know, it was a glimmer, I’m sure, to go, we could answer these type of questions, put together this type of analysis so so quickly and enter Cortex AI and Cortex code as as great ways to to solve for many of those that, you know, may not have even been possible on platform a year ago, but we’re seeing that level of innovation now. I would be remiss if I didn’t ask, really, all three of you, any advice for folks that are considering a migration to to Snowflake, what it took, who should be involved in the conversation? Maybe I’ll I’ll start off with with you, Dan. My biggest recommendation is planning. It’s it’s all up front. Make sure you kind of know what you’re looking for, how you want it structured, have those conversations, put in the the time up front before you even start pulling data. Because I think that that’s that’s where a lot of people run into issues as you go out and you grab all the data, but it’s all kind of fragmented, and it’s it’s everything. That’s the biggest piece of advice I would recommend. In a matter of who to work with, like I said, I’m biased at Atrium. So, yeah, just planning and making sure that you know what your outcome is gonna be and kind of your road map to be able to get there. Amazing. Thanks for that, Dan. Mark, how about you? Well, it’s gonna sound a lot like Dan’s. I mean, the like I said, when we took when we went blissfully slow for the first month, ironing out so many of these little details, know, like, some of those conversations were painful for parts of us to have. You know, when when Ophir and I are, you know, debating nuances of architectures, there were probably folks on the finance team that were, you know, off camera rolling their eyes. And, likewise, when we’re hearing about battle wounds from handling finance twenty years ago and how that’s evolved over time, I’m telling myself, this is great context to have, and I’m happy to be up here for the story and the progress of it. But is there any way to make this a a shorter story? But all of this all of this planning, you know, around not only just laying the information on the table, but giving a sense of who were the people at play. You know? Who were the what are their concerns? What are the pain points that have been that have been going on to make sure that the process that we put in place just just crushes those out of the system entirely. You know, being very sensitive to the needs that are being expressed and the the way that some of the way we’ve we’ve handled things over time. If we start getting into a story about one of our guys on our team who was spending nights and weekends managing individual spreadsheet data to pull together a report for what we did yesterday, it would make people’s skin crawl. And so just knowing alone that the advances we’ve made here are going to not just make decision making easier and more consistent, but also just save people time and save the company money on wasted efforts and energies that could be funneled. You know? We need the machines to do what the machines do so that we can keep the humans doing what we need the humans to do, And that is a very different world. And I think everything we’re doing with the planning and the recommendations and the, the experience and advisement that Atrium has given us is gonna make us clear to keep that distinction separate and maximize our value across the board. Yes. Love to hear. Less heroics, less, you know, Bill Bill sitting twenty four seven, making sure that things aren’t failing and and the the business is moving forward. Let the let the platform do what the platform’s gonna do. Let people do what what they need to do to drive those things forward. Ophir, I’ll leave it with you. Any other pieces as we think about critical advice for this type of migration? Yeah. I’d say when you’re on Snowflake and you’re siloed, talk to your vendors. Right? You’d be very surprised what we find out when we when we set up those conversations. Part of a process that used to take a a slew of hours, we got on a call right before we made any plans. It’s like, can we talk to these people? Can we like, who are who are the contacts? And, you know, a jump hop and a skip later, we have data shares. And then suddenly, those things that we had to wait hours for are suddenly gone, and then development works like that. So, part of it is really just, like, talking to your vendors, seeing who’s providing you that information. And today, it might be going a certain way because you had to engineer it to solve a very specific set of technical problems. Tomorrow, it’s just a data share. So I’d say, like, that’s part of it is just having those conversations, right, replicating what we do internally, you know, through the partnership, but also to the vendors and just keeping those relationships and those handshakes alive. I mean, it’s been a game changer. Right? Like, once we got those going, work was getting done the next day. Yeah. I mean, I think, from my perspective on the periphery, I think seeing the collaboration and and really that the core focus of what what do we need to accomplish with this type of migration, with this kite type of transformation, an eye on the business, how are we driving those outcomes and building to deliver on that so that it is not this vast, well, we’re gonna plan and plan and plan and plan and plan, but how are we going to sit down with that executive team and show this is what you’re able to accomplish now and being aligned on kind of those core drivers, those core KPIs, and where, frankly, you’re just gonna move the needle with the business. And so it’s been incredible to see the collaboration really between everyone here to make sure that those are well defined so that we’re we’re able to come in and and sit on this webinar and talk about how we are driving outcomes now, not just the LEGO and building blocks, but really execution. And so I think that’s where we know Hendrik is gonna be so successful even moving forward as you prioritize kind of the next use cases on on Snowflake because of that that type partnership there. I know we’re we’re running up close to time, so I wanna make sure that we are able to answer a couple of the questions that came through. Ophir, I’m gonna hand this one over to you. First question was, how did you overcome Reynolds’ data extraction challenges? Yeah. So, I guess, by serendipity, we’ve already covered a little bit of that, but the data share, I think, is the most important piece. And then really just talking with vendors to understand how they’re structuring. Right? Like, are they sharing data that they are getting shared with? Right? Like, understanding the data life cycle at whatever level of granularity helps. I’d also say know the time zone it’s processed in because that makes a big difference on a whole bunch of things, and I never thought I’d be saying that. But ask that question of time zones. It really makes the biggest of differences when you think about your overall strategy and orchestration. I’d say that was the biggest one, and then sometimes we’d have to go back and say, hey. You know, what does this mean? How is this calculated? Certain things like that. But, again, just build good relationships with those vendors, understand where the data is coming from. And, you know, when we started, we looked at the data, and we said, I we like, there’s a lot of preprocessing happening. Can we get to a more raw level? And I think that’s what facilitated the conversation. So, again, data extraction challenges was we were getting files that were hit on an API, and then suddenly, we were getting a data share and able to see the more granular form. And then we got to really think about what it means to rebuild those datasets that were hand, you know, hand delivered to us. So I’d say those are bits and pieces. Mark, anything I’ve missed? I do wanna actually speak very specifically to the Reynolds aspect of this because I think there’s I totally feel the pain when that question came across the board and interesting to get such a specific question. But I’ll since it was a specific question, I’ll give a very specific answer to it. We are pulling a variety of CSD files, which are these kind of self generated reports that you schedule and send to yourself, as well as again, for the rest of the audience, I apologize. This is very specific to Reynolds. But when they ask about RCI feeds, think of that like a and that’s the semi structured data we’re looking at. We’re getting, I believe it’s XML from that system. And so while it’s giving us this great depth of data that we want to use and pull in, the challenge is like, okay. Cool. Now it’s but here’s an XML file. What do you do? Well, the simple lift was to put a step in place to convert it to JSON. Now we’ve got an object that Snowflake has no problem pulling in, And the tactic the technique that we use from there is to take one file’s worth of feed and break that out into multiple tables, each of which represents a subset of the information inside of that file. So if you’ve got one file that’s speaking about the buyer and the vehicle and the some other details and the options on the vehicle and the financing, what each of those different subject areas, we give it its own table. Again, trying to get it back to that units of truth, those small targeted sets of information that we wanna be able to consume and process, that’s a definite part of how we pull that through. Awesome. So there’s one more question. Do you CDP CDP with Snowflake? So, if I have my acronyms right, right, that’s continuous data protection. And so it comes with a number of features out of the box. Right? MFA, time travel, access controls, certain sets of RBAC, all that fun stuff. The wonderful thing about, you know, conversations that my my co architect and I had, customer data platform. Well, I guess, with Snowflake, that is a slightly different ordeal. I’ll just quickly say on security because I think that’s also a really important aspect, and then we can slightly switch to customer data platform. We were very meticulous throughout this process to have inbuilt testing at every phase, making sure that there was a consistency and coherency across the system when we added new dimensions or we added scheme evolution or new data sources or having to, you know, rework how people have been getting the system. This has been a really methodical process of just constantly iteration. And I think that’s kind of a really an important aspect also to just let people know about. Right? Like, you should not be thinking about your governance as an afterthought. It should be part of the structure of your system for every piece to the granularity of table naming, schemas, all that stuff. And that was a big point that we also all were talking about upfront, which is that governance piece, which I think alludes to the fact of, like, how the people process goes now. Right? Like, how we communicate things. Governance is not an afterthought. Right? It needs to be upfront with the architecture from the offset. Mark, anything you wanna add? Dan, anything you wanna add? I’ll just say very quickly, you know, one of the things that I’m going to be eternally grateful for that Atrium set us up from the start was the emphasis using DBT for our transformations inside of Snowflake once the data arrives. That deep emphasis that you all put from the beginning of make the YAML files that are describing the models as complete as possible. Dump all that information in there. You know? Finally, once it got through my own head about, like, the weight of importance behind those things, you know, I I think I’d said in some meetings something like, you know, hey. The the data about the data is data just as much as the data is the data, And it’s it’s a true thing. You know? The that’s probably the biggest component about what’s gonna make our data AI ready. It spoke immediately to kind of the this isn’t our first radio type of feedback that we’re able to get from you, and it definitely set us up on the right path. Yep. That makes perfect sense. I’ll just touch on quickly from a CDP perspective. Snowflake, obviously, an incredible data foundation when you’re thinking about layering on activation and marketing as kind of that brain from real time customer segmentation, etcetera. So I think a lot of synergies between the the the key pieces, and we’ll make sure to follow-up on a little bit more from the questions that are coming in now. But we’re coming up on time, And I want to make sure that that we open it up. I know Mark, Dan, we’ve talked about this. Please feel free to to contact or reach out to Mark Ophir, Dan, myself. If you have any other questions, we’re we’ll make sure to to answer through. But I love that you mentioned the Charlotte user group, Mark. Right? This is a a community. The the Snowflake ecosystem is about learning, sharing, not just data share, and making sure that we’re we’re driving outcomes for our individual businesses. So I love that about both the two of you and about this ecosystem in and of itself that, you know, we’re here to to drive outcomes and love that type of partnership. So thank you so much for being here. It’s just been wonderful both to see your success and to get to know both of you through this process. Thanks again for taking the time. Absolutely. Glad to be here. Yes. Thank you.