The Data Stack Show - 116: Data Democratization & Self Service with Aron Clymer of Data Clymer

Episode Date: December 7, 2022

Highlights from this week’s conversation include:Aron’s background in the world of data (2:18)Recent Clients and major projects (3:30)Helping to spearhead data-driven growth at Salesforce (6:50)Stories about Marc Benioff, co-founder of Salesforce (16:12)Biggest learnings as a consultant in the data strategy space (17:58)The need for data democratization (23:33)Advice for Aron’s younger self in consulting (28:45)Current trends in data democratization and sales service (35:01)Aron’s favorite tools and platforms to use (42:19)Favorite part of the consulting process (47:45)Final thoughts and takeaways (50:29)The Data Stack Show is a weekly podcast powered by RudderStack, the CDP for developers. Each week we’ll talk to data engineers, analysts, and data scientists about their experience around building and maintaining data infrastructure, delivering data and data products, and driving better outcomes across their businesses with data.RudderStack helps businesses make the most out of their customer data while ensuring data privacy and security. To learn more about RudderStack visit rudderstack.com.

Welcome to the Data Stack Show. Each week we explore the world of data by talking to the people shaping its future. You'll learn about new data technology and trends and how data teams and processes are run at top companies. The Data Stack Show is brought to you by Rudderstack, the CDP for developers. You can learn more at rudderstack.com. Welcome to the Data Stack Show. Costas, we love talking with consultants on the show because they get to see so many things on the front lines and they craft so many technologies, especially the ones that are vendor agnostic, implementing all sorts

of technology. And today we're going to talk with Aaron Clymer. He founded DataClymer and really interesting consultancy focused on warehouse visualization stuff. But he was really instrumental in turning Salesforce into a data-driven company, which is absolutely what I want to ask him about. You think about Salesforce, I mean, they're so successful. And I just want to ask him about. You think about Salesforce. I mean, they're so successful. And I just want to know, what does it mean for them to go from being not data-driven to data-driven? I mean, that sounds weird.

You would think that they would be out of the box. So that's one. And then if Aaron is so kind, I just want to ask him if he has any good Mark Benioff stories because, you know, there are lots of those and he was there for a while. So that's what I'm going to ask him about. How about you? Yeah. What I love about consultants is that they have seen enough out there to summarize use cases, but also edge cases.

So I think it's like, I really, really like to talk about what patterns exist out there, building data infrastructure. And also like share some, you know, weird stories, like some edge cases, some things that like you really can't like to hear about. So yeah, that's what I would love to chat about with him. And of course, like also to listen on the Salesforce stories. All right, well, let's dig in. Let's do it.

Welcome to the Data Sack Show, Aaron. Thanks for giving us some of your time. Oh yeah, absolutely. Thank you for inviting me here. Okay. We'll start where we always do. Give us your background and tell us what led you to starting a data climber. Oh yeah. I've just been so passionate about the data space for so long. I had been in corporate America for 20, 25 years before starting this. And it just kind of dawned on me finally that I'm never going to be a true expert. I don't get into consulting.

I was at Salesforce and PopSugar for 10 years. So two companies, two stacks, two sort of business challenges and data sets in 10 years. And I thought, I got to do 50 of those in a year, you know? So that's how I'm going to be able to walk into the next company or the next client and be able to say, I know exactly what you need to do in this situation for this business problem you have in data. So I started Data Climber. We're a systems integration data warehousing implementation firm and started that six years

ago and just have loved the journey ever since. Very cool. And could you describe just like a couple of recent projects so that we get a sense of, you know, what's a, I know having done consulting, there is no typical project necessarily, but maybe just a couple of examples of like clients and projects that you've done recently. Yeah, sure. We're in a lot of different verticals and industries simply because data warehousing overall is very applicable across the board, right? And we can do this almost everywhere.

We happen to get into the major league sports and actually all of sports early on. I think the San Francisco Giants were our fourth client. And the, there, it's just a very interesting story that I think does expand across industries, but most major league sports teams are on a third party, you know, MSP managed service, something, you know, they're not doing their own stacks and they're all feeling a lot of pain around not being able to customize the way they want or bring in the data they want. So there's just this big need out there to own their own destiny, you know, own a stack, do whatever you need to do. You don't, you know, never get stuck kind of thing. And so that's resonated across the industry. So we've been at a lot of different sports teams we implemented for

Las Vegas Raiders, the Vikings, six or seven other teams. And so we've done a lot, including working with the NFL directly. So that's been just kind of a fun, it's almost a niche.

I don't want to label us as the sports SI because of course they're fun logos and they're fun projects to talk about. Outside of that that a lot of gosh payment providers banks financial services and trying to think of a really interesting one to talk about yeah across the board a lot of high tech too i mean i come from the bay area and high tech and uh quite a few COVID based as well,

sort of, you know, health tech as well. Colors, one that's been really interesting, just they have a whole platform that sets up testing sites for medical testing. COVID tests would be a good one, of course, that they cover. And just being able to really drive all their, again, operate more of their operations through data, which isn't a topic I'm just excited about in general. So yeah, a lot, just a different, a great, Pete's Coffee is a great one too.

That's all inventory operations. More of a West Coast coffee company, if you're from the West Coast, but nationwide technically. Yeah. Okay. So I have to ask on the sports side of it. I mean, I don't know a ton about sports, but are you dealing with, you know, if you think about a franchise, you sort of have like player data, but then you also have, you know, this is not the sexy ball, sexy money ball aspect. Yeah, yeah, yeah. So it's not the player side.

It's all the business side of these teams. It's sales and marketing that it's all about fan engagement at the end of the day, which is why that translates across the board, right? It's essentially customer engagement. Yep. You know, that would be the approaches that were applicable across industries. For sure. It makes total sense. Okay. Well, let's, so a lot of questions about current stuff and what you're seeing on the grounds, you know, because one thing I love about consulting is that you get to see such a wide variety of problems at a wide variety of different

business models and all the tools out there as well. But I'd like to rewind. So you spent a really long time at Salesforce. So I think you said it was seven or eight years at Salesforce. And, you know, as we were talking before the show, the main thing was really helping the company become more data driven in a number of different ways. And we're talking about huge scale here, even though it's an even larger organization now. But you said you worked with, I think it was 450 product managers. And give us just an overview of what was the state when you entered and then what was the state when you left and then I'd love to pick your brain on you know how you actually drove that journey

yeah absolutely in eight years does seem like a lifetime for a lot of people these days one organization yeah I came in there in 2008 and there was about 3,000 people at the time in the company I had a feeling they weren't using data as much as they could be. And sure enough, they weren't using data at all, actually. First thing I did there is actually build a predictive model against their 30 day free trial dataset. I bet I could predict with some, you know, some kind of percentage accuracy, who's going to convert, you know, which customers are going

to convert on 30 day free trial. I bet in the first five days, there's some to convert, you know, which customers are going to convert on their day-to-day free trial. But in the first five days, there's some interesting usage data, right? Sure enough, yeah. There was a signal there in the first five days. You could get a much better idea of who's going to convert. I brought that to the head of sales, presented my findings. And the answer I got was, well, why would we need data? We've got a huge sales force.

We call every prospect on the phone and we talk to them. So I don't need data to help me in. I, that was just, it just, it was a, that was a great example of what I sort of the headwinds I was facing. I'm like, okay, yeah, you know, there's no, no real need for data because we think we're doing everything as well as we can be. So, you know, it was a lot of over eight years of having discussions with mostly Prio is mostly dealing with a product usage data set.

So I was dealing with product managers and product more and talking to them, but that bled into sales, marketing, customer success as well. There's just having a lot of conversations around, let's get out of the vanity metric sort of mentality of always having up into the right number of users using my product because Salesforce always is up into the right. Salesforce grew 30% year over year, and I think it still is, you know, since day one, since 1999. So growth has not been the problem for Salesforce, but finding metrics that actually make a difference and using data that actually

is going to be actionable was the challenge. So it was more about having the right conversation about what is actionable data, you know, and getting people to really think through, if I had the answer right now in front of me, what would I do? And if I could figure out what I would do, then it's worth measuring and worth finding out what is the answer to that, you know, business problem and what data do I need to answer that question? A lot of that, those kinds of conversations. Yeah, for sure. And it's so interesting to hear, you know, you think about a company like Salesforce and you're like, well, surely it's data-driven, right? It's so great to hear that. It's like, we call every customer on the phone. We don't need data. So what did it, how would you,

I'm sure there's an infrastructure side to this, right? So was there any existing infrastructure for BI or, you know, what product managers across that many products, you know, what were they doing to get any intelligence on, you know, what they were trying to accomplish? Yeah, right. I mean, when I got there, and unfortunately, one of the handcuffs of being in Salesforce was it took a long time to get cloud technology internally in there. So, I mean, I started in 2008, and we were on an Oracle data warehouse with business objects, all this legacy stack that was very slow and very, took a big team to maintain and operate.

And yeah. Wasn't cloud Salesforce this whole thing though? Like, that was like their, you know. That's slightly hypocritical. It's super hypocritical. It came down to trust, customer trust being their number one value. And they really wanted to make sure they had tried and true technology internally that was super trusted. That was it.

That was the reason. So that made sense. I could see that, you know, from a business strategy standpoint. But unfortunately, it just meant that we had to use technology that had been around for quite a while, right? No cutting edge. So for me, it was about the patterns and, you know, luckily that is actually the key, I think, to doing it right and doing it well in data warehousing, specifically patterns, design patterns, best practices, you know, that hasn't changed all that much. It's changed a little bit, but but you know a lot of those patterns that we developed on an old legacy stack still apply today in terms of best practices so it's great to really hone those you know get those really down

especially when you have poor performance right that's when you have a lot of efficiency you have to you know build into the system so i enjoyed like coming up with how are we going to use this legacy stack? How are we going to do it well? And when it came down to was when I started, it was this data warehouse with very, you know, no really standards on how you're going to ingest data or get data or build metrics or even model your data very well. So we just had to come up with a lot of standardization around data ingestion specifically. So we had to, you know, with 400 product managers and, you know, my team was only 20, 25 people. I mean, you know, to scale that we had to go to a product manager and say, let's talk about your actionable metrics. Once we

understand what those are now, here's how your team can instrument all of this and the products that we capture all this data and just build some frameworks that made that really easy to do. Super interesting. Okay. And what was it like when you left? Like how, you know, what were, give us a couple examples of, you know, the ways that product managers were operating with data that they, you know, that they weren't when you started. Yeah. Well, first of all, I guess I should say from a technology standpoint, my, probably my biggest

win from a, from the stack itself was to be able to get Tableau in there and replace BusinessObjects as much as possible. Because BusinessObjects, I must have trained 400 people on that tool, and I don't think any of them ever used it very much. I mean, the usability was pretty bad. It's a large SAP product. But Tableau, as most of us in the data professional world know, is much easier to use. So that was great to be able to train in a lot of people to self-service their data using Tableau, at least at the time.

And so, but what was more important, I think, was really just the fact that everybody, including, again, a lot of people outside of product knew how to ad hoc query their data and get answers. You know, how to explore data and slice and dice on all the dimensions they cared about. And then build data products actually out of that. So we built things like an early warning system to detect customer health, essentially, early warning for unhealthy customers, let's get them back healthy. Healthy.

That was really more for customer success, right? But it was all based on product usage data from the product teams. And then, of course, the product roadmaps being data-driven was a big thing and making sure that a lot of product managers were making those data-driven decisions. And then the third thing we did that was actually really had a pretty big impact was just internal advanced analytics. So for instance, we would have a couple folks on the team who could do predictive modeling and we would build a predictive model, not necessarily to predict the future, but just to analyze the system and understand the drivers of success of something.

Regression will tell you, you've identified your target, what success means. It'll give you the top five reasons why that success is being achieved at certain customers, for instance. And so we actually did a bunch of analysis that led to whole new products being built with the results. So pure internal analytics, not operational production models or anything, but it was super useful. Yeah, yeah. Did you do any sort of cross-product stuff, right? Because you're collecting it from all these different products. And I mean, the interesting thing about the Salesforce ecosystem is lots of different products you can, you know,

you have the CRM in the center, but a big ecosystem of products, like very inquisitive. Did you do anything on that front? You know, I think maybe the closest thing was to develop white space analysis where you really were looking at each customer and understanding where the white space was, meaning what are all the products they're not using? You know, what are they using?

What are our opportunities to basically upsell, cross-sell, or at least fill out the picture for customers? But not necessarily like, you know, how one product usage affects the other or any of that thing. Yeah, yeah. Super interesting. Okay, last question for me

because I've been monopolizing. Do you have any good stories about Mark Benioff? Oh, man. I wish I did. You know, when they were smaller, I thought, where is he? He's got to be walking around here somewhere. You know, I have nothing good things to say about Mark. I mean, of all of the successful, you know, mega successful CEOs like that. First of all, he founded it. What kind of a founder can take it to a hundred plus billion, right? Really stand up guy found it and creating the foundation.

He created the one, one, one model to give back 1% of profit and time and, and created this huge foundation. I just, you know, just really. Yeah, just nothing but respect for the guy and what he's done and how he was, he stayed out of the headlines.

He there, you know, he was very stand up, respectable person that just drove the company to success. But nothing, no amazing stories about him. I liked the parties he threw. He threw great parties. That was really interesting.

Awesome. All right. Well, thank you for thank you for doing a Salesforce deep dive Kostas. Yes. Take it away. Thank you. Thank you, Eric. So Aaron, you mentioned that you turned into consultancy because you were hungry

like to learn more and like see more like use cases out there and like become, like, let's say like an expert motivator, right? Like in this space. So I'd love to hear like what you've learned and try to turn this into some kind of like patterns, right? And to do that, like my first question is about like the people who are coming, like, that are coming and they are asking for your help, right? So what are they asking for?

Like what's like the most common, like, let is that are coming and they are asking for your help, right? So what are they asking for? Like, what's like the most common, like, let's say project that you see out there is like people coming and saying, oh, like, we don't have like a data strategy, right, and we won't like to implement a data strategy, we won't like to start like identifying the data that we can use,, build the right infrastructure, all that stuff. Or you see more of a modernization need out there where you have businesses that they already do something, but they feel like, okay, we probably need to update a few things if we want to stay relevant. So, okay. These are like, just like two examples.

Hopefully they are like more and better ones. So I'd love to hear like what you see out there. David Pérez- Yeah, it's really interesting. I, you know, as I think about even what you're saying there and I think about how I'm always surprised at how, and maybe then it's the Salesforce story all over again, how, you know, companies are using data a lot less than you think. And so we are maybe surprisingly more working on the table stakes sort of basic stuff. You know, it's just get a data warehouse running

and self-service, get data into people's hands. Not even advanced analytics, not, you know predictive modeling necessarily let's just get a really great data model that really is a 360 view of the business or the customer to you know curate that really well with the again data modeling and best practices there but let's just self-service this data like that it still remains the number one use case and it hasn't changed in a long time. So we do normally work with kind of mid-sized companies.

So a lot of times they don't have something and we are bringing the entire stack to them for the first time. Another good quick just sports example, there is Big Ten. So a collegial sports conference with 28 sports, they didn't have a data stack at all so we're building a full stack for them but it's the innovation there is that all the schools are

going to use it too so it's going to be this shared environment that actually no other sports organization has anything like that even the major leagues don't have a really like shared environment like that they're still sharing files with sftp But so, you know, unfortunately, I don't have these amazing sexy stories of all of this advanced stuff because it really is just getting data in front of people. You know, I was thinking about what you said about from a pattern and from the business side too. Even that is pretty broad. You know, it depends. Some of our clients really want, you know, they're focused on the financial data and we just need to get financial data in front of people. They just want to actually create a P&L in a BI tool, even, you know, get out of their financial tool.

The pattern really is let's all get out of all of our analytics capabilities of our SaaS applications and let's get into a BI tool where we can do a lot more and do whatever we want. It's a lot more powerful. You know, we have some clients where we're getting their inventory data as a mess. They just need to get on top of their inventory data. For others, it's absolutely marketing is a big case study in general. Like a lot of marketers need a better marketing system and some digital transformation. Some let's get it in the cloud and go full cloud and migrate what we have. But, you know, it's just across the board. There's just a lot of almost every department in a company we work with, right?

Sales, marketing, finance, and on a company. Yeah. And do you see that like this transformation like happens, like stays, let's say in the department or is it is it more of something global that happens in the company? Let me give you an example. Make it simpler. Do you see finance coming and be like, we need BI instead of doing everything inside the SAP for whatever reason?

Or it stays there, right? Maybe it's an opportunity for you to expand in the accounts, obviously. But or you see more, let's say, broad projects where companies are like, we want to democratize our data. So we don't just want to take the data out of SAP, but we also want to make sure that everyone

inside the company has access to this data and they can figure out one way or another, like how to get far in from that. Yeah, we usually do start with a department or two, however, because it's mid-sized companies and not necessarily large enterprises. Good news is that we are able to convince them to make sure this is an enterprise data warehouse solution. It's a central data warehouse for the whole company. We might start with one department, one data set.

But even in my stories about Salesforce there, the product usage data set is the entire company can use that and find value from that. So even if we start with one department, the data is still consumed by a lot of different departments. So that's the story. We were huge fans of, again, keeping a single source of truth, one data warehouse.

And actually, maybe that gets back to your question about patterns again, is that I think is the important thing, right? Is to really centralize all your code as much as possible, govern it, keep it governed and controlled so it doesn't become a mess. Because trust, if you lose trust in your data,

that's the number one killer of data projects. You do have to put a lot of thought and governance into what you're doing. Yeah, 100%. And like, you mentioned the word like data democratization like a couple of times. Like, what does this mean in the context of the company? To me, that means as many people as possible who should have access to the data should have it in a self-service way. They don't have to be technical. They have a tool, cloud tool that they can use to then query the data, hopefully ad hoc and ad nauseum if they want to. So give you a good example. I actually back from just before I started this, when I was at Rand

Data at PopSugar in San Francisco, B2C company, and we democratized the data across every single department. And my favorite story from there is the PR department. PR department was two people who were trying to put out as many stories as possible. And they had a fashion search data set. So they could actually detect fashion trends anywhere in the world with this data set.

It was pretty fun. But the only way they could get the data and get a story was to submit a request to the data team and wait about two or three weeks, get an answer, and write a story. It was very easy to get that data set into a modern cloud stack with a cloud BI tool. In that case, it was Looker. And I was able to train them in about an hour and a half how to query that data to find any fashion trend in any place in the world at any time period, right? And so with just that one training, they were then able to produce a story every day.

You know, so that was like 15x productivity because of self-service with, you know, a pretty simple data set. The idea is to get every employee to be able to do that to then sort of incorporate data into their daily job, their business process, and make it more data-driven, that's how you change culture. Henry Suryawirawan, That's super interesting. Like let's say a little bit like longer, like in sculpture,

just like super interesting. So, okay. Making the data accessible is obviously a very important step. You need to have the data there available for anyone to go and work with it if you want to democratize that, but that's not enough, right? Like you also need the people, know, first of all, that the data is there available and also, let's say, build the kind of, like, thought process of, like, when they come with a new problem, like, to go and reach out for the data and see, like, how the data is being, like, held.

And so there's also, like, some kind of, I would say, like, educational parts in this whole process. How do you see this thing working? Because you know, like many times, like most of the time, maybe all of the time, like especially this show, like we focus like a lot like on the technology side of things. But it's very interesting what you said about like Salesforce and the insights that you tried like to communicate to the sales monitor, right?

Like, at the end, no matter, like, what kind of data system you have there, like, it's people's problem, like, to adopt and, like, use that, right? So how do we make people work with the data? Like, learn how to use the data, find the data. And maybe we don't have to, I don't know, but like, I'd love to hear like from you, like how big of a problem is and like how we can solve that. Yeah, I think that's still actually a huge problem. And it's not like I've cracked that not completely at all, but that's why I always

come back to data is hard. Even if you're an end user, it is difficult, right? Just having the data dictionary that makes sense is a huge challenge, you know, so that an end user can truly understand what they're using. I think there's always a partnership between a data team or more of a data, more of the data technical folks and the end users, a constant education. And there's always a communication there because when you want new data, you can't just use, you know, wave a magic wand and have it available. So you do need some help there. But I think it comes down to, you know, a lot of

iterative training essentially and get, and building people what they need. Once they get the hang of it and they do it a few times, but they can have some small wins quickly, then they're definitely going to keep using it. And there's also just that some people are curious naturally and some people aren't. And if you're not curious, you're probably not going to use data very much. But, you know, I think it's that small wins quickly. And at the same time,

and this is kind of what I most love in data is making it as easy as possible. So taking a lot of effort to curate a data set that is easy to use, right? That almost anybody can use. So it takes a lot of work to make sure you've got exactly the dimensions and measures of, you know, whatever you need. And not too much and not too little, right? To make it consumable and not boil the ocean.

But also low-grain data. I always go back to that. I'm not talking about aggregations of data. I mean, have access to the low-grain data, I always go back to that. I'm not talking about aggregations of data. I mean, have access to the low-grain detailed data, but make it just as clean and easy as it is to understand how to use. And I think a lot of people then finally do start to actually use it. And of course, you need a tool that makes it easy. Yeah, makes sense, makes sense. it easy. what you would say to make that person back there more successful in like this attempt to use data for some like because as you said like there are people that are like more curious like you have

people that like they are like say champions of like using data inside like the company but it's not always easy to do that right and it's very easy to get this care on. So what's your advice, let's say, to these people, how they can make it happen at the end? Yeah, that's a great question. I think if I had been a little more on top of a game plan there, I think I would have said, well, it doesn't hurt to do a little pilot, right? Let's do a little POC. Let's get this in the hands of a sub subgroup of the team. Let's get it, you know, 20, 30 sales reps. And let's see if we can come up with, you know, the top five value and benefits to them

of this data and this data set. And let's see if it makes sense. Let's try it out. So I would, you know, perseverance is one of our values. I think that's true in consulting for sure but it's true everywhere right if you persevere enough you usually can get what you're looking for and with a group that is a little bit of perseverance to again iterate through a something to get and use get this in the hands of end users get their feedback and see if you can really understand

how it's going to be best implemented. Henry Suryawirawan, Which function do you think is like the easiest one in the company, like to sell data related like projects? David Pérez, I think marketers are very data driven by nature, right? And there's so much, so much usage of that data. And then again, I think product managers as well. I mean, if you're building any kind of tech products, understanding usage of that product is pretty key. Those two, I think, are most data-driven.

Yeah. I would actually say like the same thing. Probably like I would say sales are like the less open to that stuff. Like, unless you can convince them that like you can, you know, take the pipeline and make it 10x bigger or something, that's the way to work with it. Yeah. That's what I thought. Maybe if you understood who to talk to, who to focus on in your lead funnel.

Makes sense. Right. But yeah. I don't think, but you know, Eric probably had like also some suggestions there because he hasn't worked like a lot on that stuff. But yeah, I think like sales is, it's a very interesting and silencing function like to go and like sell something like that.

But what did you think, Eric? Well, you know, one thing that's really interesting is the incentive structure is really different for sales you know than those other functions you know if you think about marketing or product you know product is motivated by you know planning the roadmap and then executing against the roadmap and, you know, perhaps even driving like feature adoption, although maybe that's, you know, sort of like a growth function within product.

And then marketing, you know, certainly a little bit closer to the sales side and then like you have numbers to hit. And so you're sort of pursuing those aggressively. But really, even in both of those situations with product and marketing, you're measured on execution and throughput. And of course, there are key metrics there, right? Feature adoption or mitigating churn on the product side or whatever it is, right? Activation. And in the marketing side, you have traffic and leads and all those sorts of things and of course there are like

performance bonuses and stuff but when you get into sales the foundational compensation structure is fundamentally different right like your motivation

is tied directly to you know primarily one vector and so I think that's what

makes it difficult is like the focus on marketing and product. You're rewarded for measuring because you're rewarded on output, right? And so measuring is in your best self-interest. And interestingly enough for sales, I'm not saying like, I'm not, this isn't like a dig on sales, but there's just less inherent self-interest and being heavily data-driven because that's not the way that you make money, you know, or make a lot of money. So I don't know, that's my take. But I would also say, I think that's changing. You know, I talked to more and more salespeople who are coming

from highly technical environments, they need to understand a technical buyer. And the more that once you see the power of using data to help you do your job, even in a sales context, like that, you realize how helpful it can be. So I think that's also changing a lot. And I think that, you know, I see salespeople more and more asking for, and even marketing data, right? Like these opportunities, what channels did they come from? Right? That's a great question for a salesperson to ask. Not all of them do. But those sorts of things are really interesting because it can help them

prioritize or, you know, sort of rank and do other stuff. So I do think it's changing. Yeah, I don't know. Does that answer your question, Kostas? Yes. So, Aaron,

let's go back to, let's get like a little bit more technical, like, and let's talk a little bit more about like the technologies that you see out there being used. Like we have, like we've talked like many times about like the modern data stack, like the clouds that we mentioned before. What are like some trends that you see out there that are really, let's say, transformative? And I would, my guess is that you will probably mention the cloud data warehouse,

but together with that, what else is out there that really enables the data democratization and the self-service around data? Alex Rauchmanis- Yeah, absolutely. Well, yeah, my world is around data warehouses. I always start there. But, you know, in my experience, getting the data, yeah, getting the right tool that is going to expose that data, self-service, make it easy, all the stuff we talked about is so critical.

And I've been, I've actually only used two BI tools that so from the BI standpoint that make that really clean easy and you just use the right approach the first one was Looker that I started actually started this company doing a lot of Looker work because Looker is just a lot executes live SQL against your data warehouse so you get instant query instant result again the most recent data you have. And you can query as much data as you want and add as much details. But I love that. Sigma computing is another one that's taken that same approach.

We work with a lot of them lately because they've just become so successful in the market. Again, because of that full cloud approach where you don't have to be technical but behind the scenes the tool is executing directly queries against your warehouse live but on top of that it has a spreadsheet interface pretty much everybody knows how to navigate a spreadsheet interface so the training in the ux is i'm finding this you know that to be pretty incredible in terms of how easy it is to get people using the tool. So I like Sigma computing specifically for BI. And then they're doing a lot more kind of blending traditional analytics with some really cool stuff. Like they can write back to the data warehouse and, you know,

eventually people will be able to use it even more like a spreadsheet, meaning they can even, you know, enter in some of their own data and mesh it with data that's in the data warehouse.

So really interesting stuff like that. And then this bleeds into like the pattern and the trend that I see that I love

is just more and more applications like that that are running directly on top of the data warehouse. So you have the most recent data you can get. And if there's new data that comes in and instantly available to you and all of that. So in the marketing space, we work with another company, Flywheel Software, that allows marketers to create audiences, run campaigns, do A-B testing, do some AI on top of that, all directly on the data warehouse.

And so I love seeing these full cloud approaches that are right on top of your stack. And the future of all of this is running more and more of your business directly off the data warehouse. But that's the most exciting trend I'm seeing out there. I have a question for you on what that trend, by the way, I agree, super exciting trend. Something that Kostas and I have talked a ton about. What does that mean for, let's call them maybe real-time packaged SaaS analytics tools. So Google Analytics, you have a whole class of product analytics tools,

you know, a la sort of Amplitude Mixpanel, et cetera, which are really useful because it's plug and play, right? I mean, everyone complains about Google Analytics, but the reality is it's ubiquitous in marketing because it automatically produces all the reports you need. And if you think about, if I said, okay, could you go rebuild all of these views that are in Google Analytics and Sigma, that is an unbelievable

undertaking, right? And arguably like not necessarily worth it. Obviously there are severe limitations to Google Analytics, which is why there's a huge movement to the warehouse. So there's this interesting gap. What do you think the future looks like for sort of the package SaaS side of things? So the gap, tell me again, the gap that you're pointing at exactly.

So like, you know, Google Analytics is much more limited in terms of data flexibility, querying data points than being able to do whatever you want in your warehouse. But at the same time, it has all these reports out of the box, right? And it's like, well, it doesn't, there's a point at which it doesn't make sense to pay an analyst to rebuild something that's already a prepackaged interface that people can just use out of the box. And so there's like,

I agree that things are moving towards running your business off the warehouse, but, you know, there still is a really big gap, like starting out of the box with Looker or Sigma or Tableau and building an entire suite of web analytics is like, just Google Analytics.

Yeah, that problem is actually being solved with templates. And both Sigma and Looker have a way for a third party to create all those templates and then just have them plug and play at any client. So I think that's where this is going to go. Like even Google, I mean, I haven't looked lately, but they may have already created this, right?

A Google Analytics suite that if you're using Sigma or using Looker, press a button and it's instantly all those dashboards and reports are actually there because it only took

one person to develop those and then they can deploy that anywhere. So it's basically a migration of a lot of that logic, but it's a one-time migration that everybody can take advantage of.

So that'll be where it goes in the long term is that, you know, that's going to make a lot of sense because the reason, again, you made the good point, is why would you even want to do it in your warehouse in the first place? Well, A, there's a lot more flexibility in the way you do it.

B, you can bring in data from any other data source right into that report from your Google Analytics report, essentially, and tack it on. And you can do that in 10 minutes rather than the five days it would take you to do in Excel. So, yeah.

So I think that's, you know, all of the forces are driving people to the data warehouse because of efficiency and, you know, this is a logical place and all of that work that's done. And so all these problems will be solved through templatization, migration of certain things and so forth. Yeah, I agree with that. I think the challenge that I've seen a lot of companies face in that process is that once you get full configurability, you start

changing. One of the interesting things about packaged SaaS is there are guardrails, right? And so they sort of force you to build meaningful reports because you are limited in what you can do. And when you remove those limitations, right, you go down these customization paths that are like, okay, you know, it's getting way too crazy. But yeah, I agree. That's super interesting.

Sorry, Costas, I jumped in, interrupted. No, no, no, no. You should more often, honestly. My pleasure. Please go on. Yeah, so we were discussing about the technologies. Any favorites? I know that's like a hard thing to ask like from someone like consulting, but let's take like data warehouses, right? Like there's a lot of evolution and innovation happening there from like starting with Redshift and having today like products like Snowflake. So what do you love to work with and what you would have, let's say,

suggestions on how to improve? David Pérez- Yeah. And it's still a really amazing, I mean, one of the exciting things about this whole space is it's kind of fractured. There are a lot of vendors and there's a lot of ways you can do this. And at Data Climber here, we are technology neutral at the end of the day.

So we're trying to use best-in-class solutions. So if another new best-in-class data warehouse comes along, we're definitely going to take advantage of that. Snowflake has just become almost a de facto for us, at least. And it's not necessarily because we made that choice early on and said, we're doing Snowflake. They just took over the market. A lot of our clients, by the time we even talked to them, which was pretty early in their data strategy, had already kicked the tires on Snowflake because it was so ubiquitous. So when I started this, it was a lot of Redshift, Amazon Redshift, and a little bit of BigQuery.

And it really shifted to Snowflake. So now like 90% of our projects plus are Snowflake. As soon as I saw Snowflake and started playing with it, I mean, yeah, it's just made a ton of sense to separate compute from storage, never get stuck with scalability, essentially. Just get rid of all the headaches. The way I described it all the time was you get rid of pretty much every technical headache of data warehousing. You know, you don't need an EPA anymore. You don't need to worry about compression and performance and scaling and, you know, compute even.

You can just keep adding compute really easily. You can do it programmatically even. So you can scale it up and scale it down as you want. So all these features and then data sharing across multiple database and data warehouse implementations so that it looks as if the data is in your data warehouse, but it's actually in somebody else's data warehouse. Stuff like that that only the cloud can do kind of blew my mind. So when I saw this, I thought,

yo, yeah, this is the place to be. And sure enough, I mean, we've never had any issues with, you know, Snowflake implementation or any regrets. So Snowflake is, for us at least, the de facto standard. And then for data, I mean, there's really three pieces

to data warehousing in my mind. There's the data warehouse itself. There's data ingestion. And you could talk about reverse ETL, getting the data back out of the data warehouse into systems. And then there's your visualization or your end user applications, right? So for the data movement, essentially, Fivetran has always been a very solid product for us. So we love 5-Tran. And we, again, I mentioned Sigma Computing for the BI layer.

We did a lot of those kind of tools. DBT as well. I mean, DBT is a really wonderful product to use. I love the fact that they came on the scene and also solved problems that you thought might have been solved by now, but were not necessarily, or not in an elegant way, like dbt sort of solves a lot of data modeling challenges in an elegant way. Makes sense. Sorry, I was going to, I didn't mean to interrupt. I was going to mention back to your

area, our discussion about templates and packages, even dbt is packaged up all these modeling, prepackaged modeling solutions, right? So there's another good example of, oh, you can plug and play a Shopify model or whatever, you know, whatever it is, because these vendors are out there and they're standard. Sure.

One last question from me about like the data warehouses. So you mentioned like Red Sea, Big Rear and Snowflake, and you were going through like the features of like Snowflake. It makes a lot of sense, like when you compare it with something like Red Sea, BigQuery, and Snowflake. And you were going through the features that Snowflake has. It makes a lot of sense when you compare it with something like Redshift, right? But BigQuery always was, and still is, a very self-service kind of data warehouse.

It scales, you don't even have to define warehouses. Someone could say that it's even even more, let's say serverless or easily use that like Snowflake. Why do you think that like Snowflake, sorry, BigQuery hasn't managed like to get more traction that it has? David Pérez- Yeah, that's a really good question. I think for me, at least early on, I would always get nervous when I realized it's still a shared resource and it's still a little bit hard to calculate exactly what your compute will be. Whereas Snowflake, it was just guaranteed. And so there was that. There was also just the pricing model of, you know, based on the scanning, you know, scanned records and how do you predict what that's going to look like for your cost model going forward? That was a little harder to get my head around then.

At least I understand compute minutes in Snowflake and I can probably get a good idea of how long something should run. And therefore I can try to predict my costs a little better. So I'm guessing, I mean, that is literally the non literally the less technical side of things, but it's really the marketing, you know, the go-to-market approach to BigQuery that made it a little more difficult for people to get their head around. That's my guess. Super interesting. Yeah, I totally agree, actually. Eric, all yours. All right. Well, we're really close to the buzzer here, as I like to say. But one question I'd love to know. So you said, OK, two big projects over a decade.

I want to do 50 of these a year, you know, 100 of these a year. Yeah. What's your favorite part? You know, and I would say like of the process, not necessarily the outcome, because of course, it's great when you know the company says yeah this is amazing like everyone's logging into their dashboards every day and we make good decisions but you know that's of course the outcome and that's great and we all love that but in terms of the process that leads there what's your favorite part like you know if you had to get your hands dirty going through that journey which part would you choose to focus on? Oh, I hands down love,

love the delight on an end user's face, you know, when they get it and you've taught them just enough to be dangerous with their data, right? And so to me, it's that last mile of like, I kind of talked about this before, right? Curating that data set to be so

exactly what a group of users need let's say that's just simple enough to use but also complex that it's very meaningful so i love the curation part really and that you step back farther that's really data modeling right you really enjoy the modeling but much a little less low of the low level modeling but the high level modeling that semantic layer sort of final modeling where you're pre-defining all your joins and you're making sure your fan outs are not going to cause problems and all the things that you worry about in terms of making a data set bulletproof. So a user can't shoot themselves in the foot. They're always going to get the right answer. Easy to use. I just love that final step of curating the set and then training them up and showing them how to

use it and then watching their eyes light up when they say, oh my God, I can do this and this and I can slice and dice however I want. And I answer questions ad hoc. That's probably why I focus on data democratization too. But I just, I do like that last mile

of the journey the most. Yeah, that's super cool. I kind of view that as a great example of sort of art and science, right? You can almost think about it as an architect, you know, where it's like, okay, well, you're building a house, but if you're trying to bring someone's vision to life, you know, you have the ability to, you know,

shape it in a way that sort of brings them delight as an outcome, which is, you know, really specific to businesses and users, depending on their use cases. So very cool. Well, Aaron, this has been such a great show. Appreciate the time. Learned a ton. I'll look for an invite from Mark Benioff

for one of his parties because I hear that they're really good. You got it. All right. Well, thanks so much for joining us. Yeah, thank you very much. It's been a pleasure to talk to both of you. It was a great discussion.

Thanks for talking shop. I always love it. You know, one thing I think that is so interesting about Aaron's story that we've heard so much on the show, and this is going to sound so cliche, but you know, the people side of things is always the hardest. And he referred back to that over and over again. You know, just in terms of like, how do you make progress in organization in terms of becoming data-driven? And it's, I guess maybe my big takeaway, this is how I would frame it, Kostas.

I love talking with people like Aaron because they just don't talk about the tools when you ask them about becoming data-driven, right? Like, it's so simple for them, right? It's like, I mean, I need to ingest data. Like, how are you going to do that? Well, it doesn't really matter. There's lots of good tools, right?

You need to warehouse data. And it's like, well, how are you going to do that? And they're like, well, I mean, like Snowflake's great, but like everything's pretty good, right? And then it's like, well, you need visualization. And it's like, well, how do you do that? And they're like, we have a couple of tools we prefer, but like everything's pretty good, right? And then it's like, well, you need visualization. It's like, well, how do you do that? And they're like, we have a couple of tools we prefer, but like everything's pretty good.

I just love it. Like for Aaron, things are kind of simple. You know, you have ingestion, you have, you know, storage, you have modeling and then you visualize it so that people can make decisions. And like they've changed preferred vendors over time, but like he wants to help people make better decisions. And it's really refreshing to just hear the perspective of like,

all the tools are awesome. And maybe we prefer some of them, but really it's doesn't matter what you use. It's actually about like getting a good data model and delivering like really clear visualization and dashboards so that people can make decisions. So I love that. I think it's so great.

Yeah. Yeah. A hundred percent. Like, and I love the focus on like the human factor, to be honest, like, because that's what it is, right? Like you can have the best technology out there, but you need the culture and like to educate the people on how to use this data and like how to think in terms of data and they make decisions

or like make it part of their work. Right. I think like, especially like the example that he gave about his own experience back at Salesforce, he went like to sales and sales were like, okay, yeah, that's cool. But why do we need it in case like. We cost every customer. Yeah.

Like, so I think that was like a really good part of like a conversation that we had. And yeah, it's one of these things that you can, you need a consultant. You need someone with like, you know, implementing, but it's not part of the system itself. Right. So it's an observer and like they can see what's going on and like they can see what the bottlenecks are.

And yeah, looking forward to have him again on the show and like discuss more about this. Yeah. Yeah. You know, if you're a vendor in the modern data stack, every problem is a nail that the modern data stack needs to hammer. And it's really refreshing to hear from someone like Aaron

who brings it back to basics. So thanks for joining us. We'll have many more great guests on the show and we will catch you on the next one. We hope you enjoyed this episode of the Data Stack Show. Be sure to subscribe on your favorite podcast app to get notified about new episodes every week. We'd also love your feedback. You can email me, ericdodds, at eric at datastackshow.com. That's E-R-I-C at datastackshow.com. The show is brought to you by Rudderstack,

the CDP for developers. Learn how to build a CDP on your data warehouse at rudderstack.com.