Data Council Week: How To Do Self-Service Data Analytics and Business Intelligence Right with Ryan Dolley of GoodData

Eric Dodds 00:05
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.

Brooks Patterson 00:23
What’s up Data Stack Show listeners. Welcome to Episode One data Council week 2024. We’re in the field at Data Council Austin for the third year in a row. Eric and Kostas both had conflicts this year. So I’m filling in. I’m Brooks, the producer of the show coming out from behind the scenes to bring you a few special episodes this week. Matt, my colleague at RudderStack, who brings over a decade of experience working in data science is joining me to dig into the details. But today, the show is all about Ryan, Dolly. He is the VP of product and strategy, good data. And we’re really excited to talk to Ryan today. Got a great show prepared for y’all. So Ryan, welcome to the show.

Ryan Dolley 01:02
Yeah, thanks for having me, guys. Glad to be here.

Brooks Patterson 01:04
Absolutely. Excited to connect. So first, let’s start out how we always do. You just give us a bit about your background, kind of how you got into data and got to where you are today. Yeah,

Ryan Dolley 01:15
so I have been working in data since roughly 2011 on business intelligence data warehouse teams predominantly. And in that time, I’ve done most of the things you would want to do. So I started out at a utility company in Wisconsin on a BI team where you build reports and metadata models and that sort of thing, and then moved into consulting for a while. Then I got into the vendor side, I’ve worked at large vendors, and now startups. So I was at Oracle for a little while. And now Nico, data is VP of Product Strategy. So lots of stops along the way, pretty much enjoyed all of them and have really loved having a career and data so far.

Brooks Patterson 01:55
That’s awesome. You left out your career before data. Went to college for creative writing, and had a career before data in the performing arts, you’ve written plays at face value, that doesn’t make any sense to me. But I know you’re going to tell us and kind of tie a thread together where this makes a ton of sense. So yeah, could you tell us just about kind of going through that how you went from performing arts to kind of working in data?

Ryan Dolley 02:23
Yeah, in undergrad I majored in creative writing in theater. So I’ve, I’ve written plays, I’ve acted in many plays, directed them, both as a student and then professionally, I was the assistant literary manager to Tony Award winning theater, wrote real play reviews for time on Chicago magazine was a script advisor, the Steppenwolf and Sarah, my own non profit theater company sells an amazing so I did all this stuff, right. And over the course of five years, like everything I just described, my lifetime earnings were about $4,000. So I hit a point, you know, you have to when you’re in the arts, you have to have a day job, right. And my day job, I was at Northwestern University, and I can vividly remember working in this department and doing data entry for donations, right? So someone wrote Northwestern to check back in those days. So I’m hard for Matt, you had to manually type in a computer system, the value of the check and who donated and that sort of thing. So I was doing that. And I remember looking around and realizing I was the newest hire. Yeah. And everyone else in that room had been there at least 10 years, and they all hated one another. And if I didn’t do something, I was going to become one of them. Right. And so I, at the time, Northwestern had an incredible employee discount to take to take classes. So I started taking night classes in it, my dad was actually in it and worked in business intelligence, he worked at Cognos. And so I kind of had seen that, you know, this was the type of career that could really take care of my $5,000 earnings in five years. And, so I did night school for that. And honestly, the connection, a lot of people listening, probably, we’ll see, yeah, there’s like, there’s no connection here. This was a total left hand turn out of nowhere, but especially when you get into business intelligence, the business intelligence side of our industry, you’re dealing with people and like, what motivates people and what helps them make decisions and that is also what I studied in my undergrad, right? Was like, what is it about human beings that leads them to take action? And it’s just, you know, instead of doing it with words on a page, I’m not doing it with lines on a chart.

Brooks Patterson 04:48
That’s good. Are there any kind of specific skills that you feel like you’ve taken from you know, certainly, we are kind of it’s probably easier the data were right. We think a lot about content. So creative writing is maybe an easier line to draw. But I also, I mean, I just imagine kind of in data rolls to your point of like, how do we get people to take action? I imagine there’s a lot you take from the performing arts. Yeah, that fits in there.

Ryan Dolley 05:16
Yeah, absolutely. I mean, just learning how, like theater more than I would say any other art is about distilling down that exact question, right? Like, why do people do what they do into an hour and a half, two hour presentation. And so one of the biggest skills that I took from it is actually every play. When you’re an actor, you read the script, and there’s what the characters say, right? But the, what the characters say, is actually just a loose roadmap to what they mean. And in the data world, one thing I found in working with end users is that there’s a similar gap between what they say, and what they mean, right, and learning how to read into that, to uncover quickly to map the gap between what your end users are saying and what they mean. And maybe what they actually need is the type of skill that I have found, was very easy for me to do because of this training in the arts. And I have found over the years that for many data people is actually very difficult. Because they come from a very kind of deterministic, correct, give me some requirements, and I’ll build what you said, then they build it, and the end users aren’t happy. And then the question is, well, I did what you told me to, why aren’t you happy? And it’s because the data team failed in this process of understanding the gap between what they say and what they mean, or

Matthew Kelliher-Gibson 06:44
need a lot of that where they don’t even necessarily know what they have no idea. They don’t, if they don’t know what they want, you have to come up with a way to try to figure out what it is they’re gonna want.

Ryan Dolley 06:54
Exactly. I often say like end to end users or customers don’t know what they need until you give them what they asked for. You know,

Brooks Patterson 07:05
Do you have any, just kind of specific stories or memories of like, instances where that was just like, so apparent? You know,

Ryan Dolley 07:16
I can tell that the one that stands out the most in my mind is that actually it was a lesson for me. It is actually a time when I failed in this regard. When I used this was back in like the OLAP cube days. So for those of you who have been in the industry for a while, you’ll know what those are. Youngins may not but I had spent a long time working with our project management office to develop these cubes. And I was this way. I was very fresh eggs straight out of, you know, Naiad school at Northwestern applying my technical skills now. And I was just building what they were asking for. So it was for the project management office at this utility company. And when I delivered it, it was exactly to spec right, a beautifully designed multi dimensional analysis environment. And of the 29 project managers, 27 didn’t use it. You know, they were just like, they were like, Oh, that’s very cool. Yeah, that’s neat. I’m going to keep using my Excel spreadsheets. And it was because I did not engage my creative mind. I did not, you know, and that was the last time I did that. From that point forward, and I think everybody has the time they get, like six months of work, right? Yeah. Everybody in data has the time they got burned, when they built exactly what they were asked for. And it was a total flop. And you don’t forget

Brooks Patterson 08:36
that. Yeah. Are there any? So moving forward from that experience? Were there any kind of concrete things you’d begin to start doing like, hey, every time I work on a project, I’m going to do you know, XYZ, these couple of things to help me make sure you know, that? I’m not just doing what they say I’m kind of figuring out what they mean, and what they need.

Ryan Dolley 08:58
Yeah, there’s a couple things I would do. Right. So there’s always the, like, some very basic, like using the, you know, the five W’s. Yeah, right, is, it seems obvious, but it’s not done enough. You know, and then the other thing I would often ask them, I would really try to drill in on like, what is the change that you’re hoping for, you’re coming to us asking for some new data. So new analytics, whatever it is, like, what is the change? You’re hoping to drive this right? What’s the end outcome? You know, and it’s, it can’t just be I want to know, blank, right? It’s got to be so that I can, whatever, whatever that thing is. The other thing I would often do is I would ask them if I could go and they could just show me what they were doing today. Like let me sit there and literally watch them do it. Yeah. And oftentimes I’d be able to derive insights about it, especially because I’m primarily focused in BI right it’s really that contact point with The data and the business people and so better understanding their process, seeing what it is they’re doing would often give me insight into how I could design a system that would fit into that enhance that help achieve their goals. Right. But doing that over the shoulder was always super valuable. Yeah.

Brooks Patterson 10:17
Instead of just having a kind of describe, here’s what, right, yeah, I

Ryan Dolley 10:21
I mean, yeah, you gotta get out of there, you can’t be in a room and have them, you know, go over the requirements list for you, and then walk away and feel like, okay, I’m good to go. Right. You’re not?

Matthew Kelliher-Gibson 10:30
No, never are. Yeah.

Brooks Patterson 10:33
I can’t remember who kind of described this, but they talked about, you know, you can get to know a city by looking at a map, but until you walk the streets, you don’t actually understand how this place works. Right.

Ryan Dolley 10:46
Right. That’s exactly applicable to this situation.

Matthew Kelliher-Gibson 10:51
And it’s something that for you, you know, like you said, it kind of came more naturally to you. How has it been when you try to take someone who’s newer at this, who comes from more of a data background? And you’re like, alright, don’t just do what they say, yeah. But they asked for this, like, how do you teach that to him? Because I’ve had that exact same convert, like that moment. And I’ve had that conversation where I teach people and it was either station that went like, I’m going to tell you to do this, and you’re not going to do it until you get burned at least three times.

Ryan Dolley 11:18
Yeah. Yeah. I mean, there’s unfortunately, some element of like, you got to touch the hot coal, right? Yeah. To learn what it means. But you know, it’s very hard. Because our industry attracts, it certainly attracts a lot of a certain type of personality. I think people who enjoy the process of certainty, yes. And finding certainty, right? And the fact that you can, you know, so much of what we do is, is ultimately deterministic, it’s about setting up the right flow. And that’s very nice. I like that too. Right? That’s super satisfying. You know, when it comes to actually, I mean, I have given this advice, similar advice to this many times in my career, and sometimes people will take to it often it’s hard, it’s hard, especially when you’re very fresh, and you don’t, you know, you haven’t gone through the stuff that you’ve gone through. You know, one of the things I encourage people to really do is like, if it’s very foreign to you, is that actually the arts and literature are the place where you can learn this much better than you can learn. If I were to write a book, you know, an instruction manual for techies on how to ask these questions that might be useful. But I actually think that would be less useful than engaging with the arts and literature and learning more about humans and what motivates them, moves them and drives them. Yeah. And I don’t think you can get you can’t get that building out a DAG and DBT. Right, it’s not going to happen. Well,

Matthew Kelliher-Gibson 12:59
I think there is the temptation from a lot of people to be like, I’m gonna read a book on like, the psychology of how people work and Right, no, go read The Great Gatsby. Exactly. Yes. Like, you want to understand what motivates people to read great novels. Yeah, you will much more, not just understand it, but like, internalize that for yourself. That’s

Ryan Dolley 13:19
That is exactly the thing, right? Because that internalization is what happens when you engage with the arts. Yeah, you can read a book about the psychology of decision making, and understand it cognitively. Right, but you need to both in your own life you need to personally experience but that’s a great thing about the arts is like this is how we imbibe the knowledge of things we haven’t personally done. Yeah, you know, and so like, I tell ya, I say all the time to people, like, every, as a data person, my recommendation is at a minimum, every other book you read needs to not be about data. Yeah, it needs to be a piece of literature, or, you know, a biography of an interesting historical character or something like that, right? Not a technical manual, not a business book, not a self help book

Matthew Kelliher-Gibson 14:11
it’s even that’s gonna get you where, like, I remember, like running a team, I had to come up with something because we were centralized that people wanted dedicated things and like, I came up with the idea of like, we’re gonna organize it like an octopus. We’re gonna have intelligence in the head. And also, like, out in the tentacles. Yeah, you don’t read about this stuff. You’re not gonna like that’s where the creative stuff comes from.

Ryan Dolley 14:36
Right? Yeah, yeah. And that’s also like, it’s a, your mind is a muscle right now. And so you need to exercise it. And it will be if you’re listening to this and you know, you’re a young data engineer you just got in college. And you haven’t been asked to do this in a while, or maybe even really, ever. It’s gonna be hard at first, when you pick up challenging novels. especially, but just like going to the gym is hard at first, it gets easier.

Brooks Patterson 15:07
You also on your LinkedIn profile, I think you’ve got Dungeon Master is one of the things there. And I would say you can find outlets like that kind of do the same thing, right? Yes. Like whatever you want to nerd out on just indulge that part of your brain, not just like, hey, at my job, I do data stuff, right? I need to learn how to do data stuff that is saying go playing Dungeons and Dragons. Yes. Like that’s gonna help you do data. So

Ryan Dolley 15:35
it absolutely will read engaging that creativity. Yes, play Dungeons and Dragons.

Matthew Kelliher-Gibson 15:42
I think also, like, I picked up woodworking early on in my career, because it was like, it was a creative kind of thing that was also tangible. Like having those hobbies there. I mean, like, you know, whether it’s Dungeons Dragons or whatever, but stuff that just creates something different, right? Yeah, you know?

Ryan Dolley 15:56
Yeah, absolutely. And, you know, and to the greatest extent possible, I think finding things that you like, you can also engage with others and right, so even if you’re doing something like woodworking, there are other people who are in when we’re right, you can go meet those people exactly. And talk to them. And so even the most solitary endeavor has a community around

Matthew Kelliher-Gibson 16:19
it. Yeah, I know, people they picked up knitting while they’re still knitting circles that Yeah, exactly.

Brooks Patterson 16:26
It’s good. Ryan, kind of going back to your just varied experience. So on one hand, you have like the very clear, okay, creative arts to data. This is different. But also within data, you have a very kind of diverse set of experiences. I work at a huge utility company, you’ve worked in massive tech companies and have also worked at a consultancy. Yeah, I imagine it is like very different experiences. But they’re also I’m sure, some common threads of like, wherever I was, like these same problems, were there. What are a few, maybe those common threads between those different experiences?

Ryan Dolley 17:01
Yeah, I mean, so the most basic problem that we had everywhere, was that we’ve got all this data, and we don’t, we feel like, there’s something we shouldn’t be doing with it. Or it could tell us something. But we don’t know what that is. Yeah, that’s the most it’s almost the existential question of the industry, right. And so, no matter where I went, I was often solving some flavor of that problem, right? And, you know, like at Oracle I was on the Oracle Sales Team. So really, my job was to convince other people that we could solve their problem, right? For about a year, I was doing that as a sales engineer. But otherwise, it was directly addressing it and, and so that I mean that, and then all the implications of that. It’s been very interesting over the course of all these stops, to me to observe, I feel like I can now read the lines of the matrix a little bit. Because I’ve been to so many places with the industry, how the industry works, the hype cycles, the, you know, the technologies that come and go, the way they get popular, the way they get popular often has nothing to do with the technology, it has to do with the people around it. And the social connections, you know, can make or break one startup versus another, even though one may have better tech. Right? But they didn’t know the right people, or they didn’t meet the right people to see all of that has been really fascinating through all these different steps.

Brooks Patterson 18:34
That’s yeah, that’s a great perspective. Another thing, so kind of going from, that’s such a, you know, wonderful, you describe is what the existential problem, we have all this data, what do we do with it? Yeah. Your career has been in the kind of BI analytics space. So that’s getting closer to what we actually do with it. And I know you’ve worked on a lot of self serve projects. Oh, yeah. Lots to talk about there. I knew Matt has some, you know, let’s call it scar tissue from self serve projects. But you know, my kind of perception is, there’s a way to do self serve. But you can’t just say, hey, let’s make self serve, like for the whole like, or can you just talk about that a bit like what’s, what is your philosophy around? Self Serve today?

Ryan Dolley 19:29
Yeah. It’s funny, you bring that up. I recently was talking with Aaron Wilkerson, who maybe you should have on your show. At some point, he’s the head of data governance at car hurt, and fellow Detroit or so on. CNN gets along, but he shares this exact opinion, even more strongly than I do, which is that fundamentally self so he would say self service doesn’t work, I would say, often doesn’t work or maybe usually doesn’t work. So this idea, especially in the BI world, if we walked down there Memory Lane, what has happened in VI over the course of my career, when I started, VI was about a centralized data team, building out very robust metadata models, and then building out dashboards, or even I started even in the pre dashboard, dashboards were new and cool at one point, believe it or not, yeah, we built out standard reports with a header and a footer and maybe dump it to PDF, that sort of thing. But it was self service was kind of a sideshow at that point, right. And, the tooling didn’t really exist. For it, it was all about professionally authored reports, that might take months or sometimes years in order to really produce. But you knew the numbers were right. And everybody could trust it and agree on what they meant. Then we kind of entered the Self Service arrow. And Tableau really took off in the early 2010s. Yeah. And what you saw there was this idea, the theory of value for Tableau, I think, is that you can take a desktop tool, you can put it on an analysts machine, you can let them go crazy with spreadsheets, and and they’re going to produce valuable outputs at a fast enough that, that it kind of replaces what we did before, because it took us so long to build stuff in that era, in this. So this is kind of the basis of the Self Service era, the tableau came out, Power BI came along along with a lot of others, it did deliver very pretty visuals very quickly. Yes. was great at that. Were the numbers correct? Is the question right? And oftentimes they weren’t, I was in a situation once where I was working for a firm, and we were a heavily regulated industry, we had to do business in multiple states. We told the regulators of state A, we had 1.1 million customers, we told the regulators of state B, we had 900,000 customers, it turns out that regulators talk. And we got in big trouble, right. And that was a result of the theory that we’re gonna push all analytics, content creation to the edge onto someone’s desktop, and they’re just gonna get it. Right, right. This is where self service really struggles. And so I’ve seen it struggle very badly. It’s one of those things I find it’s almost like a sugar high in some ways. Like, the first year, when you really embrace self service, you’re amazed at what great things people are building and how quickly they’re building them. But then maybe your five, you’ve got some, some wreckage in the past. And now you’re trying to rein things in a little bit and get more control over the data and overall

Matthew Kelliher-Gibson 22:45
data quality. Well, I think some of that comes from that idea of like, hey, we don’t need to model data anymore. You just throw Tableau on it. And it’s like, okay, but now everyone’s creating their own definition of this metric. Exactly. And now you get to sit in an hour-long debate where marketing and sales and product all tell you what they think churn is, right all got different numbers, right?

Ryan Dolley 23:10
Yes, yeah. And I’ve been in those meetings, and they’re not fun. And so I think, and this is not to say, I don’t think we should return to where things were when I started my career where there was just one data team, and they were the only people who could build anything, right. But what we need to do is kind of rationalize these two ways of working. Yeah, we’re like self service. And this is something, you know, a little plug here that we do a good job of. Self service works best, I think, when it has some guardrails around it. So like, if you have a metadata model that has been validated, and you have, you’ve gotten sales and marketing and product together, and you’ve either agreed on a value of churn or you’ve agreed to have three different definitions, right? Either way is fine. Right? But that’s reflected in the metadata model. And that self service is built on top of the metadata model. Yes, you can bring a spreadsheet in and you can merge the metadata model with the spreadsheet data, fine, that’s great. But having that foundational piece allows self service to be more effective, ultimately, even if it’s not as fun as the sugar high of everybody going crazy with their own desktop dashboards.

Matthew Kelliher-Gibson 24:23
So how do you feel about there’s some people who want to take that to like the next step, sometimes when they’re like, Okay, it’s not just about like, everyone can build on top of your data model or whatever, but they’re, like VPS should be creating everything and using dashboards and finding it all themselves because I know that like, I’ve known several data, like bi people who are kind of like the, we should never have to be asked to query they should always just look for it themselves. And yeah, I can kind of

Ryan Dolley 24:53
if there were a camera on right now, you would all see me rolling my eyes. No, absolutely not. In fact, like the last thing I want a VP doing is building their own dashboard. Unless that’s really their hobby. Yeah. Because I, they have more important things to do, right? I really do believe that. And so what I always tell data people, I have met many data people who have an attitude like this. And they get frustrated that people can’t do what they view as the very simple things, right, like I build out this insane pipeline, you can’t go in sigma, and, you know, build a visualization, like that’s easy. What I was till data people need to remember is that, just like you’re an expert in data, these people have an equal level of astounding expertise in something else, right? And if you were to walk into their realm, you would appear just as clueless as they do when you ask them to build their own dashboards, right? You know,

Matthew Kelliher-Gibson 25:59
and it’s also like the VP, he doesn’t want a dashboard where he has to click through three filters. He wants the number, he wants the answer. He wants the insight, he’s like, that, I think there’s that confusion where like, you think you’re building a dashboard. And that’s what they want? What they want are insights when they want them because they already have a day job. They don’t want to go do 14 other things.

Ryan Dolley 26:22
Exactly. Exactly. And the dashboard. I feel like there’s a lot of disillusion with the dashboard as an interface lately. It’s this in BI every two and a half years, we debate our dashboards dead. And then two and a half years later, we’re still all building them. Like SQL. Yeah, here’s like SQL, it’s dead.

Matthew Kelliher-Gibson 26:41
It’s really not dead. Right?

Ryan Dolley 26:42
Right. You know, and for bi people like the tableau. I don’t want a Mac Tableau actually, it’s an amazing tool, right? So I’m not knocking Tableau. Let me get that out of the way. But like the way Tableau kind of brought dashboards, you know, what’s like a sledgehammer for bringing dashboards to as the interface that end users interact with data. Dashboards are very good. But they’re really only good for in my mind, when you have broad agreement on what the things on the dashboard mean. And you need to operationally monitor them over time. Yes, then a dashboard is perfect. But if you’re just trying to answer someone’s question, a dashboard is a total. It’s overkill. And it’s actually confusing. Right, right. You just need to answer the question.

Matthew Kelliher-Gibson 27:36
Yeah, I think that’s a tough one for a lot of people, especially new data people to understand. It is that tendency to want to show not just what you want, but here’s the context. And here’s how I got there. And yeah, calm down, right? They want a number? Yes, just send an email with the number. They don’t care about

Ryan Dolley 27:52
anything else. They really don’t. And there’s definitely a movement I’ve seen in the data world. I feel the last couple of years around, we shouldn’t be showing our work. We should be. I mean, we should be able to show our work. Yeah. But we shouldn’t. I don’t think we should make that a center point of how we interact with our users. No, I

Matthew Kelliher-Gibson 28:14
I think that I would agree with that, like completely that it’s like, I think sometimes it comes out as trying to, like prove a value or like, look at how hard work we’re doing. And it’s like, really, they just like, they’re in the middle of something. And they need to know, what was this number from last quarter? Yeah, will you put all that in there? The number one feedback that I’ve gotten that I’m sure you got is like, I couldn’t really find where it was, yeah, I spent five minutes going through the email, or the spreadsheet or whatever, to find this thing. This is what I really needed. And like, I can’t use it. If they question it. Yes, we can show a little work. Yeah. But you don’t need to volunteer it upfront. Right.

Ryan Dolley 28:56
Right. Like I wouldn’t. Yeah, I would not. If you’re going to walk into a meeting with important business stakeholders. And your idea is, I’m going to start by building, you know, showing them my pipeline and explaining how I did this. I think you’re in for disappointment. You know, it’s just not general

Matthew Kelliher-Gibson 29:17
rule. Tell them what, tell them the answer. Yeah. And then we’ll go into some detail

Ryan Dolley 29:22
exactly if they want to know, right, if and oftentimes they will, they’ll say, Well, that’s it, especially if the answer is not what they expected. Right. Right. Then they’ll say, Okay, how did you come to this conclusion, but again, the answer to that question is not if I would not get too technical. I would explain the logic of how you came to it, but I would not, you know, show them a DAG. They don’t need to see code, right? They need to see exactly, yeah.

Brooks Patterson 29:50
Ryan, we’ve hinted at good data. Yeah. And I think we’re Yeah, I think we’re teed up to really talk about it now. But yeah, could you just tell us a little bit about what you know, what is good data? What’s your approach? Good

Ryan Dolley 30:06
data is an analytics and business intelligence platform, right? I’m an analytics and bi guy. And what sets good data apart is really that we’re trying to give you almost the best of when I did that history lesson on bi earlier, we’re trying to take the best parts of that earlier era of BI what was which was more focused on metadata modeling, data quality, Guided Self Service, and update that for 2024. Right. So to kind of bring those ideas into a post Tableau world is a big part of what we’re doing of our mission. And then the other thing we really focus on is, that’s kind of the end user experience we deliver and then the way we do it also is bi really has not kept up with advancements in software engineering, certainly. But even in the rest of the data industry. Yeah, in terms of software engineering principles, ci, CD API’s, and automation and integration at the code level. These are, of course, not end user facing things, right. But everything in our platform, we have a very friendly, easy to use metadata driven BI platform, but under the hood, we have SDKs APIs Python libraries, we have, you know, you can access our data through pandas, you can do all sorts of stuff in good data, as a developer, to make your life easier integrated, automate it, scale it using software engineering principles. So we’re kind of we’re kind of taking, you know, the wisdom of the big ages and merging it with with, you know, more a more modern approach to like data engineering, software engineering. Yeah.

Matthew Kelliher-Gibson 31:50
Like you said, I think that’s one for data, like, I have no career, it’s like, when I started, there’s a lot of things, data is different than software engineer, right? It’s like 80-90%. Yeah. And it’s a big difference in the last 10 years. Right. But it’s still not something we should be throwing out everything that we’ve learned from software.

Ryan Dolley 32:08
Well, I mean, it’s just to give you a simple example, like, like BI tools, almost universally have no versioning. Yes, right, actually thinking of that. And every bi person I’ve worked with, doesn’t use like, Gary is crazy. And so everything and good data is YAML. And you can integrate it all into your CI CD and version it and do whatever, you know, just like you would just like data engineers do with their work, we need to do it in BI

Matthew Kelliher-Gibson 32:33
right. Yeah, that’s a crazy one, when you stopped to think about it, and you’re like, wait a minute, how many changes have you made to the underlying data in this dashboard? Like five in the last week? Yeah. Well, what happens if we need to go back to one of possible, we can’t do that

Ryan Dolley 32:50
There are whole companies that exist because they make plugins that do like big companies that make plugins that provide versioning to BI tools, right. And I can tell you, I was working with one of these BI tools, I got hit with a Sox audit. Okay. And then you have to prove whether or not the code has changed. And when it changed, and it was not possible, it was not possible. And then shortly after that, we finally got budgetary approval to buy one of these plugins that allowed us to be diversionary. That’s also

Matthew Kelliher-Gibson 33:23
an argument for having some type of documentation of when you make changes, yeah,

Ryan Dolley 33:27
right. Because we could have been manually doing that you

Matthew Kelliher-Gibson 33:29
could some people use ticketing systems, I worked in places that have done that. But like having something where you can say, we made this change, it was approved by this person, right over

Ryan Dolley 33:39
right. And so like, there’s kind of there’s the way we were solving that problem, which was not at all, and then there’s like a cultural business practice way to solve it. That’s good. Yeah. Best is the software should solve it for you. Most definitely. Yeah. There, you can let the software solve it for you.

Matthew Kelliher-Gibson 33:53
It’s always better, right.

Brooks Patterson 33:56
Tell us a little more about the user experience with good data. I mean, what really sets it apart from other other options out there?

Ryan Dolley 34:05
Yeah, well, we talked about it. So there’s a couple things. The first one is when we think about this discussion we had about self service, right? Good data has everything and good data is built from a fundamental foundational metadata model. And then when you go in as an end user because of that, because we ‘ve already mapped out the measures, the dimensions, you’ve already defined metrics, the joins are all pre pre defined. As an end user, I don’t ever have to know any of that. Everything is a very clean and easy drag and drop. And because someone like me has built the model. You can combine any element in the model, any measure, any fact in any way you want with any attribute and you’re always going to get the right answer. So the user interface has been designed around that it’s Very clean, it’s easy to use. It is not, you know what BI tools inevitably add so many features like they just become harder and harder to use over time. I think that we’ve struck a nice balance between having the features people need, but keeping a clean air face. And then the reality is a lot of those features are really only used by very high level authors, right? Well, for that the whole code, the whole platform is API based and at the end we have a React library. So like, when you hit that point of complexity, or trying to design an output that requires too much would require a more complex tool than we’ve built. Yeah, you can, what we tell people to do on what they do is this, like, this is the point where you shouldn’t be building a custom data app, right? Not hacking a BI tool, right to do what you need. So the BI tool has been designed, like the BI front end has been designed to be easy to use and friendly, for business people to do this kind of guided self service within the walls of the metadata model. And then we give you all the tooling, if you’re a professional author to go and build exactly what you need, rather than, you know, hacking the API interface to get, you know, try to bend it to your will. So,

Matthew Kelliher-Gibson 36:14
So within that, do you find any questions around like, you know, you’re still gonna have to go through and define this stuff, right? You’re gonna have to define your metrics. Yeah. Sometimes when you talk with data, like data, not data people about this, they’ll kind of be like, Oh, that’s hard. Yeah. Like, can you do it for us? Can we get AI to do that for us? Or something? Like,

Ryan Dolley 36:35
yeah, yeah, yeah. Yes, it is hard. The cruel truth about getting AI to do anything is that it really helps to have the metadata defined really well, up front before the AI can do anything for you. But yeah, we saw there’s pushback to that. I mean, this is the reality is that taking the time to build the model, really well structured model up front, is going to take longer than just, you know, wrote on Let’s rock and roll, exactly, it’s going to take longer, but what you’re gonna get at the end is better, more accurate, better performance, easier to use. And so, you know, that’s a sales job, that we have to do good data sometimes, or that our customers have to do internally. When one of their business, when one of their business partners says, Look, I just don’t want to do any of this, I can’t be bothered to tell you what my metrics are, you know, you kind of have to say to him like, well, if you want to get value out of this data, then we do need to work together. And you do need to help us define your metrics. And

Matthew Kelliher-Gibson 37:38
you guys help those internally. You know, you’ve got it like the data team, and they’re trying to go to that business user. Do you guys ever offer any help to them? Hey, let’s go talk to them about Yeah, this is how you should talk to

Ryan Dolley 37:50
We absolutely do that sort of thing. I personally will get involved, because I’ve been that person. Yeah. So I have a lot of sympathy for them, when they’re fighting that fight. Yeah, so that is something we help with. And we will get used, there’s really two big buckets, we use the data platform and general purpose analytics data platform. The other thing that has emerged because we have all these APIs and SDKs. And since the user interface is highly configurable, we get used a lot for kind of data product embedding. Yeah. So that, you know, you’re trying to monetize data, or you’re adding analytics to you, maybe your assess vendor or any industry really, where you’re now building customer facing data products, we found that we slide really nicely into that as well, because you can our our platform is composable, embeddable, code driven, you can really integrate it into whatever application you’re building.

Brooks Patterson 38:48
I did just have to point out we made it 36 minutes before we even mentioned AI. That is a pretty good deal for a podcast that is being recorded in March of 2020. Man, well, we are at the buzzer here, Ryan. But this has been just such a fantastic conversation. A couple of things you, you know, you’re on our podcast, but you also do a lot of content stuff yourself. So can you tell us one, where can we find you? And then tell us about good data as well?

Ryan Dolley 39:26
Yeah. So I’m very active on LinkedIn. Just connect with me, Ryan, Dolly. And then I do every night every Thursday at 12 o’clock Eastern I do a live show with my brother Eric, called the Super data brothers. So Eric is the head of BI and analytics at Michigan State University. So it’s like a family business. And yeah, so we do interviews or we’ll do it like opinion pieces or audience feedback shows. It’s a live show. It’s like a Twitch stream for data nerds basically that’s on LinkedIn and YouTube. So check that out. And then good data, of course, good. data.com If you’re interested in hearing more about that.

Brooks Patterson 40:06
Awesome. Okay, last question to wrap us up. We talked a lot at the beginning of the show, just about kind of exercising the creative side of your brain and the benefit that has in data work. What’s on your current reading and listening list?

Ryan Dolley 40:24
Yeah, yeah. So right now I am. I’m reading Crime and Punishment. So I’m kind of in a serious literature kick. I ping pong between sci fi and right. Yeah, like literature, right? Yeah.

Matthew Kelliher-Gibson 40:36
I can be literature. Yes,

Ryan Dolley 40:39
absolutely. Absolutely. It can be. It’s just Yeah. So I read only literature. The question is, does it take place in space? Yup. And listening to this course on medieval history. So it’s like one of those. It’s like a 60 hour lecture series. That’s been very interesting as well. And I am not not reading any data books right now, like zero. And I’m loving it. So I encourage you all to join me. Awesome.

Matthew Kelliher-Gibson 41:06
I would also say that go read something that’s not data, not technical. You will be so much better for you really

Brooks Patterson 41:14
well. All right. Well, there you have it. Go read. Go read a book, not a Data Book. Ryan, thanks for joining us. We’ll see you again soon. Yeah. Thanks, guys. Thank you.

Eric Dodds 41:25

We hope you enjoyed this episode of The Data Stack Show. Be sure to subscribe to your favorite podcast app to get notified about new episodes every week. We’d also love your feedback. You can email me, Eric Dodds, at eric@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.