Shop Talk With Eric and Kostas: Data Politicians

Eric Dodds 00:05
Welcome to The Data Stack Show Shop Talk, this is a new format, where we just go off the cuff and talk about a topic that is of particular interest to either Costas already, we plan to have guests that it will be a lot more informal. And to that end, I picked the topic for this shop talk episode, our very first one, and Costas doesn’t even know what it is. So this is his first time hearing about it, which I’m really excited about. So Brooks can obviously figure on the title that this is my thought and what I wanted to get your take on process, and it relates to data engineers, or, or data professionals also needing to be politicians within their company. And I will give a very concrete example of what I mean by that again. So recently, at RudderStack, we implemented this new tool, okay? That is like a, whatever, some sort of engagement tool on the marketing side. So as the marketers, right? And over, always the marketers, and there’s a really cool tool and, you know, whatever, to help drive conversion law, right. And so there is a big appetite in the company to like, roll this thing out fairly quickly, and try it and sort of prove it out, like whether it works. So as part of that process, to optimize for velocity, which you have to do in startups, you forego, like some data tracking, right? Like, let’s get this thing rolled out. And now we have a tracking plan. And we have like, actually very sophisticated like, very clean like knives. And then of course, the thing starts to work really well. And so marketers and other people start to ask questions, I go, that’s interesting. Why law, right? Like you sort of you sidestepped the tracking process, and where to optimize for wasabi, then you have to go back and pay technical debt in order to, you know, track everything. And like Julian’s partition level, blah, blah, kind of details, all the sign in terms of companies that what really like that process got me thinking like, this is just an inevitable part, especially in startups, rolling out tooled and trying to wrangle data or whatever. But the more I thought about it, the more I realized that that dynamic is, it sounds funny event that sort of inherently political, right? Like, you have to navigate a THON, who like wants to do comprehensive data tracking, or, you know, whatever, you actually have to navigate a lot of different parts of the organization in order to build consensus around like the discipline of.if. That makes sense. So anyways, that topic has been on my mind, and I’m interested in your opinion of like, what that’s inevitable, like, what is that? What’s your experience that inside of organizations? Like? Is it even the responsibility of the data team to sort of, let’s say, like, do police and politics that makes sense around like that dynamic? Right. And of course, as organizations grow, like, the problems become more severe with link for datasets and all that sort of stuff. So I mean, I don’t think what you’re

Kostas Pardalis 03:34
describing is something that applies only to data engineers, to be honest, I think it applies like to pretty much like everyone in the company. And I think it has a lot to do with the difference between like a company that is a startup and the company that is transitioning into becoming like a real company. And I’m pretty sure that like, if you, if you pay attention, like you will select similar behaviors and like similar question. So like the remote like, also, I can mark it in your like, in sales or like, Ha, like, it’s always like this kind of like trade off between, like velocity and process, and how like, you compensate for the lack or like the difficulty of communicating. And that’s, I think, like, it’s a continuous struggle, like with organizations in general, right, like, I don’t think I’ll ever like goes away now. Well, I’m

Eric Dodds 04:34
looking back at you just a little bit like that. So I hear you. I raise you in that. That same dynamics happens within different organizations. But the reason I the reason I was thinking about it so much was that for the data team, generally they’re for view over data also impacts those stumps. Right. So like the, like, I agree that that dynamic happens, you know, whatever that happens in sequencer like that happens within marketing or whatever, right that if you have a company that leads trying to be data driven, the doubt the impact of the data team like not having purview over that stuff actually impacts local. So my, my position is that the, it’s a different problem than that happening in a siloed sense within like a particular organization like marketing.

Kostas Pardalis 05:42
Yeah, but like, Okay, how many companies have the stage? Or like the maturity levels for others that are complex?

Eric Dodds 05:50
Not very many. I mean, technically, we don’t, it’s like an entire team. Yeah, I

Kostas Pardalis 05:56
think it’s the fact that, like, the team up the other side, cares about that stuff or like has the luxury to care about LogStash is because of the nature of the company itself and the product itself, right. It’s like you’re building a product that is solving or is part of like the problem that you are discussing, right? So you have bases, and you start thinking of reads like months, months earlier than other company like, well, what companies at that stage probably don’t even have like a data engineer like

Eric Dodds 06:31
it sets up to think that this. I agree with you, but like, this is a problem that I’ve seen happen to companies of like much larger sizes?

Kostas Pardalis 06:43
Oh, yeah, of course, it’s girls. I think that’s where you stumped, like seeing these problems, like larger companies. And the question. I mean, let me ask you a question. What is the responsibility of the data engineering team in, let’s say, a big company, let’s say I owe like, a public company, that’s is b2c. And it has like, a ton of data. And they are very, like data driven, or they even have like products around data like they do in Maryland, like, whatever. So what did you think is like the responsibility of the data engineering? Not the rest? Because data teams can be like many different things, right? Like, what we’re bringing back do that in 10 years?

Eric Dodds 07:29
Yeah. Well, that’s, that makes me want to return to my original. The original question I asked, because I think my intention was probably more to fit your brain on lines. The person was, like, responsible grown, but uh, not pounds x. Actually, I don’t think the data engineering team? That’s a good question. I mean, I don’t think the data engineering team would necessarily be involved

Kostas Pardalis 08:00
in that decision making process, right around like the implementation. Does that make sense? Yeah, I don’t think so. either. I think the deadlines engineering team has an infrastructure that they need to take care of. And they need to make sure that whatever pipelines run there they run and the data that for whatever reason, is delivered through like these pipelines is correct. Okay. Well, I don’t think that they like the data engineering team can go like much farther than that in terms of like responsibilities. Right? That’s when you start getting into like poise at the end, the consumer of the data. And who’s going to benefit from this data? So let’s, let’s say it’s like, the market demands, or like the product monster like these, all of the people at the end will be like, Okay, why we don’t have mistakes, while we need it, and how we can get it and why we didn’t, let’s say do that, like, from the, from the beginning, right? When we when we started, like designing the service or these products, or whatever? Well, so let’s

Eric Dodds 09:13
So let’s dig a step deeper in the context that you’re talking about? Because I think that gets closer to the tension behind the question, which is the especially in a larger organization, like one team does something in order to optimize towards whatever their goal is, right? Let’s just say, because I’m in marketing, well, we’ll make steps. Your stylist presents. So sale does something or implement something with good intent because they’re trying to optimize towards their goal right and more revenue revenue data. That’s right. But what they do create that byproduct of that as a data problem for marketing set, right? It’s just like, Okay, well, now marketing is data sentencing, because they don’t have whatever new life data. That’s right. The sales team is producing three new ends up, which again, like isn’t enough in the inevitable problem. And I need your point is really well, man, like, probably the description and RudderStack, RudderStack cron, like over indexes for like maturity because of the nature we do and like the team leader, or whatever. So the question is not like, what do you do in that situation? Where it may not? That’s inevitable, right? And there are ways to solve that. But I guess more of the heart of the question was, do you think that there is someone in a data focused role, who should be the arbiter of all of that centrally in the organization? I understand you’re worried about data engineering. But I think part of the problem is that if sales a document a optimizing towards it needing to do, then they shouldn’t be thinking about the larger like data layered structure, right. But because of that, they can’t know that there’s gonna be like, they’re gonna blow stuff up downstream. Right, then in just because of the nature, I mean, I don’t know there’s also Leon’s right. If you think about it, like there’s like, a pull request for implementing a new sales thing that creates data, right? Like in the context of soccer, like they’ll just do it or plug it into Salesforce. Yeah. Yeah, I

Kostas Pardalis 11:39
think. I mean, at the end, I think it’s like more of like a roll duck ringland, whether you’re describing and I don’t think that cost to do necessarily like with data roles. And I’ll tell you what, I mean, like, someone who’s like dealing with data, let’s say, let’s take an analyst, let’s say you ask them, like a marketing you there you calm, and not only that works with you to make sure that you have all the insights that you need.

Eric Dodds 12:04
Yeah, and like a lot of times analysts are like, can be a centralized function, right?

Kostas Pardalis 12:08
Yeah, but let’s, let’s, let’s assume that, like you, as Eric, you have, I don’t know, like Psalm on there, like a data minion that.

Eric Dodds 12:19
They just like ETL very informed,

Kostas Pardalis 12:22
you aren’t reading all of them. But like the work that I mean, the median exists to answer all the questions that you have, right? Without assumption that these questions can be answered using data. Okay. Now, I don’t think that like this person can beforehand, right? Know what, first of all, what questions you have, or you might have, and what data is going to be needed for that. And that’s why I’m saying that this is like more of like a product problem. Because think of whatever function you like, whatever you’re trying like to build as marketing or sales as a product, right? Like, at the end, why do you need data because there is a lot you need data is to figure out the success of what you have implemented. Right? Like, that’s why you need analytics there. And the impact that whatever you built has like in the business, so it’s your responsibility as the stakeholder like at the customer, okay, to sit down with whoever is doing like to build that, and figure out how you can measure the success of whatever you’re doing. And from there, figure out what data is needed. Yep. And build the infrastructure for lat. But this is not something that let’s say, the data role, or count can do for you. In a vacuum, right?

13:47
Yeah. Yeah,

Kostas Pardalis 13:51
I think you’re asking too much. Like, it’s like, what do you expect from your minion? Like, the minion cannot like, like a genie come out? And just answer like,

Eric Dodds 14:01
well, after the part of the reason, part of the reason, and this is why like, no, no, no, no, no,

Kostas Pardalis 14:07
no, because you know what you’re doing right now. You do exactly what sales do to marketing. Once I something goes wrong, sales will be where are my leads? Right. And then you go to the data person, and you’re like, what are my insights?

Eric Dodds 14:24
So I thought I was the one who is first to get insights from the data.

Kostas Pardalis 14:29
That’s what I’m saying by UCAV. Like, the problem always starts from sales. These are the bad guys always right.

Eric Dodds 14:41
Now, I mean, that is a balanced way that maybe, maybe it is. Number one, I think it’s a really helpful to think about the various teams as producing. producing products right? I like that’s really, really all on data as a key input to that. And maybe it’s idealistic. But why can’t there be someone who has a team agnostic, like, it doesn’t have to be good engineering or analysts or whatever. But I think to some extent, like ops roles can do this type of thing, even though those also tend to be segregated, right. Like, you know, revenue ops generally tends to like support the sales organization, heavily marketing as marketing, engineers, marketing. But it’s compelling to me the idea of like an individual or a team that has purview over those things, to avoid the pain of like, having to go back and pay technical debt in the product that you’re delivering on marketing, because of the decision that sales made, because they don’t know that what they’re doing creates input problem for the products at market. Am I just an idealistic person cost this?

Kostas Pardalis 16:06
No, sorry. But like, my, my instinctive reaction was like, You’re not idealistic, you probably have daddy issues. Because you’re looking like for a daddy who’s going live to solve the problems for you, like, no, like, you don’t need some are like, has been like to do that, like, you have to be aware of, like, how you can succeed in what you’re doing. And make sure that like you, you know, that like, data comes helps you with that. So you can create the requirements, that then these data folks can execute for you. Right? Like, nobody knows the context or the problem, or like, what is at stake better than you? Right, and car like, and that’s the difference like between the customer. And like building products for like real customers and for in demo stakeholders, right? The difference is that it’s like you as the stakeholder, you should be much more engaged. And you should be much more proactive in like giving the rights hunt, say that, like, the right information. For the people we’re going to execute to build the infrastructure and everything that is needed around that to help us succeed. Right, it’s for your best interest and like you should drive that love that drive, like between going live to an arbitrary, like building a product like to go and see what’s right.

Eric Dodds 17:26
Yeah. Okay. I’m gonna turn this back. I’m going to turn this back on, again, to try to return to my original. Sorry. What do you have daddy issues? This the type of thing you did? This? Is this is why our relationship? No. No, you’re saying that it’s not a centralized data professional that needs to be the politician. It’s actually the person in the downstream team who has a vested interest in delivering their products. That needs to be the politician and politics with all the other functional leaks. Yeah,

Kostas Pardalis 18:11
you’d have to build on those. And only the borrowed legs take care of you. Like that’s, yeah, like, politician are two different categories. But I like how you’re like, Yeah, I

Eric Dodds 18:27
mean, like, a politicking inside of the work. That’s yeah.

Kostas Pardalis 18:33
I mean, jokes aside of the end, like you can build with say, any strategy around data sounds like the stakeholders truly believing that, like, these data are going to show up in whatever they are doing it. Right. I have my like, marketing campaign might be sales campaign, it can be a product, if me as let’s say, a product manager do not believe that, like having data is going to help me succeed. Like, you cannot, I mean, you cannot force like the implementation of lotsa, right. So that that’s what I’m trying to say that. You can’t have just someone who’s going to, let’s say and force the company to be data driven. You need like, every individual is like, responsible of creating initiatives to believe that using data to optimize and build these initiatives is like, a good thing. And it’s not always a good thing, right? Like it’s not always loves that data cannot solve everything.

Eric Dodds 19:33
Yep, yeah. Would you describe that as maybe like a human data mesh? work.

Kostas Pardalis 19:49
I mean, I think that data mess is an organizational signal the technology thing, that’s what the Vonzell use of data Miss say? Right. So yeah, I mean, I’m Yeah, so I think it again, it has to do a lot of like with the problem you’re trying to solve and the like how like educating the people in the value of like using data to solve their problems. And again, not always data can help, right? Like, that’s something that I think we all need to keep in mind. And it’s one like it’s a very, I think, like a very standard example of this is you’re a b2b company. You start having your first signups and you’re like, oh, yeah, like, I’m going to instrument like the signup page, because this is data driven like crazy insights. And you get like three signups a day and like, okay, yeah, sure. I mean, and by the way, drove them are like Gmail accounts, so they don’t even, like, mothered rights. So, yeah, you’re not like, you just don’t have enough data like to go like build your product that way. So, yeah, that’s like, I don’t I don’t I don’t believe that. Like someone should have like the full responsibility of like, let’s say enforcing of the game’s data surroundings in the company. I think it’s, it’s something that’s part of the culture of the company itself.

Eric Dodds 21:04
I love it. All right. Rex is selling ads. It’s time to end and I need to go call my dad. Yes, it’ll set up.

Kostas Pardalis 21:10
I don’t know. And then I think you’re old enough to call the federal prisons a little bit that

Eric Dodds 21:15
fill in at all. Alright, thanks for joining us on dataset Soda Shop Talk. And we’ll catch you on the next one. We’ll have some guests coming up for this form. And also we’d love your feedback. Did you like what we talked about and do you want to talk about it? Give us questions shoot me an email Eric at data section.com And we will catch you on the next one.