Why Niche Data Tools Fail, Broken Hiring Processes, and the GUI vs. Command Line Showdown with The Cynical Data Guy

Eric Dodds 00:06
Welcome to the data stack show.

John Wessel 00:07
The data stack show is a podcast where we talk about the technical, business and human challenges involved in data work.

Eric Dodds 00:13
Join our casual conversations with innovators and data professionals to learn about new data technologies and how data teams are run at top companies. Welcome back to the data stack. This is one of our favorite times of the month. It’s where we get to sit down with Matt Keller, herd Gibson, the cynical data guy, and hear his perspectives that have been shaped through over a decade of time spent deep in the bowels of enterprise data in corporate America. Matt, welcome back as always.

Matthew Kelliher-Gibson 00:51
Thank you. I’m glad to be here for my monthly Festivus. Yes, I had a lot of problem with you people you’re gonna hear about it

Eric Dodds 00:59
I cannot wait. All right. Well, one of the hashtags in this first post is hot takes. So I think we’re going to start out with a bang here. For those of you listening to a cynical data Guy episode for the first time, we have Matt on the show. John and I go through LinkedIn and we pick out posts that we think would rule the cynical data guy up and we do Lightning Rounds. John’s company is called the agreeable data guy. So my goal is to pit them against each other and get differing opinions. And today, actually, I have four posts. So if we can wrap the first three up, we can do a lightning bonus round. Yes, bonus bonus round. Okay, round one. Ding, ding. Okay. This post is actually going to name the author of this post, Evan wimpy. He was on the show, an amazing episode. Go back and listen to it. He’s from older data. And this is a great topic. I actually am so excited about both of your responses. Okay, I’ll read it quickly here. Are there any short sellers on LinkedIn? My feet are peppered with folks pushing their preferred tool or service, sometimes explicitly, sometimes sneakily. I’m as guilty as any. But are there any short sellers or anti sellers? I’d love to read messages like, don’t buy X, here’s why, with some solid reasoning behind them. Okay, I want some short sells here. Cynical data guy,

Matthew Kelliher-Gibson 02:18
anything with AI, no. I would say one that comes to mind is especially since we’re starting to get to the consolidation. If your thing is, like, we do this one really thin slice, but we do it really well. I’m sure it’s selling you at this

Eric Dodds 02:36
because we’re going back. It’s like bundling, the cycling of bundling and unbundling. We’re getting into the bundling stage. Yep,

Matthew Kelliher-Gibson 02:43
we’ve kind of reached the absurdity limit of unbundling. And so bundling is going to start happening. So you’re like, well, we only do one type of ingestion in this specific way. It’s like, I don’t think that’s going to work.

Eric Dodds 02:58
Agreeable data guy.

John Wessel 03:00
This is sad for me to say, because I think it’s a problem. But the data quality that people have never wanted to spend money on,

Matthew Kelliher-Gibson 03:13
standalone data quality, yeah, you’re like, we’re just gonna sell data quality. That’s like,

John Wessel 03:18
and nobody cares. My second one, which is which has one major exception to it, my second one would be places where there’s a really or spaces in the dataset where there’s really strong open source adoption and one or multiple players that also have like been able to like, make the transition commercialized, to bring in there and commute and compete there against those players, and not innovate like,

Matthew Kelliher-Gibson 03:46
that’s a short you got to be really innovative. It’d

John Wessel 03:49
be really good and some unique way if you want to short sell people that have really large open source communities and have already, like, started on the road of like, capitalizing you

Matthew Kelliher-Gibson 03:58
to not lie to yourself about it. More different. No, you have to be like, two, 3x better, yeah, to do that, yeah,

Eric Dodds 04:04
or 10x okay, if you could invest in an area right now, what would it be? Obviously cynical data guys not investing in AI,

Matthew Kelliher-Gibson 04:16
no, well, I think you get back to where it’s like, what are we in? We’re into the bundling stage, right? So I would say there’s two ways you could look at that right? One is, I’m going to invest in the people who are likely to be the bundlers, yep, or the people I think most likely to get bought by the bum bundlers, yeah, at a premium, yes, because the first ones are going to be at a premium. And then as we go down. Yeah, the disc, that’s

Eric Dodds 04:41
right, yep.

John Wessel 04:43
I think for me, a little bit of a different approach to data sharing. I think FTP is finally on the cusp of getting disrupted, finally, and finally, data shape two, two different forms, one like corporate data sharing, like back and forth between companies, and then, like all of them. Like audience sharing, clean room stuff. I think it’s

Eric Dodds 05:02
gonna be that’s gonna be huge Interesting. Okay, we’re gonna, if we ever start a data stack show fund, yeah, right, exactly,

Matthew Kelliher-Gibson 05:10
exactly. It’s gonna be full of people saying, we’re gonna change the world, and me going, I don’t care.

Eric Dodds 05:17
Okay, next post, round two. Ding, ding. I’ve written before that titles are meaningless in data, and all that matters is your capabilities. Here’s a deeper explanation. I’ve been a VP with zero reports. I’ve been a director with 100 data people reporting to me while owning the PnL of a nine figure data business. Lower title, quote, unquote, but way more actual responsibility. Then there are people who get a senior data title after 12 months of work experience. I know a quote, Senior Manager, end quote, who ran data strategy at a greater than $50 billion company after previously leading data strategy at a 100 billion plus company and countless other examples. This is why, in data, anyone competent will calibrate your seniority based on your actual capabilities, achievements and impact, not based on your title.

Matthew Kelliher-Gibson 06:10
Recruiters don’t listen to that.

Eric Dodds 06:16
Just a recruiter this LinkedIn posts say,

Matthew Kelliher-Gibson 06:18
But no, my title doesn’t matter. Well, but it but it but we want you to have director level experience. And you go, yeah, and a senior man, you’re not even getting a call. Here’s

John Wessel 06:27
the thing. So here’s how LinkedIn works, right? You buy the LinkedIn recruiter package, and then literally, you plug in the fields like a certain title range, it’s cert like, that’s how it works, yeah. So I’d love to have a more agreeable take on this, but that’s just how it works. Yeah?

Matthew Kelliher-Gibson 06:44
I mean, from a growth standpoint, I agree that it isn’t like that, you know, and I’ve worked at places where they underpaid, so they threw around titles like crazy, and it’s a problem. Yeah, I think it’s also one of those that, like, when you’re hiring, it’s on you to realize this fact, and not kind of go the lazy title match route. But realistically, if I’m talking, if you’re talking to someone and you’re like, Oh, don’t worry about the title. That’s usually coming from someone who has, like, 10 years of experience and has been a VP twice, and it’s, you know, it’s like billionaires talking about, you know, money just doesn’t really matter. That’s like, oh yeah. I’m sure you feel that right now at $50 billion yeah,

John Wessel 07:26
All right. So no agreeable take, okay, so I think the only like potentially positive on this would be it might be worth like, as far as like title not mattering it’s there are situations where it is worth taking, you know, getting to work in a better situation, getting to work with better people, or a company that’s on a better trajectory, where, quote, like, cool. Let’s just say, like, okay, title doesn’t matter. And, or maybe a startup, you know, there are situations where I do think this is the right thing to do, but it’s not the right one. If you’re in corporate America, and they’re telling you, like, Oh, we’re gonna give you all this responsibility, but we’re not gonna change your title also,

Matthew Kelliher-Gibson 08:06
because in the bigger companies, you could say, like, well, but you know, you’re getting all this responsibility, Yeah, but you’re gonna be paid at what your title is, yeah, right, right. I do think, is it the thing for you individually? And I think it’s more once you’ve had a title, yeah, then I can kind of look around like I’ve been a director. I can go look back, because I can go back and be a director. Yeah, things like that. So I think it’s one of those that it’s like, it depends on where you are and what you’ve done, but like, telling someone that coming right out of school, I’d be like, No, don’t listen to that. Titles equate to your pay band. That’s why they’re important. And in certain companies, they equate to your political prowess and power. So, yeah, it’s not something to be taken, it’s not important, like it’s not going to affect you as a person, but it is something that is in the context of every company and business, and you have to take it into account. I

Eric Dodds 08:53
I agree. Man, that was open and closed. I need to do better on Yeah, you got another chance right here? I have another chance. Ooh, yeah, I forgot. Okay, this is going to be great, something I’m thinking about as I redo my using Git and GitHub with our course. I’ve come to strongly believe in teaching git using a GUI. Teaching newbies to use Git in the terminal seems like the equivalent of teaching them to use R without an IDE, teaching them git with a tool like GitHub, desktop is like teaching them to use R with RStudio. Thoughts,

Matthew Kelliher-Gibson 09:30
I don’t think that’s a one to one analogy. I think it’s inherently limiting. Right to do that. If you just learn it on a GUI, you it’s gonna, yeah, you’ll get up and running soon, but it’s going to be in it’s going to be limiting, in the wrong in the long term. I have less of an issue with doing this in a newbie course than, like, I don’t know where there’s you know, you’re going to learn how to do it later on. Because, like, that’s usually the response when you’re like, well, but you need to learn how to use the command line. Well, yeah, but they. Can learn to do that later. How, when? Yeah, I don’t see that. So we end up getting a bunch of people that, like, you know, they don’t know, they don’t know how to go above a certain level, because they’re just, they’re stuck on these small tools and these gooeys and stuff like that.

John Wessel 10:18
So for data people, I agree. I think I agree with the post, use the GUI. Well, just use Version Control, please. Right? Like that would be my number one thing for data people. It is so common and he’s talking about R so that’s kind of leaning toward data science, and they’re more traditional, like computer science, like engineers. So I would say, yes, just use version control if it’s true. If it’s a developer, like, no, like, you’ve got to learn the command line. That’s but for a data person, like, I yeah, I can get behind that.

Matthew Kelliher-Gibson 10:51
Well, it’s and, like I said, I will stipulate if the choices are no control, or use the GUI control. Please, people, version control. But I would even say for I mean, if you think, if you’re going to be a data scientist or a data engineer or whatever, like, we’ve gotten into this cycle of where like, there is you don’t go to school really, to be a data scientist, there is no like, there is no function or institution that’s going to force you to learn things that are hard and you don’t want to learn. Yeah. So it’s this amazingly consumer driven market which then moves everyone towards the you know, the person has to pick what they want to learn. They have to pay the money for it. So it creates these incentives to make things a little simpler and a lot more accessible, which, in of itself isn’t bad, except for then we get all these tool makers who then turn around and say, like, Oh, we’ve built all this stuff so you can deploy, you know, notebooks or whatever, and it’s like, well, that’s not a good pattern to use, like, well, but that’s all they know how to use, right? Well, you’re just reinforcing, you know, the kind of, like, immature patterns of this. We’re not unlearning and then moving on to higher things.

Eric Dodds 11:59
So you’re gonna short gooey for version.

Matthew Kelliher-Gibson 12:04
I mean, given our current state, I don’t think I would store it, to be honest, but I just, I don’t see where all the incentives are now kind of lined up against it, where it’s like, hey, how do you mature in your development practice? They’re not there. Everything is like pushing you towards doing these things that can work but aren’t great in the long term. I want

John Wessel 12:25
to take this a little bit different direction, combining the title thing with this question, Matt, what is a data scientist? Oh, you know,

Matthew Kelliher-Gibson 12:35
I don’t have an answer. It changes so much I don’t completely know at this point, though. I believe it has something to do with making a predictive model of some I

John Wessel 12:44
I mean, this is, like, a real problem I’ve been facing. There’s been talking with somebody that’s trying to hire, and they’re like, I think I want a data scientist, but I might need a data analyst, but they but I wanted to be, like, a good data analyst, yeah. Like, it’s a real problem. And, like, there’s this range of, like, PhD, like, level, like, statistics, machine learning, you know, like, I can build models from scratch, and then there’s this, like, I think I’ve heard the definition before, a data scientist is a data analyst that lives in California. So, like,

Matthew Kelliher-Gibson 13:15
there’s a static Silicon Valley. So, so,

John Wessel 13:20
yeah, yeah. I mean, like, there really are bright people that, like, or, I mean, nothing wrong with data analysts. It’s more of a data analyst type role.

Matthew Kelliher-Gibson 13:26
Oh, and I think part of that is also like, data scientist was originally, when it came out, it became webmaster, it was everything, yeah, sure, yeah. And then it split off to data engineer. And now we’ve got, like, the, you know, like ML engineer and things like that, or the BI developer or those things. I still don’t think it’s completely settled down, but I also think there’s that gap between, because data analyst is another one that means this whole wide variety of things. Data analysts can be like, I manually update data in Excel and then email it to people. Yeah, sure. That could be a data

John Wessel 14:02
analyst, or it can be, like, I’m doing, like, really complex analysis, and, like, even, like, regressions and some sad stuff that would probably be data science and some right,

Matthew Kelliher-Gibson 14:11
or, like, a decision analyst type of idea, yeah? So, yeah, there’s a lot of ambiguity. I still haven’t figured it out. All right.

Eric Dodds 14:18
There’s this really great I guess it turned into a meme I haven’t saved on my computer, and Brooks, we should dig it up and put in the show notes. But going back to the title thing, and you mentioned, you know, webmaster, there’s this great image of Tim Berners Lee and this guy from AOL. So this is a while back when AOL was at its peak, right? And they’re talking about technology on some talk show, you know, as guests, right? These are expert guests, and under their titles, they had their titles under, you know, it’s like, okay, you know, so and so title, right? And so, this guy from AOL, also, he was, you know, he had, like, he was. Is dressed super provocatively, right? Like he had, you know, his hair was spiked up and, you know, died and whatever, which is great or whatever. And Tim Berners Lee just looks like a dad, yeah, the title for the AOL guy was like, internet profit, I think. And Tim Berners Lee was just a web developer, which is so great. I’ll dig it up.

Matthew Kelliher-Gibson 15:24
I’ll see if I can dig it up. The important thing to remember, data scientists do pay more than data analysts. Well,

John Wessel 15:29
I was gonna speaking of titles, yeah,

Eric Dodds 15:31
for sure, that’s the thing. Okay, I think we have time for a lightning round. Brooks hasn’t messaged me saying, land the plane yet. Oh, man, this is a good one. Okay, I’m just going to read the interview. I’m going to read this graphic, and then I’ll read the post. Okay, so the graphic is a screenshot of a document. The h1 is the interview process, and then it’s a list of bullet points. So interview process, application review, recruiter, screen, hiring manager, screen, virtual on site, interview, one, interview, two, interview three, final technical interview references offer, and the post is this many rounds of interviews only tells me that there is no trust between employees and management. The company seems to believe that HR and the hiring manager will fail at their job. So they require three more interviews, because the people conducting the first and second interviews will supposedly fail as well. And of course, you need one more technical interview, because apparently the person conducting the third interview is also expected to fail. So

Matthew Kelliher-Gibson 16:38
first comment, you know, that seems to believe that HR is going to fail. I don’t trust HR in any of these. I have never worked. I’ve hired a lot of people. I’ve never worked with an HR that was like, oh yeah. Or, you know, the internal recruiter type that I’m like, oh yeah. You got this. You know what to look for. I want you to know how to know what would be a good fit for this role. We’re cool now they know outside of that. I mean, that’s really similar to what I used to run, and I would and I hemmed that down from what a lot of people did. A lot of people try to stick to it, like nine people interviewing there, but you need to have some sort of process in this. I don’t know what they think. We’re just gonna we’re gonna like, you know, we’re gonna vibe this and the hiring managers be like, you seem cool. We’re gonna vibe this one and be like, yeah, you seem cool. Like to have a beer with you. Here’s $200,000

John Wessel 17:44
we’ll see if it works out. YOLO, agreeable. Oh, man, I’m still trying to understand the situation here. I guess the complaint, it seems like it’s twofold. One, it’s too long. And two, like there’s an assumption here where, like, the first two steps are done by, like, HR, recruiter group, and then like, the hiring manager, and there’s kind of this, like, pass off back and forth. So first point, like, it’s too long. Like, maybe, like,

Matthew Kelliher-Gibson 18:17
if it’s too long, it’s too long by, like an interview or two, like, it’s not, you’re not cutting this in half, right? It’s

John Wessel 18:24
not seven interviews, maybe I but something I have heard again talking to people trying to hire right now is there is way less patience with candidates with long interview processes for whatever reason. Yeah, which is interesting. I

Eric Dodds 18:38
heard that recently too, a good friend of mine runs product at a great company, and he said, we’re, we have to move so fast to get talent, which basically means it’s a full time job, because you’re just, you’re running multiple people through a process as fast as you possibly can. Well, I

Matthew Kelliher-Gibson 18:55
I think that’s the other part of it. Is if you’re taking three months to go through this process,

Eric Dodds 19:00
yeah, sure, whatever. That’s the problem. Yeah, that’s a great point. Yeah,

Matthew Kelliher-Gibson 19:04
I can do that in a week to week and a half, right? Yeah. So, I mean, it’s not, if it’s important and you’re doing it, you should be moving fast, right, right? It’s not her job, like there’s a lot of that, where they’re it’s the recruiter’s job or whatever. No, this is your job to hire. Yeah.

John Wessel 19:19
The other thing where this gets sloppy is handoffs. I do think with that many steps and the handoffs back and forth, like screen, like recruiter pass up to the manager, like, all those handoffs, if it’s not, like, really clean, then it will take months to go through that. And

Matthew Kelliher-Gibson 19:34
If you’re the hiring manager, that’s your job to make sure that it’s right. I mean, I think there’s an assumption a lot of times, like, well, that’s her role. Like, no, it’s your team. And I think a lot of times when you get the really long ones, part of what you’re getting is we don’t have a process, and we don’t actually have, like, a well defined framework of what we want from a role. And you see that partially when they’re like, you know, the job description is really generic, yeah, whatever. And I don’t. Really know what I want, so I’m going to throw more people out of it. We’re going to get into a room, and then we’re going to how did you feel about it? And it’s going to vibrate again at that point, right?

Eric Dodds 20:12
Vibe. HR, also, great name, yeah, the domain names available. We can start a company.

John Wessel 20:17
The other thing he mentions, like, fails a bunch of different times here. Like, you know, I think that’s probably not fair. Like, the between the application review, recruiter screen, if all of this happens quickly, that would be my main measure of success here, to be honest, right? Like, are you like, like, you said, Are you going to be able to give a recruiter HR manager? Like, hey, here’s what I don’t need. I trust you. Like, I’ll hire whoever you say. Like, no, nobody would do that, right? Or, almost nobody would do that. But a lot of

Matthew Kelliher-Gibson 20:47
times you’re trying to translate, hey, here are these mindsets or skills that I want to hire for, and they’re like, Okay, what keywords do I look for? Yeah, yeah, sure. I don’t have keywords for you, yeah. But

John Wessel 20:59
Hiring managers can be really lazy writing job descriptions, which makes it really hard to recruit. Well, yeah, the

Eric Dodds 21:05
best, the best, I’ve worked with multiple internal recruiters, and the best one was actually a coach. They’re like, okay, you know what you want, and I’m gonna tell you, like, here are all the things you can do. You know, interview, like, all, it was like a coach, like an advisor who is like, here’s, here are different frameworks to run in the process. Here’s what to look for. It was like, This is great, right? It’s like an advisor who’s helping you, like, you know,

John Wessel 21:34
Well, and yeah, especially somebody that can coach on the like, what makes a good employee things because you like, as a manager, you have specific technical skills you want, maybe even, like, soft skills, but it’s like, Hey, I’ve been doing this a long time. Like, this is the process I use. Here are the things that, for me, like, make a good employee at this company for these reasons. You know that I think that can be really helpful, and that’s like

Matthew Kelliher-Gibson 21:57
A second level hiring thing is the idea of a lot of people who view it almost like you’re building a baseball team, right? I’m collecting talent. Yeah, they got different positions, but they don’t interact that much. There’s a handoff, and they’re kind of specialized, or whatever, where the reality is, like, if you’re a small team, you’re really making a basketball team, and it’s got to have a high level of like, I understand, and you and we work together, and we can, you can’t just throw people into it to be effective. And if you’re running a really large team, you’re running a football team at that point. Yeah, and you got to really define what alignment looks like? What is a wide receiver? Yeah, like, I mean, and they still have to play together, yeah, you’ve got to really work at it to play. You got to drill. There is no such thing as a well oiled machine, because you don’t have to train a new iStan how to work with it, it doesn’t work. Yeah, right, yeah. And I’ll give and kind of with this, also give kind of my free advice on this, because I did a lot of work on because hiring processes suck most of the time. If you’re going to hire someone to make this all effective, it all starts up front with that idea of, like, what are the three to five things you have to have? Yeah, because everyone else is going to have a but unless you’re Google or one of the big ones, you’re going to get people that have, like, a well, but so you need to know those three to five things, yeah. You need to hire for the strengths, not for lack of weaknesses. Yeah. And you need to, like, really hone in on what those things are. And then when you get into it, it’s, you know, you got to remember, we’re not collecting talent. It’s got to fit in with this team that we have. And what is the role that we need? Not just, I need an analyst, I need an engineer, whatever. But like, how would they have to fit in with the rest of our team? And how are you going to actually evaluate that process? Yeah, because that’s why you have people on your team, interview them,

John Wessel 23:40
yeah. And in addition to that, like, who are you as a company? Where are you at? And, like, is your role scoped? Well, like, in, like, Is this a real job, right? Because, like, especially with like, fast growing companies, oh, we’re gonna, like, hire this new role. And then, like, we’ve never had one of those before, yeah. And nobody spends any time scoping like, what are they going to do? How do they fit in with the team? Yeah. And then, like, a lot of times they completely miss on budget. They like, want to hire for like, 50% of market right, like, market rate. So if you do those two things wrong, like that, that creates train wrecks. Because often, like, you’re hiring maybe even for, like, a director, more senior position that you scoped, and then you didn’t even, like, scope it well, and then you’ll have the budget for it. Like, so can then pretty badly. So

Matthew Kelliher-Gibson 24:23
A quick story on that early in my career, I worked for a company that did not pay well, okay, but like, you know, whatever it didn’t pay well. They mostly hire out of school. But the VP who had a degree in, like, journalism, to give you an idea of what we’re dealing with, she really wanted to hire people who had, like, Ivy League degrees. And we looked at her budget, and I remember telling my boss at the time she, you know, she’s like, she wants people from Harvard and Stanford. I’m like, based on what we are paying. I don’t want that person from Harvard or Stanford, if they’re willing to accept that they. Are in like, the bottom of that class, right?

Eric Dodds 25:01
It’s like the doctor who graduates with DS, but they’re still, you know, still a doctor, a doctor one slightly, since we’re still getting no warnings from Brooks, I’m actually, as the show host, going to interject a slightly cynical take on this one. I get what this guy is saying about trust, right? And it’s like, okay, well, if you don’t trust, if there’s a lack of trust between the people who are doing these things, you’re just going to add a process to try to control that one. Yeah, right. So that’s not untrue, but my gut reaction when I saw this was, you have never raked over the coals of your own horrible hiring decision, and felt the pain of how much money that cost you. Like,

Matthew Kelliher-Gibson 25:46
I mean, not even like someone else doing it, just like sitting there and going, like, I made my team worse. Yeah, I have a giant headache, totally.

Eric Dodds 25:53
Or you’ve never inherited a team before, yeah, totally. It’s like, I am working two jobs because this was just the hire, is such a misfit, right? And, like, brutal if

Matthew Kelliher-Gibson 26:05
you’ve never gone through the hiring process like leading it on the other side, it’s really easy to sit there and be critical of it. It’s really hard to do in real life. It is really hard. It is, and most people solve that problem by trying to outsource it to the recruiter or HR,

Eric Dodds 26:21
yeah, yep. Okay. Well, easy summary, short sell. If it’s not open source, always chase the title and use it gooey, forget. So this has been a great one. Thanks for joining. We hope you got a couple laughs out of the show today. We’ll be back again next month with a single, cynical data guy. Subscribe if you haven’t, so you get notified of new episodes, and we’ll catch you on the next one. The datastack Show is brought to you by rudderstack, the warehouse native customer data platform. Rudderstack is purpose built to help data teams turn customer data into competitive advantage. Learn more at rudderstack.com you.