The Data Stack Show - 104: A Decade of Change in the Data Space with Benn Stancil of Mode

Episode Date: September 14, 2022

Highlights from this week’s conversation include:Benn’s background and career journey (2:28)The problem Benn sought to solve (4:48)Data engineering a decade ago (9:58)Technology inside vs. outside Silicon Valley (18:11)What’s next for data (24:42)Mode’s evolution and journey (29:31)Challenges of getting enough context to create (39:21)Current trends that won’t see long-term benefits (48:44)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. Hey, Data Stack show listeners. We have another live recording coming up next Wednesday, September 21st at 1.30 Eastern, 10.30 Pacific. This time, we'll be talking about building modern data teams with a great lineup. Sri, the head of data

infrastructure at Robinhood, Paige, a data analyst who was on the Netlify data team, and Sean, a cloud data engineer at REI, are all gonna join us. They've all been on the show before. We can't wait to have them back to tackle this subject because they're gonna bring a really diverse perspective here. They've all been on different teams in different roles.

So they're really gonna help us understand what works and how can we build high performing teams in such a rapidly evolving space. So visit datastackshow.com slash live to register today for next Wednesday, September 21st at 1.30 Eastern, 10.30 Pacific. Costas, today we're going to talk with what I think is actually a really interesting company in terms of the context of both the problem they're solving and then how long they've been doing it. So we're going to talk with Ben Sancil from Mode. And they've been around for almost a decade

doing data visualization and sort of analytics workflow stuff, which is super cool. Ben writes a very popular blog. Lots of people, including myself, love to read his thoughts on everything. And this isn't going to surprise you, but I want to know why he started Mode.

There were some huge enterprise incumbents when he started Mode. It was just right at the genesis of the cloud data warehouse, which is a really interesting time. You actually have similar experience with this, starting a sort of data company right around the same time. And so I just want to hear about that. What was he thinking? What was he doing?

And what led to him starting it? Yeah. I mean, for me, it's a special episode, to be honest, because Ben is one of the first people that I've met in person while I was still building Blendon. Really? Like on your first trip to San Francisco? Yeah.

No way. He's a person that, I mean, he's kind of special to me. And also he's a person that I really appreciate both his opinion and the way that he expresses his opinion. Because that's also important. So yeah, I'm really looking forward to chat with him today and talk about all the things that have changed these past 10 years in the industry. And I'm pretty sure we're going to be surprised by his thoughts and opinions. I agree. Let's dive in. Ben, welcome to the Data Stack Show.

Super excited to chat with you today. Thanks for having me. Excited to be here. Okay, let's start where we always do. Give us your background and kind of what led you to what you're doing today at, at Mode. So I am one of the founders of Mode. It's a, basically a BI product built for analysts. Can get into more of specifically what it is in the conversation. Mode's been around for a little while.

So we started Mode in 2013 or so. So one of the kind of early cloud data tools or modern data tools that we now have plenty of. Prior to that and kind of where it originally came from, I worked on a data team at a company called Yammer along with a number of other folks. Three of us left to go start Mode. So basically got a lot of exposure

to how people were thinking about data. That company saw some of the things that we thought might make an interesting product, an interesting business. Started Mode. And then since being at Mode, it's kind of bounced around from a number of different things. Most of my time has been spent in either thinking about our own internal analytics

and kind of data infrastructure and spending time on kind of the landscape and thinking a lot about like what's going on in the data world, the data community, what it's talking about, what kinds of things are the trends that are melding everybody else's writing, what might be popular, what won't stick, that kind of stuff. And then in addition to that, as you tend to do at startups, busting around for other jobs and being involved in marketing or customer success or solutions or products or, you know, running the exec team. Whatever kind of stints you have and kind of your various rotational programs over the course of eight or nine years as a startup. Yeah, absolutely.

I mean, that's a long tenure for a startup. So congratulations. I'd love to actually dig into what you saw as the problems you wanted to solve, right? I mean, even back in 2013, the cloud data warehouse is still fairly new, kind of emerging. But there were still massive incumbents in the analytics space, right? So you had, you know, whatever Tableau and like visualization had sort of some major incumbents at like the enterprise level. So what did you see that made you say like, okay, I want to go build something new because there's some sort of need

or like something that's not, you know, some itch that's not being scratched in the market. Yeah. So the basic answer to that is the problem that we had at Yammer. So we were a data team at the time Yammer was, so Yammer got acquired by Microsoft in 2012. We were, we were, the company was about 400 people or so where it got acquired. We had a data team of 20, roughly,

including data engineering. And we represented this kind of like new version of what data teams were becoming in Silicon Valley, where our job was to work with other people who were in the business to help them make decisions. We were not a BI and reporting team.

Like we weren't there to just build dashboards and sort of binders that execs are supposed to look through for their weekly meetings. But we also weren't a to just build dashboards and sort of binders that execs are supposed to look through for their weekly meetings. But we also weren't a team full of statisticians. And we're trying to work in sort of capital D data science type of places. And so what we're trying to do is like sit next to businesses, like people in the business and say, which marketing campaigns should we run?

Which products should we build? How do these AV tests perform? That kind of stuff. Yep. The way that we needed to do it was we wanted tools for ourselves who were relatively technical data analysts. So technical in the sense that we could write SQL,

R a little bit, a little bit of Python kind of stuff. Yep. But we also knew how to share that stuff with other people. Our job was very much business oriented. We weren't deploying code to production or anything like that. And so the tooling to us was either the traditional BI tooling that you're talking about. So Tableau was kind of the cutting edge of that, but also all of the micro strategies and business objects and those types of things. Or it was stuff that tilted very technical. So basically statistical tools,

data and SAS and those kinds of things. Or RStudio, Jupyter notebooks were around that point. It was like Python, I think, at that point. Yeah, like visualization or really gutsy, heavy-duty stuff. Yeah, and while we were like desktop SQL editors, like DataGrip and that sort of stuff, SQL Workbench, those tools would have worked for us, the more technical ones, because that kind of fit what we were trying to do. But we couldn't actually share any of that stuff with anybody else.

Like it didn't work for the rest of the business because we can't send to the CEO, here is an IPython notebook, like, you know, here's the instructions that have sent up and run it. We couldn't send them a SQL instructions that have to run it. We couldn't send them a SQL query and have them run it. We couldn't send them, like, we weren't going to pay for SAS because we were a startup. So all those things

didn't really work for us. And the BI tools were very aimed at, like, these non-technical folks were just like, our job is to build dashboards that you put up on a TV screen. Like, that's not the job

we wanted to do. So we basically, like, we need to do this ourselves. We need to, we ended up the team at Yammer built an internal tool that looked when it essentially like a very strict down version mode, that was basically a SQL editor in a browser with charts on top. And so what we would do is we're like do analysis and SQL write queries, save those like charts, send them to people and be like, you know, here is why campaign

A did better than campaign B or here is some analysis that tells their story. And so what we ended up seeing was two big trends that were emerging around this time. This is more or less why we left to go to start mode. One was cloud data warehouses. So at Yammer,

we were spending a whole bunch of money on data infrastructure. We had a Vertica cluster. We were probably spending half a million to a million bucks a year on it. Like that was kind of par for the course. If you're using Vertica or Oracle Teradata, whatever. The other thing is we, we had this kind of team that we thought was specialized. It was like, this is not many companies have teams like this.

Both of those things changed where cloud data warehouses made the cost of the infrastructure way cheaper. So you could run Redshift for a few thousand bucks a month. And then the team that we represented, this kind of like analytics team that thinks about data, not as a reporting function or not as like an actuarial function, became much more of the way that a lot of companies thought about data. That people started pulling off what was the Facebook model, the Airbnb model, that kind of stuff and saying, our data teams are not IT teams.

They are not also a bunch of statisticians in the basement. They are like people to help the business to kind of deploy around the business and to be data experts to help people make decisions. And those were really the two trends that we were running. Yeah. Super interesting. Okay. Question for you so like the term data engineering is actually like relatively young how did you refer to the team back then because that was sort of you know maybe say like before data

engineering became as formal a term as it is now in terms of like you have a data engineering team right or like a head of data engineering role or whatever. I'm just interested to know, like, I mean, so that's like a decade ago, right? How did you refer to it?

Because it also sounds, what's interesting is like, it also sounds that at Yammer, the data team was operating in a very forward thinking way. You know, it's like, we've heard it called like structured embedding or something, right? like, we've heard it called, like, structured embedding or something, right? Where you have, like, someone from a data team,

like, sitting with someone, you know, in a business function and, like, working on that stuff. So, just interested to know, like, the ways that you thought about it and maybe even the terminology you used, like, a decade ago. So, actually, the terminology

was easier then than it is now. We... Yammer was a... Yammer was, like, a relatively forward-thinking company. the terminology was easier then than it is now. We, Yammer was a, Yammer was like a relatively forward-thinking company. I will give it credit for that. And in a couple of particular ways. The two kind of, these are the ways that we are writing are sort of like the data moving to the cloud

and like the rise of these sort of new data teams taking new ways. A couple of things that Yammer was writing was what was referred to as the consumerization of IT. So basically like business products are getting built like consumer products,

where instead of it being a thing that you build and you go sell IT and like nobody likes, but IT does because it checks various boxes, you should go build something that the end users themselves

will like and then pressure IT to get. The other thing that Yammer did, sort of pioneering, kind of a little bit of a wrong word, but was one of the early companies in, was thinking about data for a business the way the consumer companies did too, where we did a lot of A-B tests. We did a lot of this stuff that was meant to say, how do we build a better product, a stickier product, a more viable product in the same way that really like gaming companies had actually pioneered, you know, like Zynga's in the world where a lot of the people who really thought about this stuff and in sort of very robust ways. Yeah. And we were basically picking off some of the easy stuff from what they had done.

One quick question there. Do you think that what Yammer, and actually maybe this sounds funny because I remember Yammer, I remember using Yammer, but do you mind just giving a quick explanation of what Yammer is for our listeners who might not know or what it was who might not know? Especially as it existed pre-Microsoft. Yammer, yeah. So Yammer was Facebook for work before Facebook for work existed. It was like quite literally, it was intended to be something

that felt exactly like Facebook, where you have a feed, you've got groups, you've got like the experience was meant to say, hey, Facebook works really well for getting like a bunch of viral adoption around your friends. What if we do the same thing for work? In effect, I think it's interesting. The product actually worked really, really well if you used it right. Because I mean, right. If you used it like, it basically went all in on it. Because it was, I am a, personally, I hate Slack. I think Slack is a disaster for society.

It was, but it does some things well, which is like unstylos communication or whatever. It gets everybody in one place. It has some nice stuff, like a bunch of email threads are bad about, but it does it in a way that it's just this like 40 fire hoses that are constantly hitting you in the face to the point where you just mute everything and like pray.

Yammer actually worked a little bit like how Facebook actually works, where you can check Facebook. And if you log, like at least how Facebook used to 10 years ago, if you log into Facebook and like kind of read it 20 minutes a day, you'll end up picking up pretty well on like what people are doing. You suddenly have this ambient awareness of what's going on with all of your friends. Because you don't see everything they say, but you'll see this, you'll see this picture, whatever. And that was basically what Yammer was trying to capture was create the same kind of feed that you're not supposed to read everything. But if you check it periodically, you'll see a bunch of conversations from other people around the business. It's like, oh, I see this thing about marketing.

If it looks interesting to me, I will kind of read the thread. So it's actually like a kind of digestible thing, similar to like reading the news news but it was very much like the product itself was very much like facebook's work like it was just a feed threaded messages out yeah it's just super interesting because it's when you think about that happening a decade ago it's a really interesting context of having a b2b like sass product in terms of like the way it's going to market, but like delivering the product is very much a beat, like a consumer experience, right? It just happens to be like inside of a business. Do you think that that dynamic sort of helped catalyze some of the ways that

you were approaching, like working with data where like it was a B2B company, but you had like a very consumer mindset? I think it did. I think it, yes and yes, I certainly did. Like that was certainly, you know, me and the other two folks

who started Node as well, some of the early employees who came from Yammer all had the mindset that we were kind of instilled with from Yammer, which was around this idea of build a consumerized product,

think about marketing in a more consumer way, think about marketing in a more consumer way, think about growth in a more consumer way. I think that served us well in some respects and badly in others, to be entirely honest. And I think like- Interesting. That still applies to data companies today. And I think we've moved a little bit past this idea of like,

you know, growth hack your way into success, which is where that goes. That was like at the height of this little growth hacking oh yeah yeah i remember that would you say that's like what the negative part was like it served you well in some ways but like not in others what were the not in others i think it so i think it serves you well and like you think about the end user you think about building a product that people like like the product is the focus. The downside I think is businesses buy things differently. And like you develop kind of a disdain for traditional marketing and like the value of the

pay to have a coffee with a CIO and talk to that person. And like there is just some like write the white papers, do the stuff that people have done in sort of B2B marketing forever, talk to analysts, do all of that kind of legwork that you're sort of like, we don't need that. If we have a good enough product, it would sell itself. Right. Or better than that. Exactly. I mean, yeah. And there are every once in a while companies that seem to do this, that you're like, oh, of course we can do that. Like Slack to some extent did this. But it's hard because these products, there's sort of a natural limitation to how

viral they could be. Businesses are still like the purchasers are still often IT. There are a lot of concerns that businesses that are not especially data or like security and privacy and all those sorts of things matter where you're not selling a consumer product. Like you could kind of think of it that way, but you're not. And so I think that there are some places where like early we had this idea, like, yeah, we'll build a viral thing. We'll focus on all that. And I think there is some legwork to be done

on building an actual enterprise grade products that you have to start on pretty early to do it. And thinking about like sort of growth hacking or whatever you want to call it as the actual path of success i think i think it's a little bit of distraction yeah super interesting okay costas i gotta i'm i'm gonna hand the mic over to you because i'm having too much fun please please jump in i mean you can you can join the conversation whenever you want but yeah it was very interesting

like to listen to both of you and like i have a question based on the stuff that you were discussing towards the end. So you've talked about B2C, B2B, growth hacking, viral growth, and all these things. And what comes to my mind is, as a question question, like how things are different about like the stuff that we talk and care about in the valley, in the Silicon Valley and outside Silicon Valley. Because, okay, we build, let's say products in a kind of like bubble, right? Like it doesn't mean that the rest of the world out there like operates or or works in exactly the same way. And I think this is even more evident when we are talking about B2B. Because, okay, B2C is a different kind of beast in how you can, let's say, approach people and do marketing and

selling and all that stuff. But with B2B, things are different in terms of how fast things can change and behaviors can change. Ben, you said some things about the CIO drinking coffee with them, talking to analysts. All these things are still there. And even with the success of Snowflake, it's not like Snowflake didn't do that stuff anyway, even if they had a more product-led, let's say, growth strategy or whatever. So based on your experience,

like how different are things inside and outside Silicon Valley when it comes like to building technology and selling technology? I mean, I'm in Silicon Valley, so I have no idea. Like my general take on that is there's like a couple of ways, which is very different and a couple of ways it's actually not. I think there is this sort of impression. And you see, if you read political journalism, there's a lot of political journalism about the D.C. elite going out to the diner in Kansas and talking about this is what real people think. And they mostly think the same things that people in dc think they don't pay as much attention but they're not like sitting around you know the only thing i care about is like gas prices

and that's it like they are entertained by sort of the political theater the same as everybody else's i think the same is sort of true for for tech where they don't spend as much time thinking about this stuff they don't have podcasts about data infrastructure tools. Most of these people are not driven that much in their jobs by, I care about the craftsmanship of the technology that I am using. But I think a lot of ways their problems aren't that big. These businesses are trying to solve the same sorts of things. They see the stuff that people in Silicon Valley have done just as startups look at Airbnb and want to emulate some of the stuff that they did in their success or Facebook or Uber. So does such and such company in Dallas that is trying to build a similar business. They care about what Facebook is doing

and they talk to the Gartner analysts. The Gartner analysts are looking at what Facebook is doing. So I don't think it's that. Those things aren't that different. They want products that work. They want products they enjoy using. They want products that make them not miserable when they have to use it. But I think it's been a lot less time talking about it. And to me, the biggest place where that actually manifests itself is there are certain things that we probably, like in Silicon Valley,

my contention is we overvalue certain things because we ourselves value them. Like craftsmanship and software is part of that, or sort of philosophical stances about software. And one of the ones that I more recently have been thinking about is modularity of the data stack. This has become a thing that is like a little bit of a best practice about building data tools is you want stuff that's modular. You want things where you can like plug and play and you can have kind of a Mr. Potato Head type of deal where you choose the mouth and the ears and all fits together really great. And like philosophically great.

I think that makes sense. I think everybody wants that. But if you don't care that much about it, you actually, there's a thing that's even more valuable, which is you could buy all the pieces of Mr. Potato Head in one package. And like, I don't have to go to 10 stores to buy a leg and an arm. It's like, ah, I do want the nice leg and i do want the nice arm but it also saved me a lot of time to just buy them in one place and if like this place has like the discounted leg but it works i'll take it and so i think there's stuff like that it's more around how people buy software the things they value that is different but i don't think it's like some crazy different world

where they're you know where aliens and they're not or whatever, like it's, everybody's still trying to kind of solve the same problems and fundamentally driven by like the same principles. Yeah, no, makes total sense. So, okay. You mentioned like the modularity and of like the modern data stack or whatever we want to call it. Do you think that the reason that we ended up with so many

different, let's say, tools that you have to use in order to build, let's say, this stack is mainly what you mentioned about this kind of software engineering kind of mentality in Silicon Valley? Or there are also other reasons that contributed to this unbundling of the data stack or whatever like it's called. What are your thoughts on that? I mean, I think we have all of these tools because it was easy to raise VC money. Like, I mean, I think it was basically that Silicon Valley poured a whole bunch of money

into the space. You could start a company and raise 20 million bucks with a pitch deck. Silicon Valley, poured a whole bunch of money into the space, you could start a company and raise 20 million bucks with a pitch deck. Especially if you were someone who came from a reputable data team in Silicon Valley. If you were head of data infrastructure at X notable startup,

VCs would throw money at you. And there's this kind of interesting dynamic too, where if you are, say you are a data person, your career is sort of capped. Like it's hard to know where to go. It's hard to know where senior director of data infrastructure at, you know, what's that kind of, I'm suddenly blanking on the comedy.

Who cares? Some $10 billion startup goes. You know, say you that kind of, suddenly blank on the comedy. Who cares? Some $10 billion startup goes. You know, say you work at Coinbase, the senior director of data infrastructure at Coinbase. Where do you go next? Like, what's your next step? I don't know, actually. Like, you move into just sort of like

general business leadership, maybe, but like as a data person, the career path is sort of like not there yet. Like, I think it will be, and I think these folks can become just generally leaders of the business or whatever. But if you're an engineer, like you can move up to being, you know, VPs and CTOs and things like that. If you're in sales, you can move all the way up to the top or you can just make a whole

bunch of money as like an IC sales rep. As an analyst or a data engineer, it's like, ah, you kind of hit the ceiling and then your boss is now a CTO or your boss is the VP of finance. And like the data engineer does not get promoted to being a VP of finance. And so I think in combination of that sense, plus the fact that there's a whole bunch of money out there and it's clear that like the market was basically willing to accept anything, drew a lot of people, and it doesn't take that much to really saturate the market with tools, but drew a lot of people who were at that point in their careers to be like, yeah, I'll start a company. Why not? Like these things seem like

they basically hit the ground as successes. It gives me a chance to do a thing I've never done before. That was part of the motivation behind why I did it was like, I'm a data person, but I'm never going to be on the ground floor of a startup. Like you don't hire data people until like employee fifth. And so I think it was kind of a why not. And so as a result of that, we got a lot of people who were looking for small problems to solve or trying to solve like the small wedge of an issue that they solved within their other companies, which is, I'm not sort of putting any moral judgment on that. Like that's fine. That's what exactly what I did. But when you do that a thousand times over, you end up with a whole lot of tiny wedges that can't be viable businesses unless they go into big wedges.

And then you have yourself a problem because all of them are going to be there. Okay. This is like some excellent point. So, okay. The career path is not there yet. So people in data, they're, I mean, they will be like, let's say, stuck in this ladder for a while more. But at the same time, I mean, it seems at least that like there's not that much money anymore out there, right?

Like, so what do you think that it's going to happen next? I mean, there's like already quite a few companies out there, like pretty much every category has a number of vendors already. And what's next? Like, okay, for the VCs, I mean, okay, they have their portfolio anyway. Like I'm not that much worried about them and their returns, like they will figure it out, but like for the founders who have started like a data company out there, like, and especially first time founders, what your intuition says that it's going to follow? Like, are we going to have like mergers? Are we going to see like companies

shutting down? Like, what do you think? I mean, it inevitably hit to some extent. Yeah. Like, and I don't know how bad that'll get. Like I, if I knew that I would have a different job and be making a lot of money doing that. It's interesting because, so if you go back to like the prior tech bubbles and things like that, obviously there was a lot of pressure on startups then and a lot of them shut down. The two things now that feel very different about it is, especially in the data world, is a lot of these companies have a lot of money. Like they went out and raised a whole bunch of money and probably haven't spent that much of it. Now, there are some that probably raised money a year ago that ramped everything like crazy.

And those are the ones that are going to be in a little bit of trouble where like, oh my God, I thought I was a company that had a hundred million dollars in the bank and it was worth a billion dollars with 5 million bucks in revenue. And now I have a 200 person team supporting that and that's way too big. And the market actually today would value me a hundred million dollars. And like, what do I do? Okay, that's like getting out pretty far,

like in dangerous territory. But if you raise money six months ago, say you raised $100 million and you're not worth a billion dollars, company's not worth a billion dollars anymore. If you were to raise today, it'd still be worth 100 million, 200 million, whatever. But you have a whole bunch of

that money in the bank and you probably have to change your business that much and so really what you're doing is like well now let's just make our runway six years and you could do that and so i think like that's a very different world where a bunch of companies have runway for six years to be able to get through something like this in a case where like you have runway for two and you're desperately trying to figure out how to like basically catch up to your to your valuation so that you're not completely underwater. So that's one dynamic. The other dynamic is I think the data industry is still really big.

Regardless of what happens with a recession slash correction slash whatever you want to call it, these companies are still selling to something that's really big. I don't think it's big enough to support all of them. Certainly not to support them at the valuations that they had six months ago. Certainly not like all of these companies that had their wedge. This is a company that's worth a billion dollars because that's a wedge that we think can turn into something that generates $100 million in revenue. There's a whole lot of wedges that are going to overlap with each other. But the industry, I think, is big enough to support a lot of this. And so my guess is the ones that are in relatively good financial footing will figure out ways to come out of it. They won't all necessarily be like $10 billion companies or

anything like that. Being a billion dollar company is actually very hard despite the last couple of years making it seem quite easy. So I think it'll basically just be like, that stuff is hard again. But I don't think these companies are going to necessarily die. They're just going to be a lot of, a lot of VCs that make like pretty middling returns on some aggressive investments, I suspect. Okay. Okay. That's, that's pretty cool.

So, okay. You've been, I mean, Mold started like in 2013, like today's like we're in the 2022, many things have changed. And you have also gone through like, I mean, back then, like we were talking about like the BI market. There was a lot of competition there. We had like Looker, we had Periscope data, we had like a whole group of like companies that they ended up either like getting acquired. Many things happened in this market, right? Like it was like very interesting for me to observe as an outsider, to be honest,

because I wasn't working like an FBI too. But Moot is still there and obviously like has done like has evolved, right? So can you share a little bit with us this evolution and also the journey through this evolution together with how this is affected by the market out there, right? Because you don't build a company in the product like in a vacuum. That's why I did also all this introduction about how the market has changed. So yeah, I'd love to hear from that because like, that's like a pretty unique experience that you have.

And I think it would be awesome like to, to hear from you, like how you experience this. Yeah. I mean, so I think Mode has been the, there's a question of how these have changed, a question of sort of like, why is Mode still standing? Why are some of these other companies not around or been acquired? And why is Mode still standing in that regard?

I think a lot of that is luck, to be honest. I think that we had the initial thought for what Mode was, was something that was born out of our experience. And I think that saved us a lot. Where in these sort of frothy versions of the market, you hear a lot of different things from customers

and everything is moving really quickly. And I think we always had some grounding of the product that we wanted to use. And I think that those guesses that we made about what makes that product important, about what it is that analysts will actually want to use, what is it they won't want to use? What are the things they value versus

the things they don't value? We got lucky in that I think they were mostly right. That doesn't mean we like did everything perfect by any means. There are plenty of places where like, you know, we didn't execute on things we should have executed on or made bad decisions or built the wrong thing or whatever. But generally I think like in broad strokes if if we obviously didn't know exactly where the world was headed but if you we had looked at the world in 10 years ago and said this is what it looked like in 2022 we would have been like that's not a bad one for us obviously if we'd known that we would have done some things differently but it's a world that generally fit the direction that we were kind of i don't want to even say betting, like we were,

that would give us more sort of like agency in this that I think was really true. Like it was a world that we were intentionally or not building towards. And so I think we got lucky in that where things like SQL stayed popular. One of our big bets was like, yeah, SQL-based BI tool, basically.

Sure, lean on SQL. That could have not happened. And at the time, that was a little bit unconventional. At the time, it was like, no, look, ML is the right way to do stuff. No, people are all going to move to Hadoop-based stuff. And we were like, no, we're pretty sure SQL will be good. And that was one where we did

actually make a bet, and it landed pretty well. But things like, like, Node is a BI tool without much of a governance layer. We didn't focus a lot on semantic modeling and things like that. We focused on like rapid iterative analysis. Things like DBT

introduced a semantic layer that we could use. And so that made it such that the version of BI that we had built, which didn't depend as heavily on this sort of built-in semantic layer, actually kind of fit. Had that not happened, had there been nothing like that in the market,

had there been just like, oh, actually the way to do this is you build stuff like LookML or you build really deep versions of that. Like, yeah, Node would not have, have not fared well in that world. Now again, one of the reasons we did do that is because within, within Yammer where we built these tools, we'd actually built an internal thing that sort of looked like dbt. And so we were building kind of to the world that we created we got lucky in the sense that that bet partly like you know we built a thing that that just had some some it works but we got lucky in that bet that that was actually the direction that things panned out so i think that like part of it was you know that's the big reason to me why mode has been around was like

in a macro sense a lot of these trends were things that played out. No, you know, I would also be remiss to not say another big reason why it's paying out obviously is because there's been a ton of hard work by a lot of people, there were a lot of very smart people on the mode team that, that made good decisions that just put in the hours to make it work, you know, startups are both a combination of luck and sort of the grind and, and I am very fortunate that, that to have been part of a team that was able and willing to put in that grind.

So I'd be sort of remiss not to mention that. But, you know, I think in terms of how it's changed, like, mode is actually kind of interesting and it has never been through a pivot. Like, a lot of data companies

go through sort of pivots in their life. Like, mode is never really pivoted. It's sort of like winded a little bit there have been some some big swinging turns in some ways where it's like oh let's focus a bit more on this focus a bit more on this but if if you just showed me again mode the moment we founded it compared to like if if the day of founding me had seen mode as it exists today i I would be like, okay, yeah, fits.

Like it would have still fit the general direction. So, um, you know, for us, like not that different, I guess, but, but again, a lot of that was just sort of like lucky macro guesses largely in the direction that the market was going to go. Henry Suryawirawan, Yeah. You actually like pretty much also answered like the next question that I had, which is like, based on your experience, like, do you think that there was, let's say, more change that was required for the product or for the business of mold because of the changes out there?

Like what, where was like more pressure? And from what I understand, like, yeah, okay. Like market changes, maybe we have also like to adapt on how we do business and how we go to market, but you didn't have like really to go through like some kind of big change on like what the product is, right? Like... No, we haven't gone through that. And I think that there have been changes on the market and it's more of like market position. It's more of like, if you just like were to map out the landscape, mode as a product is still similar to what it would have been, you know, day one, really.

Like, again, obviously it has a lot more capabilities and things like that. But like, it's not, if you squint enough, it looks the same. However, the particular place that it fits has become very different. So 2013 was a time when people were skeptical of data in the cloud. There was no sort of like, ELT was barely a thing. Most people did not have cloud data warehouses that they were starting to. There was nothing like DVT or transformation. There was certainly nothing like orchestration, certainly nothing like data catalogs and

observability and all these other things. There was no real concept of there was a very strong concept of exactly what BI was. There was no real concept yet of analytics teams and what they were supposed to do. Our first customers were data science teams. What are they trying to do? It's a little bit different. We had to explain to a lot of people that there's this thing, like I remember the very first pitches we ever gave.

It was like, oh, there's this thing that's outside of a BI team that are the people like writing SQL queries, answering questions. And like one of the things that we got a lot of questions from those pitches was, how many people write SQL really?

Like, does SQL really matter? And so that obviously, like that positioning has changed. There's like data teams, analytics teams, what analytics engineering represents was not a thing at all. So like all of those worlds have changed in a way that, that again,

actually grows around us in a decent way. We've gotten sort of lucky in that, but, but I think that's, that's largely the thing that's changed. It's like, how do, how do we position exactly what load is in the market? Positioning is both about what we have and what the market has in itself.

And so I think that has certainly changed. But again, the specific of what it is, isn't actually up there. Okay. And how much does the market change since 2013? I've been kind of interested in your view on that too, as someone who was in it then as well. My view is a ton. I mean, I think that there's a handful of things

and I think they're very different. One is the cloud, probably the biggest one. That has become much more ubiquitous, much more widely accepted. Certainly there are some companies that are still hesitant about it. You can't sell AWS to Walmart type of stuff is still there. But like, hey, we have a cloud tool used to be like a deal breaker for a lot of people. Now it's kind of expected. And I think that's up and down the stack. The other, I think, really big change is, broadly speaking, the idea of analytics engineering.

That's a little bit of a nebulous term. I don't quite mean it like exactly as the people who do what you would find on like DBT's blog is what analytics engineering is. To me, analytics engineering is more representative of like the new structure of what data teams look like and this gets to to eric your earlier question about like data teams and how they're different when when i was a yammer we had data engineers which was just responsible for like building everything building etls keeping up a warehouse basically being dbas all that sort of stuff and then analysts were just strictly writing queries on top of it and built nothing.

It was an easy split because they were like the builders and the analysts and that is very clean. In this world, it's like a lot fuzzier. But as a result of that, you get a lot more of like people creating stuff with fast tools and there's broader acceptance that a data team isn't just a bunch of like business

analysts. It's this broader team that's trying to like solve big, hard problems. Changes a lot of the landscape in terms of how people like build tools, what they're trying to solve with them. So mode, which is a tool that sort of fits those types of people, fits a little better in that landscape

than one where it's like, well, we have data engineers and we have a bunch of BI analysts. One question for you on that, actually. Did you, so when you had the builders and the analysts, like, and you had sort of a, you know, a really clear delineation there of jurisdiction, say, were there context problems, right? Because the people building, you know, are like building to spec, right? Say like, okay, we need to get data from here over to here. But the analysts like have a lot of context from the business users. And I agree

that today, like, I think there's a really healthy crossover between like the building and the analysis. And like a lot of the tooling allows you to like combine the context from the data to the business problem. I mean, Moat's a great example of that. But back then, do you feel like there was a challenge with the builder getting enough context to create a data set that serves the analyst's needs? And then ultimately the business user users needs or the consumer. So I'm sure that like we, we didn't have that much problem with that at Yammer because of the way the team was structured and because of the tooling that we had.

So the way our team was structured was data engineering and the analytics team all reported up to the same person. Like it was all part of one, one org. And so we were, they were, we were their customer. They were our, we were their customers. I don't know, whichever one we were sort of like the people they sat next to. And the data engineers were basically building to serve us as analysts. And so it was like a pretty tight relationship there. The other thing is in, in like the analytics engineering world, we had built a tool that

was a SQL based transformation tool. We had built a tool that was a sql-based transformation tool we had built a tool that again like looked similar to what dbt does today so that the pipelines were like they're just a data transformation the pipelines were built by ants like we ourselves were building that i think that solved a lot of that problem where the data engineering folks were much more focused on how do we give you the tools to do that how do we give you query tools dashboarding tools? How do we give you query tools, dashboarding tools, and those kinds of things? Whereas whatever went in those things were our responsibility.

But I mean, yeah, this was a problem back then. There was, I don't know when this blog post is from, I think it's like 2015 or 2016. A blog post on Stitch Fix's blog, which I think is kind of one of the big trajectory changing things that has happened over the last decade or so. There was, I think the title of it, something like analysts shouldn't, or data science, or no, data engineers should not write ETL.

Where it was essentially arguing for data engineers being out of the building pipelines and transformations because, to exactly what you're saying, like you can't, it's a hard thing to spec out. And this is very bad back and forth of a very capable data engineer writing a bunch of SQL to transform data in ways they don't really understand, and the analyst writing the spec, it just doesn't make sense.

And so the reason that blog post probably had to get written was because the problem you were describing was exactly what was happening. We were a little bit ahead of that at Yammer and the folks who came before me, who built all this stuff, anticipated that and built some solutions for it. But I'm sure industry-wide, that was not the case. Yeah, 100%.

That's also my experience. By starting a company back then, I also had, let's say, the luxury, let's say, of the experience of being one of, let's say, the first categories of this bundling, right? Because, okay, it's one thing like to go out there and be like, I'm building a BI solution. Okay, it's going to be like a very different way of doing BI, but it's a BI solution. And it's another thing like to do what like companies like Fivetran were doing back then, or like Stitch Data and including like Blendo, where we were going out there and we're saying, well, you know, like you can move your data around by just clicking like a few buttons and you can do

it on the cloud and you don't really have to write your own transformations to move the data there and all these things. And one of the biggest changes that I have seen is in the behavior of the markets towards this kind of tool. If you're not like Ben, we started the conversation, we were talking about how many different tools we have today, right? Like back then you were saying that outside of like BI and the warehouse, I'm also giving you like a platform that's going to do like this one single thing, which is like moving the data and like doing the extraction and the loading of the data and even that was like, why do I need that? Why like Looker is not going to do that?

Why is not the AWS with Redshift is not going to build that? Or why I'm not going to just ask my data engineer to go and maintain and build the Python script to do that, right? Today, we don't talk that much about that. It's pretty much like a non-existent conversation. And a big change that has happened, definitely in terms of the perception that people have on how this infrastructure should be built and maintained and i think it has a lot to do with like the empowerment and like the different let's say position that like data teams have in the companies right now right it's like

it's not like an it function it's like something else and that i think like has made like a huge difference, both like from the user perspective, because of like all the different tools that they are like available out there to use, but also like from being someone who wants to build it, right? Like it's a completely different experience to go to market right now.

And I would say that for me, it's probably like a little bit of more of a technical like catalyst that I think that has happened. Is the cloud warehouse changed things a lot. And it's not that much because people moved into the cloud. It's because of like the elasticity that it provided for computing that made the ELT like reality because it's a completely different game to chat about, to discuss about data infrastructure

like 20 years ago when you needed appliances. You needed, let's say, writing ETL meant that any logic is part of the pipeline and like, do we want like to remodel the table? Like, yeah, we have like to go through like huge deployments to do that, right? None of that stuff would have happened if we didn't have, let's say, a way for people to be free from like how many resources they are going to need and like budget, like the computation and like the infrastructure upfront. So these are like the things that I have seen like change the lot and like really like acted like as catalysts to get today to where we are.

And so together with the money from the VCs, like we have ended up like having all these like wealth of innovation happening. So that's how I see it. So the one thing I would add is, is it also compounds in, in ways that I think are, are hard to anticipate. You know, that there's like the, when Uber was early, there was a lot of stuff of like, oh, well, it's obviously the cap of it's the taxi industry, right? Like we will use Uber for it's the Uber and taxi industry that are zero something.

And basically what Uber was like, I guess proved to some extent was if you make this easy enough, people will do a lot of things they wouldn't typically do with taxis where they'll commute to work on Uber because it's just straightforward and you can schedule it now or whatever.

I'm like, I would never do that in a taxi, but with Uber, why not? And so I think that we're like, you know, if I live in a neighborhood that doesn't always have taxis, I may not take it to the grocery store. I think a lot of stuff happens around like data infrastructure like that too, where once things like the cloud storage became so cheap and it became fast enough, we started building a lot of things on top of it that we wouldn't have really wanted otherwise. Or once Fivetran, the stitches and blenders of the world became a thing. It was like, oh, actually, yeah, I'd like to get data out of my applicant tracking system. I don't think I need to do analysis on that.

But if it's straightforward, yeah, I'll do that. We sort of catalyzed a bunch of demand because we made stuff that was so much harder easier in a way that the ecosystem really kind of bloomed from. Yeah, 100%. Eric, my turn to say that I monopolized the conversation. So it's yours. No, it's great. Ben, okay.

So I've read a lot of your writing over the years and really appreciate it. So thank you for the time and thought. And by the way, if our list, I'm sure that a lot of our listeners read your stuff, but if they don't, where can they find your musingsings online so most of them are a sub stack it's been.substack.com need to work on the branding i like it you know it doesn't have a punchy name or whatever i never got past that point yeah that's like i if you want to follow me on twitter great that's much less interesting and infrequent i don't know i'm you know i keep myself sane by not saying too much on twitter yeah yeah the the longer form thing if you have trouble sleeping or looking something to

help knock you out at night and you can check it out at the end that's no it is really good stuff but you know right sort of writing like is a form of in some ways at least in my opinion like sort of really testing your thinking you know because you're actually putting something down into words and one thing i'm interested to know like you think a lot about the data space in general. Trends like you've seen it over a decade, like. You know, I'm tempted to ask you the standard question of like, what's interesting or whatever, but what I what I really want to know is. What is a trend that you've seen that you that you look at and you're like, this is happening, but it's probably not a good thing? You know, in the long term. Probably a couple of things.

So first of all, I'll say that, yes, people say that writing is meant to help clarify your thoughts and test things out. I do most of my testing in production on that. So you read the blog. Yeah.

Yeah. Apologies in advance. Force push to master. Yeah, exactly. Hope for the best. See what people say. Sometimes it works, sometimes it doesn't. So, okay.

So I think, I think there's a couple of things. These aren't really trends. In terms of like big trends, I think there's like an of things. These aren't really trends. In terms of big trends, I think there's an oversaturation in the marketplace. There's a lot of small problems that we're trying to turn into big problems that just don't need to be solved in that way. Yeah. Or they need to be solved. They are better office features and not products.

That doesn't seem super interesting. I think most people kind of agree with that. There is usefulness in having companies trying to solve those things that eventually we'll figure out good ways to solve them and incorporate that solution into something bigger. Yeah. Which is like as an economic reality, I think it's hard for that stuff to become viable. One of the things though, that I think is interesting about that, there's another trend in the data world that I think is kind of a bad dynamic. That mode is, like, every data company is and I think it makes for ultimately worse products,

which is everybody tries to be Switzerland. There is this, like, general sense, and I think DBT did this really well, and I think a lot of people basically are trying to do the thing that they did, where you are competitive with nobody, and you're like, oh yes, we are agnostic to every other part of the stack. We work great with every database. We work great with every ETL tool. We work great with everything else.

There is this kind of like reluctance to compete unless you are so squarely in someone's space that you can't do it. Like Stencils and Hightouch fight each other. It's kind of fun. I like it. I like that they like yell at each other about performance nonsense. I'm not here for it. As of the side, I'm also a small personal investor in census, but don't care. I'm just here to watch these people fight it out. I think that's good. I think it's actually good that Snowflake and Databricks yell at each other about performance and stuff like that and are not shy about the fact that they're trying to do the exact opposite. By expensive billboards, right? Yeah, which snowflakes billboards are terrible, but

clearly some 60-year-old dad wrote all their billboards and it's just terrible. I love it. That's one of the best spicy takes I've had on the show in a while. Which I guess you know, here we are talking about them, so job well done. But I think there is this general thing of like, we all have to play super nice with each other.

And I don't know that that actually leads to the best experience for customers where you sacrifice a lot of, well, if we just worked really well as other people, if we focused on making really great experience with these two or three other folks, we could actually solve really good problems in really, really robust ways.

Because like there is this kind of fragmentation problem where the experience is kind of all disjointed. I think that's a problem. I think the way you get around it is not by trying to make everything work with everything. And I do think there will be some, that is maybe a positive change around what is happening in the market now is there'll be winners that start to emerge. There will be clearer like, all right, I'm going to align myself with the winners and say, actually, I don't want to work with the people who are struggling. Again, that's in some ways like it's a cutthroat dynamic. It's a dynamic that will lead to some people losing. Yeah, maybe mode.

I don't know. But the point is, I think like in terms of as the customer of those things, I would rather that dynamic and think we end up with long term just a better product than if we end up with something where everybody's kind of like trying to play a little bit nice and it's not the thing is it's not even nice it's like trying to play like passive aggressively nice where yeah the positioning is like oh yes we're all friends like i think there's some just like now let's actually just bite it out see if we make the best products actually is going to be useful to me. Yeah. Love it. Yeah. It's interesting because that's like a very like strange flavor of hypocrisy in some ways, because, you know, your go-to-market is like, we're all friends and your investors are like,

we're not all friends. Yeah. Like if you get the outcome that I want, like we're not all friends. Ben, this has been such a fun conversation. It felt like we were talking for like five minutes, but it was an hour. So thank you so much for your time. And we'd love to have you back sometime to come back and give us some spicy takes.

Yeah, thanks for having me. It's been fun. And I think my biggest takeaway from that conversation, Kostas, was just how transparent and I think real Ben is. I mean, you kind of get this from reading his blog, but it's just so refreshing to hear someone say,

man, VCs are pouring a bunch of money into companies that, you know, are basically should be features of other products. Right. And just to hear him talk about that and even sort of, be dispassionate about mode, right? Like he was very clear,

like maybe mode gets it wrong you know and sort of like goes to the grinder you know they've survived 10 years so i don't think that's likely but like i don't know i just appreciated that like he was just he was probably one of the most transparent guests who like will just say say it how it is and like is okay with that also incriminating like himself and his company. Yeah. Yeah. Yeah.

I think that Ben is like probably one of the most, let's say, clear and consistent voices out there in the industry. And that's something that like, I think everyone appreciates. It's not just me and you. And that's one of the reasons that I'm so happy that we had him today, like on the show. What I keep from our conversation, to be honest, was like the last part of it's about competition and how competition is important to build at the end, like a better product for the customers out there and how this particular industry right now, as it is like, it's at least at the surface level,

like tries to avoid competing. But I have a feeling that like reality is going to accelerate things. So we'll see more competition happening out there and I'm looking forward to it. Indeed. All right. Well, thank you for joining us.

Tell a friend about the show if you haven't and we will catch you on the next one. We hope you enjoyed this episode of the Datastack 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 Rutterstack, the CDP for developers. Learn how to build a CDP on your data warehouse at rutterstack.com.