What even is the modern data stack

I am joined today by Benn Stancil, whose Substack I’ve been following for quite some time. I know it’s been a while now, because I had to go look up some of my favorite posts and I had to scroll quite a ways. Benn, you are a writer in the data world. I’m not sure how you refer to it, talking about the modern data stack a lot… Things that I barely understand, but I really enjoyed your writing, and I thought our audience might enjoy some of your topics as well, so thanks for joining us on the Changelog.

Yeah, for sure. Thanks for having me.

So if I got your story right, you were in the data world for a while, you were a hesitant writer, and then you started this company called Mode. And in the early days of Mode you started writing more for the Mode blog as a co-founder. And I don’t know if that’s where you’ve found your niche, or somehow you got inspired to write your own thing later, after Mode was established. I know you’ve sold that to somebody else at this point, but tell me about your journey into what is now a weekly authorship… Is that what I’d call it? Maybe it’s just a blog, I don’t know. It’s a Substack… You’re writing weekly, long essays, really, and you’re pretty consistent with it. Can you tell the story?

Sure. So like you mentioned, I started a data company; it was a BI tool, basically, that was called Mode, about 10 years ago. When we first started it, there were three of us. One of the persons was the CEO, who was a good face to the company, and out talking to investors and customers and things like that, and was the sort of person that we could probably stroll out as the external face of what we were building.

There was a person who was the technical co-founder, really, who was chained to his desk, building the product, and then there was me, who was neither of those things, who neither was an engineer, nor was sort of fit for external consumption, I would say… And so I didn’t have anything to do. Why I was a founder there - who knows? That’s a question you have to ask them. But my job – basically, my background was in data, and as an analyst and things like that, and really what I was doing was kind of representing the customer in a lot of ways, where the product we were building was for people who were like me… And so I was to some degree PM-ing things, or helping the person who was the engineer do the “Here’s what we think we should build”, and testing stuff out. But that leaves you with a lot of time, in the early days when you’re basically moving as fast as one or two engineers can build it… And so I had a lot of time to spend on stuff, and so I started basically writing a blog, that - we didn’t really have a grand plan behind it, but what it ended up being was kind of like FiveThirtyEight-style analyses of pop culture.

[08:15] We wanted to do something that would get notes and visibility… I kind of just started doing this because I needed something to do, and it was kind of entertaining to me. And so the very first blog posts we ever wrote were things about like Miley Cyrus, and the VMAs, the Video Music Awards, and various things that were just like interesting things to me going on in the world, from a data-driven perspective. So it was sort of like “Here, let’s take a data-driven look at x thing”, or whatever.

People seemed to like it. I think it worked because this was a time when content marketing and sort of having company blogs was becoming pretty normal… But most of those blogs would be sort of transparently thought leadership with the intent of somebody clicking on the button saying “Download our white paper” or “Check out our product”, or whatever. It’d be like “Here’s five tips to build your engineering team”, and then the fifth tip would be like “Use our product”, or whatever. And so this was not that. There was no real call to action for anything related to Mode, it was just “Here’s a bunch of charts about the baseball playoffs”, or whatever.

And so people seemed to like it, I enjoyed it, I thought it was kind of fun to do, and the writing part of it was something I’d never really done… But it was like “This is kind of interesting.” Eventually, within, I don’t know, six to nine months of starting doing that, Mode grew, my job expanded, I started doing a lot of [unintelligible 00:09:25.09] started having customers… I basically didn’t do it for that long, because there became a point where writing a blog about Miley Cyrus is not the most important thing that you can be doing to grow startup.

Eventually, yeah.

Yeah. No, honestly, I think actually probably it should have been the thing that I kept doing. It probably was more important than me answering customer support tickets, and somebody else could have answered customer support tickets… But at the time we certainly didn’t think that. And so for a long time, this was always a little bit in the back of our minds, of like we thought that was a fairly successful thing, and at some point I’ll get back to writing that sort of stuff. And honestly, it took – we thought it would take two years, and it took eight. But there was a period much further down the road with Mode where – my job at Mode basically bounced around to a bunch of different things, I was kind of like… The internal joke was my title should have been Chief Interim Officer, where I basically like filled the roles of whatever executive we didn’t have… So it’s like “Oh, we currently don’t have a head of marketing. Go do that for a bit.” Or “We currently don’t have a head of support, or product, or whatever. Go do those sorts of things.”

At some point we had hired somebody to do all those things, and we sort of built out the executive team, and a lot of like the sort of good and much better leadership than I was in place… And so I was like “Alright, I now have a little bit more time. I’m gonna go back to doing this blog stuff.” The original intention was to do what it was back in the early days, of “Alright, I’ll write these kind of like data-driven-y things.” For whatever reason, that’s not exactly what I started doing. I had all these rants essentially about the industry that we had been working in… And so I wrote a handful of those posts, people liked some of them, and at some point you get sort of captured by your audience a little bit, I would say, where you recognize that these are things people like, that you have stuff to say or stuff that you kind of entertain yourself with saying, and then at some point it kind of takes on a life of its own.

So there is no sort of grand plan behind it, it never really became a product marketing thing, or an actual marketing funnel for Mode… It just became a thing that people kind of paid attention to, and so I kept doing it, and that’s as far as the plan goes.

[laughs] Is there an end game? Do you have an end game to this? Is it just like “Serve your audience forever, until…”

No, there’s definitely not an endgame. That doesn’t mean it’s – like, do I imagine I will do it forever? No, probably not. I don’t know what will be next, but no, there is no “Okay, it’s time to initiate step two” kind of thing. Step two is it’s Friday, and I guess I’ll publish that then.

[laughs] Well, I’m often impressed by how reliably you publish on Fridays, but also how deep you tend to go into your thoughts and your rants. I would characterize your writing as somewhat irreverent, definitely pop culture-y… Sometimes meandering, and I say that in a positive sense, although some people don’t like meandering… But then always, always coming back to the point. I’ve enjoyed it even being a bit askew from your world. I think our worlds overlap, but aren’t one to one, and so oftentimes I find myself just kind of like with a view into the world of BI, and data, and whatever they’re calling it these days; the data cloud, I’ve heard recently… You seem to refer to “the modern data stack” a few times. Can you define that? Is that a real thing, or is it just like everybody who’s in the industry knows what it means, and nobody else does?

[12:36] The answer’s sort of yes to all of those things. It does not have a quite clear definition. The most accepted definition of it, I would say, is a collection of tools that are all somewhat oriented around the same philosophy of how companies should build their data infrastructure. So things like cloud-first, tends to be things like modular, where they’re not these one giant behemoth type of suites, but are kind of like specific point solutions for solving data warehousing, or moving data between certain tools, or data visualization, or BI for a particular subset of types of consumers, versus sort of technical tools for data analysts and data scientists… Like, there’s a lot of products for each one of these kind of narrow verticals, and there was a period of time where data was one of the hotter VC spaces. It was a place that you could raise a lot of money, and so there was a lot of draw of getting people to found companies in that space, and so a lot of people who were - myself included… Mode sort of predated this fad, but probably in the 2017 to 2021 days if you were a data practitioner or an analyst, or someone who had worked on an internal data tool, let’s say a Facebook or an Airbnb or whatever, it was pretty straightforward to go out and say “We’re going to turn this thing into a product, and we’re gonna raise money”, and you’d get pretty good valuations, and all that sort of stuff.

So there was a collective sense of we’re all somewhat in the same cohort. The modern data stack roughly refers to that. My view of it is – this sort of like jokey, but actually sort of serious definition of it to me is it’s data tools that were launched on Product Hunt…

[laughs] There’s the irreverence right there.

Well, I mean…

Yeah, I get it.

Yeah. It’s like, Product Hunt was sort of a marker in time. Tthere was a point at which that was a big thing, and now I guess it’s still there, but people don’t really focus as much on it. It also is like – a particular type of tool gets launched on Product Hunt. Oracle doesn’t launch their stuff on Product Hunt. And the modern data stack typically is not referring to Oracle’s latest releases, it’s referring to very like Silicon Valley-oriented, bottoms-up things that have some ambition to build like great user experiences, and product-led growth, and all that kind of stuff. And all of those ideas are sort of jumbled together… But the modern dataset is basically some collection of that set of things.

Right. So another thing that I guess is in my world, but not exactly as a software developer, which I find a lot of parallels, is the cloud and the Kubernetes world. So I don’t know if you’re familiar with Kubernetes and that whole ecosystem of cloud-native things, but when you go to KubeCon and you attend that event with 3000-5000 people - and that’s all open source projects, and open source companies, and all these things… And you’re like “Wow, the money is here. This is where the money is in the open source world. This is where the commercial open source companies are.” Everybody else who’s not in like the Kubernetes land or the cloud-native land, we can kind of get some money with our open source companies, or maybe not, and we struggle, but the money is there. And I felt like for a very long time, in the world of data, and led by Snowflake at least publicly, there’s all these companies that either IPOed, or are like well funded, creating data tools for data analysts… And I felt like the money is also there. That’s the parallel I’m drawing. For a long time, the money was there.

[16:02] One of your recent posts - I guess it was this year, even; yeah, January. “It’s time to build”, in which you’re referring to a change, a shift in your world… And I think it was the shift of Open AI, Chat GPT, large language models, and really the hype that has moved into, and stayed at least for now in the AI world. Did that suck a lot of the money and the air out of the modern data stack?

Yeah, yeah, a lot. And so the Kubernetes parallel, which I don’t know – I mean, I’m familiar with Kubernetes; I certainly don’t know the vibes of Kubernetes conferences, but… There definitely was a period of time, to some extent pre-pandemic, and very much so in the 18 months sort of post-pandemic bottoming out in mid 2020, to market turning in whatever that was, early 2023. For that one and a half, two-year period there definitely was the same kind of like “This is just a crazy amount of money in the data space.” Lots of people starting companies, raising in obscene valuations, numbers that are infinite multiples, essentially, of companies raising 9 and 10 figure valuations on a handful of millions in revenue, or even less in some cases…

And so yeah, there’s a particular line I remember from a conference, there was a conference – I want to say it was the first conference post pandemic that was in-person. That was like the first one that really was like “Alright, let’s go out and do the stuff.” I think it was early 2022, where it was a pretty big conference, there was probably 1,500-2,000 people, so it was not huge, but it was not some tiny meetup… And somebody asked this question of “How many hundred-million-dollar businesses are there to be built in this space?” There was a panel of like VCs, or whatever. And the answer the VC gave was basically “Infinite.” It was like “Everybody in this room could build a $100 million company in the data space. The data space is that big.” And that was sort of the attitude, was like “This stuff is all huge. It’s all going to be like these enormous companies”, Snowflake just had this IPO, and was like the biggest tech IPO in a long time, or ever, depending on how you kind of counted…

Then two things really happened that sucked the air out of the room, kind of at the same time. One was the market turned, and so there was just a lot of “Oh, wait, maybe this was a big bubble. Maybe the company that is making $50,000 a year isn’t worth half a billion dollars.” Like, there was some just like “Oh, wake up from that fever dream.” There was also AI. Suddenly, all of the VC interest turned from “Oh, data stuff is going to be the future” to “AI stuff is the future.” And so people got kind of hit twice with “The market’s gonna put a lot of pressure on things”, but also just it wasn’t cool anymore. It wasn’t the place where – VCs aren’t hosting dinners for data founders, they’re hosting dinners for AI founders. Everybody talked about why they’re an AI company, and not a data company. And so it became much more of just like “Okay, this is just one of the pieces of the tech industry”, the same way that CRMs are, or marketing tools, or backend finance infrastructure stuff.

Right.

Data is just one of those stats. And so fine, that’s probably good, it’s probably good for us building better things… As me, as someone who yells on the internet about this stuff, it’s fun to yell about, because it’s not – I don’t know, part of the appeal of it is there’s energy in it. And so that post ultimately was kind of like – okay, is it fun to have a blog about CRM software? No, it probably isn’t.

[laughs] Well, in that post you confessed how you had changed even the content of what you’ve been writing about, because - well, you were gonna follow what’s interesting. And of course, there’s parallels, or I guess there’s touches between AI and between the data world. You wrote this in that post, which I thought was – for me, at least, it was the nugget that I thought “Okay, this is insightful.” Not that the whole thing wasn’t insightful, Benn, but this part I thought I would read back to you and have you expand on.

[20:00] You said “Though it can be demoralizing for the air to leave the room, there’s a lot of opportunity in the slowdown. Startups just need to change their tactics. Don’t build something new or go after major incumbents. The wilderness is too hard to tame, and the cities are too hard to conquer with a lot of money. The better targets are the helter-skelter frontier towns, built by frenzied founders, who wanted to stake their claim on any piece of open ground that they could find.” I’ve found that to be insightful, and I’m curious, what are the helter-skelter frontier towns? I like the wording of that, but I’m not really sure… Are there concrete examples? This is like where people have kind of stated claim, and then they’ve failed and moved on… Is that the idea?

So taking the modern data stack, for instance. The primary thing that the modern data stack sort of did was say “Let’s move a bunch of data tooling to the cloud.” That’s probably the biggest thing that it was – its biggest sort of philosophical bent, was cloud-based SaaS software versus some sort of like heavy infrastructure that you buy from Oracle or Microsoft or whatever. And so there’s a lot of things that people did to make that easier. Data pipelining can be now moving from SaaS – rather than these sort of big, heavy data pipelines that you have to write yourself, to sort of push-button stuff from SaaS products like Salesforce, into data warehouses, and easier ways to be able to share data back and forth between other tools, or share data between… Like, how does the data team share their results with the marketing team easily, and stuff like that… All those sorts of things.

So there’s a bunch of stuff here that like – the way that data teams work kind of changed. They also became – there’s this sort of new team… Like, prior to 2010 data teams were kind of either capital D Data Science teams that were doing hard math, or they were kind of like business intelligence reporting teams that would be building just like binders of reports for executives… There also kind of was this rise of like the analytics team that was supposed to help people make better decisions by doing a bunch of analysis that wasn’t “We’re gonna build some crazy model”, but was “We want to help you decide how to be smarter.” The easy analogy for this is like sports teams. There’s analytics teams in sports franchises now, that are data scientisty types, but they’re not saying “We’re gonna build some crazy predictive model that we can plug into anything.” They’re saying “Oh, we want to help our coaches make better draft picks.” Businesses try to do the same thing.

There were a bunch of tools built for all of those things, for those people, for those different workflows, things like that, that were kind of new and novel approaches to it. But during kind of the frenzy years, those things, one, were all new, and so they were still figuring out how to make it work, they were still figuring out “What’s the best way to do this and move it to the cloud? What’s the best experience for that?”, all that kind of stuff. And because there’s so much money in it, people are trying to move really quick, they’re trying to like have some – like “We’ve got to build a giant business.” And as a result, you build kind of shoddy products. They’re not bad, but they’re frontier; you’re figuring out what to do. You don’t really know what works. The ground is changing underneath you some…

So for instance, Mode got built – started originally in 2013. There’s this really popular open source tool called DBT, that’s like a data transformation tool that helps you define how to basically build data pipelines. That thing got popular and became pretty ubiquitous 2017-2018. Now what Mode does would be better off served if it also kind of understood the way that DBT works, because a lot of customers are using DBT. However, Mode was built before that. And so if we were to rebuild Mode today, you’d kind of rebuild it knowing that “Okay, there’s this new set of technologies that most people are using.” There’s a lot of examples of that sort of thing, where products got built when the ground was still evolving, and everything was still changing… And so they’re not quite built for the world that exists today.

And so I think there’s there is an opportunity now to basically say, okay, the ground shifting has settled. The earthquake is sort of over. There’s a bunch of half-built buildings that were built on the ground before it shifted entirely to where it is, and they’re a little bit shaky, and all that kind of stuff… And actually, you can now go back and basically say “We’re just going to build with those things built.” The ideas were good, but they were early. There were products that we’re still figuring how to do it; they were products that were built for a slightly different time. Let’s just say we take the really good ideas and rebuild it for the landscape as it exists today. And you can’t do that when everything’s crazy and there’s so much money everywhere. But when things are calmer and more settled, there’s a lot of opportunity, I think, to do that.

[24:16] Yeah, kind of a second wave of products, having learned the lessons of the first wave. So are most of them like Mode, where they were acquired and now owned by something else, and maybe like brought in? Or some of them have small customer bases and they’re chugging along, or are there ones that are actually dead on the side of the road? Is it a whole mixture of all those things?

It’s kind of all of it. I think you could have examples of all of it. There are some that got acquired for huge numbers, there are some that got acquired for less huge numbers… There’s some that got acquired in fire sales. There are some that are dead along the side of the road, there are some that are walking dead, that are going to be dead in three years, but managed to raise enough money in the good times to keep chugging… There’s are some that’ll probably make it through, that are semi walking dead, but will figure out some way to make it work… There’s probably a lot that are just going to not ever quite figure it out, and not be able to grow into the valuations that they had… So in some ways I think this is just like the Silicon Valley circle of life. The same is gonna happen with AI, I’m sure. There’s thousands of AI companies that, one, were built for kind of the same thing, built for a world in which AI models were GPT 3, that was the best they were, or they were built for us assuming we’re all going to use RAG. I don’t know, maybe we do, maybe we don’t, but it’s certainly possible that RAG is a fad, that in two years actually there’s something way better, that makes way more sense, and everything that was built for RAG will have to sort of figure out what to do. That’s basically what happened with the data world, except instead of RAG it was a handful of other kinds of paradigms that we have evolved our way through. So there’ll be a lot of companies there that figure out places to land well, there’ll be some that figure out how to survive and thrive in the world, and there’ll be some that’ll die.

If you are interested in executing on this advice, a) is there still enough money floating around to where you could raise if you had a solid plan? Or would you have to bootstrap? I guess I’ll stop at a and let you answer that before I ask b.

Yeah, I think there is. I mean, I think there’s still a lot of money in venture. You might have to tack some nonsense AI pitch on there to really get people excited…

That’s unfortunate, but yeah. It makes sense. [laughs]

But I think that if you have – people are generally aware, I would say, that the data industry is full of a lot of tools that were propped up by the good times of 2021-2022, and probably are vulnerable in a number of ways. They’re like businesses that are not really designed for a slower market, they’re businesses that were sort of spending as though there was no tomorrow, and stuff like that… They’ve made the adjustments and sort of – they’re not still doing that necessarily. But it’s hard to sort of restructure a business to that degree.

I think there’s a lot of understanding that these products, some of them are like good ideas, but there’s space to reimplement them, or like just build really good versions of it… And I think there’s a lot of investors that are chasing the Notions and Linears and Figmas of the world that are kind of this polished craftsmanship version of an existing tool. I think you can pitch that. You can be like “Look, we are the same as these other things, but we’re just gonna do a really good job of it. We have this great team. Let us show you the quality of our craft.” People will always – I think you can sell that pitch.

So no, I don’t think you have to bootstrap. I don’t think you can raise at the sort of astronomical valuations by any means, but yeah, I think if you come along with something that’s like “Yeah, this looks exactly the same as the other thing, but our goal is we’re just gonna make it really good, and we know how to do that, and we’re great at our craft.” I think there’s always sort of money to be had in that pitch.

[27:56] Okay, so part b would then be imagine that you’re not you, you’re me - and hypothetical me, because I’m not gonna go do this. But a software developer, who doesn’t understand the data world very well. How would you identify a target to actually go and do something like this? Now, you have to take away your own knowledge, which is really hard to do… Like, what would I do? How would I start? How would I figure it out? Is it possible, or do you have to be in the world already to know?

I think that’s really tough, partly – so there are a lot of products in the data world that are traps, that seem like things that aren’t too bad to build, and can be solved, and “Why doesn’t somebody just do this?” And my God, they are messes. So Mode is a BI tool, basically. And BI is like dashboards, right? It is an endlessly tempting thing to build, because it’s just charts. Charts seem pretty simple. There’s some cool new open source library that makes that easy. What if instead of asking questions and stuff, we now have AI, and we can do like natural language? There’s all these things about it that seem pretty straightforward, and it’s just… It is a product with 1,000 edges, that everybody wants a slightly different thing, that every customer is gonna have slightly different preferences about the way they want stuff to work. Charting, for instance, visualization - it is the biggest cookie you could ever give a mouse, where people will want an infinite list of customizations about visualizations…

So I think there’s a lot of things that look easy on day one, that once you start doing them – and like we suffered. We thought like “How hard can this be?” And it’s 10 years later, we’re like “That’s pretty hard.” There are versions of that famous Hacker News comment, when someone [unintelligible 00:29:33.10] “Well, isn’t this just like an SFTP thing that I could run myself?”

Right. That I can build in a weekend.

Yeah. There’s a lot of data products that you feel like you can build in a weekend. And you kind of can. You can build like the basic versions of it. But you can’t really sell it. And so my advice to an engineer would basically be like build something you really know, because if you only sort of half know it, chances are you’re gonna find out there’s a whole bunch of skeletons in that closet. We knew BI okay, and we still found a ton of skeletons in that closet. And so many people who end up in this space - there’s just skeletons everywhere, and it’s kind of like, you should make sure you really know what you’re getting yourself into. Maybe you’re fine with that. Maybe you’re like “I just love visualization tools, and I’m happy to do that forever.” Great. You can build a great visualization tool if you are willing to invest a ton of time in it. But if you’re like “I can do this quick, and get rich quick” or whatever, it’s a slog.

Yeah, get rich slow. What were the hardest parts with Mode? Was it the endless customizations, or were there more hairier problems that you had to solve?

Well, I think the hardest – technically, there’s not a ton that’s that challenging. I think it is a technically complex product, because you’re basically building like an application that needs to do a lot of stuff. That there isn’t one thing there like “If we just do this one thing really fast and really well and really reliably, then we’re great.” It was a very feature-rich thing. And so similar to like a marketing automation tool, or CRMs, or whatever - those things just have to do a lot of things. And so it is technically hard to build that and keep it performant and make it like a good UI, and make it understandable; it’s just like easy to make a messy product.

I think honestly the harder part of BI - and somewhat related to that - is it is a very preference-oriented thing, that people… The way that you think about data and interact with it is probably different than the way that I want to interact with it, and the way that someone who is a marketer thinks about it and probably wants to interact with it. And so some of that is driven by like our abilities and our backgrounds, some of that is just driven by personal preference. Tableau makes a ton of sense to me. It’s just the way I think. Or some people will be like “Tableau doesn’t make any sense. I can only think in a spreadsheet.” And so you end up getting pulled in a ton of different directions, where everybody kind of likes something, but like needs it to be a little bit different… And so it just takes a lot of discipline to build a product that is really good for a certain group of people. You end up building often something that’s like fine for a whole lot of people, and I think it’s really hard to always build something that’s just like really good for this subset, because there’s always gonna be some adjacent group that’s 90% the same as the subset you’re building, and if you only add these one or two things they’ll love it, and you just drift.

Right… It’s a fractal.

Yeah. So you end up – it’s easy to basically spend a long time building something, and five years into building it you don’t know who your customer is anymore, because you’ve built a handful of features for 50 people.

Break: [32:27]

Alright, let’s go up a level now and talk about this other post, “Disband the analytics team.” So up a level, I mean not talking about building the tools, but like, is the whole endeavor worthwhile? This post is where I was just like, get out my popcorn… He’s going after his own industry, the industry that he’s a part of. You kind of liken it to a Ponzi scheme, but you say it’s not really a Ponzi scheme, but then also you drew the analogy to a Ponzi scheme… And some of this resonated with me, because I’m not a big data guy. I can see narratives well illustrated with data. You say a thing, and then you show the thing. And I’ve always enjoyed that. But when it comes to like data-oriented decision-making, I always find it kind of like – I don’t know, the cart and the horse, or the chicken and the egg… I feel like the tail is wagging the dog as I try to actually make decisions based on data. That’s just my personal experience.

And this post here, you’re talking about like “Do analytics actually do what we all say they’ll do in practice?”, which is inform you to make better decisions. This is like the version of the analytics team you’re talking about, where they’re like informing the leadership to make better decisions. And then it’s like, is that actually doing what it’s supposed to be doing? And you say this in that post: “Analytics, not as an industry or a technology ecosystem, but as a discipline might not work. The average company may never be able to make better decisions by hiring a team of average analysts. We can make dashboards and be operational accountants, but the fun, exploratory, valuable work may always be an indulgent, empty desert, and never the entree that we want it to be.” I know you have a follow-up post of that, I know that you’re very much analyzing and considering your own industry… But quite a question, and one that I was like “Dang.” I mean, is the whole thing a pipe dream? What do you think?

Yeah, kinda…

[laughs]

I mean, it’s like individually maybe not, collectively yes. So the backstory a little bit is – and anybody who’s worked sort of anywhere, but certainly worked around tech and stuff knows that data has been for a long time this kind of like promise of smarter organizations. Companies have to be data-driven or die, and to some extent - back to the sports bit - there is some “Oh, look what’s happening with sports. If these people don’t have data teams, they become terrible sports teams”, and every company is going to have that same evolution, where if you don’t Moneyball your business, then you will be dead.

And so analysts full-time, and like data scientists kind of – this is like the sexiest job of the 21st century. I think part of the reason for that was like – this was a famous blog post by a couple of people, DJ Patil being one, the guy who kind of coined the term data scientist… That data science is gonna be the sexiest job of the 21st century. And generally, there was a mindset a while back that these people are gonna change the way everything operates, and like just being smart with data was this huge advantage that everybody eventually needed to be able to take advantage of.

And it’s kind of been like a No True Scotsman thing. The No True Scotsman thing is you say this person’s a Scotsman, but they don’t like whatever, and it’s like “Well, no, a true Scotsman wouldn’t like that.” So you basically just like reject anything that isn’t a Scotsman by saying the person that you’re talking about actually isn’t a real one. Everything with data has sort of been like if a data team struggles to get stuff right, or a company tries to do this stuff with data, but it doesn’t work, it’s like “Well, they’re not doing it right. That’s not a true data team, that’s doing this the way that we all know is possible.”

And I think like this has happened for long enough, and enough companies have sort of like put investments in their data infrastructure, and are not that much better at how they operate, that I think you have to start wondering if it’s worth it. And to me, that isn’t necessarily saying the potential isn’t there. It’s not necessarily saying – like, if you could interrogate this data so smartly, you may find things that actually make you better. You may actually be able to find patterns in this that could make you a fundamentally better business. I think yeah, maybe that’s true; I don’t know. The problem is maybe that’s so hard to get at, that most people don’t have the ability to do it. It’s just such a hard problem to do that the idea that we all need to do this is like – it’s just not gonna work. It’s just like as a general discipline, it’s not valuable unless you are like a very, very, very sort of top-tier analyst, or whatever. And there’s not that many of those. So for most of us who are trying to do it, it’s like [unintelligible 00:41:43.23] We don’t really say this about other disciplines. We don’t really say “HR is the thing that every business needs to have. An amazing HR.”

[laughs] Right.

[41:53] There probably are some companies that HR truly is a differentiator. That they truly are better businesses because they have great HR leadership, that really genuinely does – like, the appeal of working here is like “HR has built this and made this an amazing place to work.” I am sure that is true. But we don’t really think of HR – it’s like, okay, but most HR is like “Do the job, it’s fine.” It’s possible data is that. It’s not a transformational thing. It’s not a thing that most of us can do as well as like the top, top-tier people, and so we shouldn’t try. We should acknowledge that we’re just kind of out here doing the kind of mechanical work, and that’s the extent of the ceiling that we have. And so I don’t know that I fully believe that, but I think I do… And really, it’s more of a question to me of just like – we’ve had this “Data is the new oil” type of stuff for a long time. And in reality, maybe data is just really low octane. It’s likee, if you’re willing to invest a ton of money in it, you can actually extract some energy from it. And if you’re a Google, or you have like very particular problems, you can probably actually make that extraction worth it. But if you’re a random business, I don’t know… The analogy I’ve used before - it’s like peat bog. Like, is peat that useful? Not really. You probably don’t want to put that much investment in and getting a bunch of peat, because it’s kind of like hard to get anything of real value out of it.

So yeah, I think we’ve talked about the promise of data for so long that we’re still kind of waiting for it, that at some point you’ve got to wonder if that’s ever actually gonna materialize.

Yeah, that’s interesting. I guess it’s kind of disheartening, because the story makes total sense. Even the heroic moment in which the causation was attributed to this thing, that was only found because of this dashboard, that changed the course of the business… Those are the stories that we tell ourselves. And like you said, I’m sure those things do happen from time to time, but that as a promise – you know, because there’s a lot of money that goes into it. There’s a lot of time, there’s a lot of risk. I mean, data as the new oil, also been called the data as the new toxic thing that you don’t want to have, because there’s all kinds of drawbacks to holding onto other people’s data…

Yeah, is it all worth it? And I mean, if it’s not worth it, then holy cow. But if it’s maybe kind of worth it, then it goes back to like “Well, do we need an HR department?” And then “Do we need a data solution?” Or what do you guys call yourselves. Do we need a modern data stack?

Yeah. Analysts, or whatever. Yeah.

Right. Do you need BI? I mean, I think maybe the answer is like low-hanging fruit for everybody, because there are things now where it’s relatively tractable to get a certain amount of insights that everybody gets, for a relative amount. But like the super-deep, expensive - I don’t know what it looks like - data stack is the one that maybe people in the future will opt out of, or I don’t know.

Yeah. And I don’t know that you need – it doesn’t mean like data is useless. It’s like “Yeah, reporting is valuable. You’ve gotta know how much money you’re making, and you’ve got to know how people are using stuff.” All that stuff I think is – it is really useful to do that. But a lot of times – and data teams do this themselves, and this is obviously to some degree self-serving… Reporting is seen as – I think this is sort of mentioned in the little bit that you read, or sort of referenced… Reporting is seen as kind of the prerequisite to doing the important stuff. That a lot of times people say like “The point of things like BI is so that data teams do less reporting, and work on more valuable, high-impact work.” And this high impact work is the thing you said, of like “We’ve found the nugget, the bit of insight that caused the pivot that changed the business”, and all that stuff. And those stories do happen. But so much of like what data teams do is “How do we get these prerequisites off our plate, so we can work on this valuable stuff?” And I think oftentimes, they get to the valuable stuff, and it’s like, there’s not that much value there. They just never actually deliver on it.

So if you’re a data person in this camp – I get why if you’re a data person, you would want to do that. It’s the fun stuff. It is fun to go digging through things, and trying to find stuff, and [unintelligible 00:45:54.26] stories to tell… That’s what I did in the very beginning of Mode, was writing blog posts about this with Miley Cyrus. But I don’t know that – a lot of times those answers are not trajectory-changing things for thee business. They can be, sure. But it’s hard.

[46:11] I do love those narratives that have data behind them, though. Do you remember the old OKCupid ones?

Yeah, that was great. It’s fun to do that stuff. There’s some appeal to it. But is it that valuable? I don’t know. Maybe.

Yeah. Maybe every once in a while. So in light of all of these – I guess these two discussions that we’ve had so far… The first one being the air is sucked out of the room to a certain extent, there’s opportunities in the slowdown, but they’re not going to be the new hotness necessarily… And everyone’s focus is on AI, and in light of the fact that analytics and BI as a – did you call it a practice? …as a discipline - that’s the word - is maybe not all that it cracked up to be… Like, what are you interested in today, here, mid-2024? You’re looking forward, you’re doing stuff, I’m sure… It looks like you’re in an office, so you haven’t completely gone to Mojito Island based off of your Mode sale, or anything… What’s interesting to you these days?

To the extent that I’m interested in sort of the data space, or the adjacencies around it, or whatever, I think it’s just the dynamics of like a big ecosystem like this are kind of fun. I don’t know that – I was talking to somebody about this a couple days ago… Data itself - I am not someone who is attached to like data itself. Some people are like “Look, I just love SQL compilers. I think they’re fun. SQL compilers are awesome.” Great. Some people love data pipelines, and think like the idea of just like figure how to move big things around faster and reliably is cool. And like, okay, that’s cool. I am neither of those things. I do think the OKCupid type of blog analysis stuff was kind of the first attraction I had to the whole data world, and that stuff is fun… So like just being an analyst I think is actually kind of fun. But in like the tech ecosystem side of it – part of the appeal, and part of the reasons I wrote about this stuff, the main reason was because I was working in it for a long time. But I think the thing that made it interesting to me was it’s a big ecosystem with a bunch of products that kind of have weird places to go, and we’re trying to figure out what to do with them, and it’s like, you have all these parts that you need to bring together in some way, and there’s a Rubik’s cube of sorts to figure out there. And I think that’s interesting. And I think, to the point of the air has been sucked out of the room - the gossip of Silicon Valley is interesting. Just the various dynamics of “Such and such raised money at this crazy number, and now there’s crazy stuff happening…” FTX is a fascinating story, not because crypto is interesting, though it kind of is, but it’s like because it’s just a crazy drama.

Yeah, yeah.

So I think that AI is interesting to me, not so much because it’s like “Oh my God, this is gonna become some super-intelligent thing”, or whatever. Sure, that’s kind of cool… It’s more that there’s a whole bunch of stuff to be like “What do we do with it? What are ways that we imagine this crazy new thing? What kind of other worlds can we think about that might happen?” You can come up with all sorts of thought experiments that aren’t quite thought experiments, but are maybe reality in these sorts of situations, and I think those things are fun to just think about. It’s also there’s a lot of drama. And again, I am a sucker for the drama. I love a good soap opera. So I think part of the appeal of Silicon Valley is that it’s kind of a soap opera.

Well, how might those interests then manifest here over the next 12-18 months? Are you just gonna continue to write? Are you currently doing analysis for – I mean, are you going to join the… What’s that, [unintelligible 00:49:42.04] journalist people? They’re kind of covering things, more of a journalistic endeavor… Or what are you up to, man?

I’m still figuring that out. I don’t know. I mean, yeah, I’ll probably continue to blog. There’s sort of the obvious adjacencies of what I did, of like “I’ll do another startup”, or go join some data thing. Or sell your soul and become a VC. Those things are there. I was talking to someone about this earlier… The ideal – this is very much like “What am I doing with my life…?”

[laughs]

[50:11] The thing, actually, like sort of the ideal is like these banks – you’ve got like an investment bank or whatever, and they have like these rotational programs where you end up “Okay, go work on the trading desk for a while, go work in fixed income, go work at the deal desk”, and M&A stuff, or whatever, and you spend like three months in each of these things… Basically, I’m like “I’ll just go on some professional rotation program around different things in the world.” Basically, I think tech is interesting, but there’s so many other kinds of interesting problems out there, that just seeing – even if it’s data-related, but like doing data stuff for an industry that isn’t selling software to other software people… There’s lots of interesting things in that. You just get exposure to like “Oh, this is like a fun problem.”

The person I was talking to, we were talking about casinos. Casinos are like a little bit of a greasy industry, but the problem there I bet is pretty fascinating. The things that casinos have to deal with I bet are pretty interesting. I have some friends who work in various political circles; they do the same thing, but it’s like a very different sort of set of problems. So all that kind of stuff I think is more to me about like “Is this a fun thing to think about?” And I say all that, and then probably it’s like “Alright, fine, I’ll sell my soul and start a new company”, or whatever…

Exactly. [laughs]

…but, but these are the things you dream about before you do the thing that is the boring thing.

Right. Until you just start your next dashboard company.

Yeah, exactly. Until you do like “You know what? I’ve figured out BI. I know what the right thing is. I know what people –” And it’s like, “Oh, God…”

Yeah. “From the co-founder of Mode, here comes…” You know?

Yeah, BI 2.0. Great.

Exactly.

Break: [51:42]

Alright. Well, before I let you go, I do want to talk about your post you wrote, I guess last year around this time, which I loved, and was one of the only posts I actually shared with our audience here on Changelog News, because I was like – a lot of your stuff is adjacent, but this was right in our wheelhouse. A lot of people giving conference talks, a lot of people making speeches, having to demonstrate their work… And you wrote “A Gambler’s Guide”, speaking of casinos… “A Gambler’s Guide to Giving Talks.” Some posts you can just tell like everything you need to from the subtitle, and your’s says “A bewildered audience is better than a bored one.” So I think that your premise is well known, just from the subtitle. But then you go on to back that up with some argumentation. Can you unpack the short version of that? Of course, we’ll link it up for people to read the whole thing.

Basically, if you give talks, there’s a lot of canned advice out there for how to give talks. A lot of it is things like “Don’t have too many slides, and talk slow, and repeat yourself” and all these sorts of things, where it’s like the point is to be very deliberate in your communication, and expressive, and all of those things. Okay, great. I am sure that is good advice. I am sure that if you implement that well, then you will give good talks, and people will say you’re giving good talks. I personally hate it. I struggled to do it that way. I also find those talks typically pretty boring… And so like I had to give a bunch of talks inside of Mode, just as like sort of company all hands types of things… And one of the things, partly because of the blog, there’s like a data conference circuit kind of thing that people end up on, and so I’ve given a number of talks at those things. And sort of somewhat accidentally, somewhat because I once when I was in college saw someone give a talk with this style and couldn’t turn away from it, I developed a style that was like the opposite of that. So basically, it’s like an outline of how I think about giving talks, which is essentially counter to all the advice that you typically get. No, it’s not totally – like, there are some people who have like the same sort of style that say this is the way to do it, so this isn’t like some totally novel thing. But basically, it’s okay to talk fast, slides, the main thing - you just have an astronomical number of slides… I probably averaged 15 slides a minute, and so it’s like a talk that is structured in that sort of way. And there’s some other stuff in there that’s a little bit more to your point of like the title and the subtitle.

Right.

I view talks as your primary enemy is just people getting bored and tuning out… And basically if you talk loud enough and fast enough, and slip slides fast enough, then people won’t get bored, and then you’ve won 80% of the battle.

Yes. I first saw this in practice by a fella whose name I believe is Giles Bowkett. He was in the Ruby community, and worked on some interesting open source library called Archaeopteryx. Now, talk about things that are memorable; like, just that word… And it was like a MIDI library that he was into… And he would give talks at Ruby meetups and conferences. And exactly what you described, with regard to the slides per capita was just completely insane.

[58:24] And it was the first time I ever see anybody do that, and I couldn’t forget it. I was like, every sentence, there’s some sort of reference. And he’s not talking about the slides at all. They’re literally just an adjunct or a sprinkle to what he’s saying. But he has it all timed out, to where like, every time, similar to what you’ll see on the Daily Show, or comedic things like Saturnday Night Live’s Weekend Update, where as the punch line hits, the background updates. And someone’s job is to time that sucker out, you know… Sometimes you’re watching it and it’s alive, and so it’ll be a little bit slow, and they’ll wait for the punch line to land, because the slide hasn’t… But Giles had it completely timed out. And it was like 45 minutes of pure action. I don’t know what he was talking about, I think it was Archaeopteryx, but this was probably 15 years ago and I still remember the talk… It was amazing. And I’ve watched probably dozens, if not scores of other talks throughout the years, and it’s like “Yeah, I’m sure it was good, but not gonna change my life”, you know?

Yeah. And I would love to see this, because that’s basically the exact way that I try to do it. The slide is sort of its own – to me, I view slides as basically like there is a second conversation going on on the slides. There’s a talk, but the slides – in some cases you have to have slides that are like “Okay, here’s a diagram” and it’s gonna help me explain it, that’s fine. But a lot of the slides are references to – they’re indirect references to whatever it is that you’re saying. And so if you’re trying to say something about “In the future we’re not sure we’re going to continue to need the same infrastructure we have today”, you show like a Back to the Future thing that has the future [unintelligible 00:59:56.12] And like you don’t really explain the joke. If people get it, they get it. Great. And people I think appreciate the sort of like [unintelligible 01:00:03.25] of it. If they don’t get it, they’ll be like “Whoa”, and then they’re bewildered. But they probably keep paying attention, because they might wanna see what the next –

Then they’re on to the next slide anyways, right?

Yeah. And the other thing is - this is to your point of flipping slides… You can’t do this unless you flip your own slides, and you do kind of know what’s coming. You don’t have to memorize it or whatever, and it’s bad, I think, to memorize it outright… But you do have to know how to time it, because there’s this bit in this blog post about rhyming off the beat. To me good, good rap songs are the songs that aren’t sing-songy. Back in the early days of rap it was very like sing-songy, where they’d always have it rhyme at the end of the line… And it gets really boring. And so now it’s like the rhymes are all interspliced within the bars, and like they’re off the beat… You have to pay attention, because there’s all these sort of rhymes that come at different times. And I think jokes basically have to be.

There’s a lot of people who will be like “I’m going to deliver a joke. New slide, here’s the meme slide, and now I make the joke about the meme, and y’all all are like…” It’s just clunky. It’s like, yeah, if you flip it right as you’re delivering the punchline, then the joke can land, as opposed to this “And now let me tell you a joke.”

And so yeah, I think there’s a tendency to think “I break up my talk based on slides, where I flip a slide, I talk to the slide, I make my point, I flip the slide, talk about the slide.” I think it’s much better to basically make 20% of the point in the previous slide, and when you hit the punch line and the point, flip the slide. If that punch line is an actual punch line, or that punch line is like something else, just like that, knowing that the slide flip is going to happen at an unknown point I also think keeps people much more engaged, where if I turn away I’m going to miss the transition.

Yeah. No, I love that style. And I definitely wanted to just expose our audience to that idea. Now, that being said, that particular conference talk or talk style is just one way that you can be entertaining and risky, and try to be memorable. That’s one way of doing it that I’ve seen to be very successful. And obviously, Benn, you appreciate that as well.

There’s also other ways you can go. Kelsey Hightower, who has been on the show a few times and is keynoter for all kinds of things, he goes no slides, no anything. The guy’s up there just talking. And he can get an audience engaged, and stay with him throughout 45 minutes to an hour just based on the storytelling. So there’s other ways of doing it.

[01:02:19.20] And the things that he does - sometimes live demos - are considered risky. Most people wouldn’t even have the guts to try this stuff. And I think that the overarching thought that I appreciated from that particular post is like “Take the chance. Take the risk.” Because anybody can be like average or slightly above average, and you’ll get the pat on the back or whatever, and you’ll feel good about yourself, but you don’t have very many opportunities to capture people and do something that everyone’s going to remember. And so maybe step out on the ledge a little bit and see what happens.

Yeah. I mean, to your point of like you have seen talks for forever, and you remember basically one. It’s like, that’s basically what happens. I don’t remember any of these things. You go to a conference and the thing is in one ear and out the other, and it’s like “Oh, that’s fine.” And even the ones that are super-practiced and super-polished, and like you have no feedback on, that you can’t be like “Well, that was bad, because they messed this up”, or whatever… It’s like, it was all good, and yet you remember none of it. And so I think it’s – yeah, I would rather go to a conference where I remember the thing, and to remember it you’ve got to be a little bit… Yeah, it doesn’t have to be this particular thing by any means, and I am not nearly charismatic enough to hold a room without like flipping through slides, and basically doing it with volume and speed…

Sure. Yeah, most of us aren’t. [laughs]

I have to have that cheats. But I think that - yeah, the goal is more of like “How do I make sure people don’t forget this thing?” Because that’s really – again, the enemy you’re fighting is attention, and just like memory, and being a total void to most people, instead of something that’s “Oh, yeah. I remember that. That was interesting.”

And in the software world oftentimes our topics don’t help us out very much, because they are dry, and detailed, and very specific oftentimes. And so it’s a struggle to make a talk both educational and actually entertaining, and off the wall, and like all the things that would become memorable. And so it certainly is a challenge. I don’t decry anybody for not trying, and going the traditional route. But I would encourage folks to, given the next opportunity, go out there and step out and take the gambler’s guide to giving talks from Benn’s Substack and see how it works out for you.

Yeah. And the last thing I’d say also on that is like - this is maybe not good advice. And this is true for the blog, too. I would rather people enjoy the time they spent reading it, then walk away feeling like they’ve learned something. To me, I very much would approach talks that way, where like I’d rather you just be like “That was a fun 20 minutes”, than be like “That was smart.” It’s really hard, I think, if you just have a smart talk, but it’s boring, it’s not going to matter… And it’s like, I’d rather be a talk with no point, that at least keeps people entertained. Obviously, if you can do both, great. But that’s tough.

Right.

But I prioritize basically like “This will keep people awake” more than “This is smart.”

Why don’t you prioritize it that way?

Partly because I don’t like smart – well, there’s a few reasons. One, I don’t think people remember smart. Again, you still can’t win people’s attention, even with brilliant ideas. Like, all the smart things that people remember, I think, overwhelmingly come from people who they are biased to believe are gonna say something smart to begin with. If you think about the smart things you remember from a podcast, or from whatever else, oftentimes those are like “I went in knowing that person. It was Jeff Bezos. I’m gonna listen to Jeff Bezos, and I’m gonna say the thing he says is smart, because he’s Jeff Bezos. And I’m paying closer attention.” But if you go to a random conference and see a person that you don’t know giving a talk, I think it’s really hard for people to be like “That was really memorable and smart”, unless you were memorable in the style that you gave it, too.

Right.

The other thing is - I don’t know, there’s a million things out there that are essentially teaching you stuff. I’d rather people just be like “That was enjoyable.” Enjoy your time – I don’t know, how many conference talks do you go to where you learn something that you actually, even if you thought it was smart, you actually do anything with it? It’s like, you take notes, you take pictures of slides, and then you forget they ever existed. I have never implemented something from a conference.

No…

[01:06:22.16] It’s like, yeah, you have like a little nugget… And so I don’t know, why chase that, I guess?

Yeah. So that has turned into my conference strategy, which is to go to zero talks, and hang out in the hallway, and meet people, and talk to them, and have fun conversations. Because while I know there is valuable and useful information in those talks, like you said, I’ve never left a conference talk and been like “I am going to go implement this in my business or in my software today.” I know there’s people who have done that, I’m just not one of them. And so I think we’re kindred spirits in that way. So also, hanging out on the hallways, man. That’s where the action is.

And that’s the thing I would want to fight against, is people thinking “Oh, I miss something by –”

“I’d rather be in the hallway.”

Yeah. It’s like, can you make it so that people be like “Oh, I missed out on something fun not being in that talk”, as opposed to “Oh, I missed out on that smart talk, but whatever, I can get a picture of the slides, or see the deck, and whatever supposedly useful thing is in there I’ll get in five minutes.”

I have a friend who works in TV, and he does like broadcast for Major League Baseball. And he says his goal is to make people who attend the games feel like they’re missing out by not seeing the broadcast. It’s like, he wants the people at game to be like “I am missing a better show on TV than I’m seeing in person.” And I think there’s a little bit of that to me, of like, yeah, the real value of conferences is gonna be like the hanging out and stuff with people afterwards. I want people to feel like they’re missing the better show by not joining the talks that are usually boring.

Yeah. I think good commentary can do that. I know I’ve been at the Road of the Final Four, because it came here through Omaha this last spring… Round one and two were here in Omaha. And so I’m at the games, and you’re seeing it live, but you’re wondering what the commentators are saying about it. Especially when you couldn’t really see that particular play, like did he actually travel or not… And you just wish you had the commentary in your ear. You still want to be in the stands, versus at home, unless you have a very nice setup at home… But missing that.

And I know there’s people that are going to baseball games and they’ll still turn on the local radio station, because they want to hear the commentary from the biased version, [unintelligible 01:08:28.18] because that’s fun. But yeah…

Yeah… It’s an ambition anyway.

Yeah. Love it. Love it. Alright. Well, Benn, thank you so much for joining me today. The website is benn.substack.com. That’s Benn with two N’s. I held off and didn’t ask you why there’s two N’s in your name. I just figured people ask you that all the time; even though I’m deadly curious… Benn, why are the two N’s at the end of your name? Is that from your parents?

It is. My name is not Benjamin, it’s Bennett.

Bennett. Okay.

So…

And so did you opt into the second N, or did your parents opt you in, or how did it work out?

My parents, yeah. I have never not done it. So when I was learning what my name was and how to write it, that’s what I learned. I don’t know if they were like “It’s been with one N”, and I’m like “No, I have two” when I was just able to talk… I’m not sure who was the originator of that, but I can never remember writing it any other way, so we’re gonna go with it.

Awesome. So benn.substack.com. Of course, all the links to all the things will be in the show notes, including the three essays that we talked about today. Anything else, Benn, that I didn’t ask you, or that you wanted to say before we call it a show?

No, I think that’s good. I appreciate you having me on.

I enjoyed it. I enjoy your writing. Keep on doing it. Keep entertaining me, even if I learn nothing. Just keep them coming every Friday, and entertain me along the way.

I’ll do my best.

Alright, that’s our show. We’ll talk to you all on the next one.