Jonas Christensen 2:40
Benn Stancil, welcome to Leaders of Analytics. It is fantastic to have you on the show.
Benn Stancil 2:50
Thanks for having me. It's good to be here.
Jonas Christensen 2:51
It is good to have you because, Benn, since I started researching for this podcast, I have grown increasingly fond of your writing and more interested in who you are as a person and all the thoughts that you have in your head around this whole world of analytics, data, the technology that drives it, and so on. And we're going to get into that throughout this episode. So I won't reveal too much because I for one, expect that we will hear it all from you. But two, you have thought in so many directions that I'm not quite sure where this conversation is going to go. But all I know is it's going to be highly interesting. So now over to you. Could you tell us a bit about yourself, your career background and what you do?
Benn Stancil 3:35
Sure. And I make no promises about this being that interesting, but happy to try. Yeah, so my career background and stuff. So I'm one of the founders of a company called Mode and we make a product. It's a BI product that is used by data science analytics teams to distribute work around our organisation. That's basically a BI tool. It's an analytics-focused ABI tool. That's been around for a while. So I started that with a couple other folks now, in 2013. So about eight years. At Mode, I do cover a range of different things as founders of companies, kind of, end up doing.You don't really have one particularly sticky role. You end up sort of bouncing around between different places that they need to be. So that's included some roles in product and marketing and like support and customer success and solutions, things like that. Currently, I'm more focused on, like, the marketing product side of things, a mix of, kind of, product strategy stuff. And basically, as you alluded to, heckling the internet about data stuff. Prior to that, I was on a data team at a company called Yammer, which was acquired by Microsoft in 2012. It was a early B2B SAS product that we had a data team that's sort of shaped very similarly to a lot of data teams are today, where our job was to help the business make decisions, make recommendations to people who are marketing a product, things like that. About which products to ship, how A/B tests perform, those sorts of things. And before that, I actually worked in Washington DC at a Think tank doing Economic Policy Research, which is a very different world than kind of the tech and analytics world. In a lot of ways, it's same page, different book. Like it's taking data trying to make recommendations and figuring out like what decisions to make. It's just the recommendations you're making are like ''What the Fed should be doing?''. And the Fed doesn't particularly care what random 20-something year old guy in the Think tank says to them. So in that way, like, same kind of part of the brain to think about. But you kind of cast it off into the ether, and nobody ever pays attention to it again.
Jonas Christensen 5:21
I've always wondered what Think tanks actually do. And maybe you've just described it to us here. Is there more to it than that?
Benn Stancil 5:28
I had no idea what they did until I joined one. I joined right out of college. It was kind of like ''Well, this is the job I got. I guess I'll take it''. It's basically like the bridges between academia and policy. Like in a perfect world, what they would serve as, like, academics doing research, writing papers, doing things that often move pretty slowly and have, sort of, like, academic aims, where it's ''What is the relationship between this thing and this other thing?'' or, like, trying to figure out stuff for the point of, kind of, advancing knowledge in some ways. Think tanks are basically trying to apply that to policy, where it's ''Okay. Because academics X, Y, and Z said these sorts of things, therefore, we should pursue these economic policies''. And so they have to, sort of, straddle the line between like ''What are other researchers say? And what are the political realities on the ground of what policies we can pass?'' and try to make recommendations to policymakers about - given the political constraints and sort of what we believe is the sort of academically right thing to do, here's what you should do. In practice, does it work that way? Not really. Because they end up basically becoming tools for politicians to make the case for the policies they want to pursue anyway. Where it's like ''Oh, such and such, I believe in gun control'', to take a random example, ''Therefore, here's a paper that says something that supports my policy conclusion''. Like to some extent, that's how they work today. So I think in theory, they can be a positive thing. In practice, it's kind of a insider DC industry. A little bit of like, people scratching each other's back and getting kind of cush jobs. I don't know. It's sort of loaded on the dole.
Jonas Christensen 6:53
Well, you certainly got out of that. And moved into a different world here. So you described how you went to Yammer and then worked in analytics there. And then at some point, you must have sat there and somehow decided that you wanted to start a BI platform, analytics platform with, I think, two of your colleagues from Yammer at the time. What were the moments or was the inputs that sort of crystallised for you: One, that this was the thing for you? Maybe the three of you to do personally. But also that this concept, the product that you'd thought up was something that the world needed at that time?
Benn Stancil 7:29
Yeah, so the product part largely came from the success that we had seen over an internal tool that our team had built. And one of the other two co-founders was the primary person responsible for building that tool. So we had this data team at Yammer. As I said, we were responsible for working alongside folks in operations or product, or marketing, or sales, or customers, etc. Basically, like help them do their jobs day in and day out. And so we ended up building a couple of internal tools to help us do that. There were basically like, query tools for us to go to write SQL queries and put charts on top that we can easily share with those folks and let them kind of extend the analysis that we've created. Which is basically what Mode does in a much more polished and powerful way. But like at its core, that's kind of what Mode is. We basically saw that be successful at Yammer. And then started to see a bunch of other companies were building similar internal tools like Facebook, and Airbnb, and Pinterest, and Spotify and Uber and stuff. All built kind of tools that were roughly shaped this way. And so because of that, that was really where the idea of like ''Hey, maybe there's a thing to sell here''. Once we saw a bunch of these sort of leading data teams starting to build tools that looked like this and it was like ''Hey, if these tools are going to be built, should somebody just make a product like this and focus on how to do this?''. As for, like, me personally, like why do it: I didn't grow up in San Francisco. I didn't, like, think about, sort of, the whole tech entrepreneurship thing as any part of, like, my identity. I think it was more of an opportunity to work with people that I enjoyed working with as a data person, in a lot of ways. You don't have that many opportunities to work at early stage startups, because you don't have any customers or you don't have any data. And so it's kind of like ''Yeah, why not?''. I think that I was excited about, like, the people, the opportunity. But I think the biggest part was, kind of, it sort of when in Rome, let's try it. Let's see what happens. You know, I think Silicon Valley, for better or for worse, there's a lot of things about that are - It's not a perfect place - but one of the things it does do is it gives people, certain people anyway, like the opportunity to do those sorts of things, at relatively low risk. Like, it's not like starting a restaurant where you gotta go collect $100,000 from your friends and family and take out a second mortgage on your house to be able to do it. Basically, the ecosystem of Silicon Valley is such that you can do that at relatively low financial risk to yourself. There certainly are other kinds of risks to worry about that aren't necessarily things that are obvious, but financially speaking, it's Silicon Valley makes that relatively easy. Yeah, and I was fortunate to be able to give it a shot.
Jonas Christensen 9:51
So it'd be a bit more specific on that. Did that mean that you had some sort of seed funding quite early in the journey? Basically?
Benn Stancil 9:59
Yeah and this was vice where we - part of the reason again that we did this was because we got lucky in that regard. So we were coming out of Yammer been acquired. It was a good acquisition. It wasn't like some astronomical thing. Certainly not by like today's standards. Yammer was for like $1.2 billion. Me and the other folks that I started with, me especially, like I joined Yammer three months prior to that acquisition. So this was not a thing that - I didn't make any money off of that. But certainly, like some of the early folks at Yammer did. And you know, the CEO and a couple other, like, very early employees or founders made a good bit of money. And so basically, in Silicon Valley, as soon as you make a bunch of money, you turn around and you, like, find random startups to invest it in. Like, that's kind of how the Pay-It-Forward part of it works. And so we had sort of the good luck of, after that acquisition, having people we could immediately turn to who knew us, understood the product we were building, because they had seen something kind of like that at Yammer. And we're basically looking to write checks for exactly that sort of thing. And so yeah, so we were able to like, raise enough money to basically pay ourselves from the beginning. And then from there, you end up in the sort of Silicon Valley venture. You're, sort of, tumbling down that rabbit hole.
Jonas Christensen 11:07
Yeah. What an interesting ecosystem that feeds on itself.
Benn Stancil 11:10
Very much, very much.
Jonas Christensen 11:12
Step 1: Come up with the idea. Step 2: Get funding. Step 3: Build product. Sell for a billion dollars. Step 4: Become a VC investor, and then the circle of life continues.
Benn Stancil 11:23
Yeah. You don't even have to do it in that order. Sometimes it's, like, Step 1: Raise money. Step 2: Come up with idea. There are plenty of versions of those two, so.
Jonas Christensen 11:31
Okay.
Benn Stancil 11:33
Once you're in it, like once you're kind of in that ecosystem, it is one that treats you very differently than when you're out of it. There's definitely an insider's game in it, for sure.
Jonas Christensen 11:44
Yeah, very interesting. We, the rest of the world who are not in it, observe it a bit from afar, and hear the stories and anecdotes that you can find, if you look deep enough on the internet. And some of these are in your sub stack, I think, Benn. But other than that, it's the wild stories that actually do meet mainstream media. Now, Benn, you're the co-founder, and also the chief analytics officer at Mode. I'm interested in a little bit more specifically what the company and the tool does, and what problems you solve for your customers specifically.
Benn Stancil 12:18
So the thing that we - As I mentioned, this came out of this thing we had built it at Yammer. Like, the idea was sort of inspired by that. The problem that we solved there and the thing that we wanted to translate into what we built with Mode which we have is - The reason we built a tool at Yammer and didn't go out and buy something was because we were a data team that tried to solve problems with relative technicalities. Like, our job was somewhere in between the, kind of like, capital D data scientist who's off building complicated models and doing PhD level math and the BI analyst who's supposed to be building a dashboard for an executive that they look at every morning on the way to the office. We wanted to use tools that were more tilted towards the first row, where it's like we were writing SQL. We wanted to write R. We wanted to write Python. We wanted to, kind of, ask hard questions of data that required us to, kind of, have like a very hands-on approach to how we manipulated it and we did a lot of like granular things in that way. But we weren't trying to build prediction models. We weren't trying to do things that were particularly hard math. We're trying to just answer questions that were the things that the business wanted us to answer. So it wasn't ''Show me a dashboard. How many users we have?''. it was ''Why is our user number down this month?'' Or ''Why did this particular cohort of customers perform so much better than other particular cohort?'' Or ''We just AB tested this thing. What do we learn from it?''. And all of those things required, kind of, the types of tooling that were typically useful for the, kind of, data sciences folks. But they also required us to have this, like, ability to interact with other folks around the business who weren't data people. Like, we weren't just, sort of, academics living in a hole, writing hard code, and then building some asset with it every six months. We're, like, sitting with folks on the marketing team being like ''What's the problem you have? Let's talk through it and try to solve this together''. And so really, what we needed was a tool that allowed us to work the way that we wanted to work,without alienating those other folks. Like, we couldn't send a Jupyter notebook to the CEO and be like ''Here's the answer to your question''. Like, they would not know what to do with that. And they'd be like ''Send me this other thing. Send me a PowerPoint''. But at the same time, we couldn't do our work in Excel. Like, Excel wasn't the right tool for us, or BI and Tableau. So the thing that we tried to solve was: What if we build a tool that has, like, the technical interfaces for the analysts. But then can be presented with nice charts with sort of easy, zoomable interfaces and stuff like that, for the people who are actually using it so that when we have something to share. We can package it up into something that's easy to say ''Hey, here's what we discovered. Here's what we think we should do about it'' and kind of collaborate through that. So that's basically what Mode is. It's meant to be something that's comfortable for Analysts and Technical users that want to live in SQL or Python or R. But also, is it alienating to other folks who don't necessarily know those languages and are just looking to help understand what's going on? Sometimes, they want to poke around, explore more. Sometimes, they want to do a little bit, like, self-serve type of stuff. But it needs to be something that, kind of, flex across that spectrum of: Good for the very technical folks but also good for people who just want to look at a chart every morning and not think too much about it. Just want to, like, know what's going on and don't need to see SQL or any of that.
Jonas Christensen 15:19
Yeah, nice. So it's really that there is a chasm that's still, but definitely back in what 2013: You said when you started Mode, yeah, a huge chasm between the regular BI and then this, what I called Analytics, which is answering the why, not the what, at the more extreme, the more data sciences and machine learning type of work. At that time, you wouldn't have had anything like Power BI. Definitely not in its current iteration. There's a plethora of tools that I don't want to list to risk not including someone but all the cloud providers wouldn't have had the their offerings and all that stuff, certainly not to the same extent as now. So, you would have really been quite early in market at that time. I see that market as a very competitive market now. There are quite a number of BI tools that have matured a lot in that period to, I suppose, try and incorporate things, like, what I'm sort of imagining when we need to describe Mode. The ability to pull in R, Python, Code, and so on, and generate things then and there. But also, the big guys have woken up to it. And I know, certainly what I experienced when I go from organisation to organisation is the usual Microsoft strategy of ''Hey, you can get this for free''. And then all of a sudden, there's world dominance with the Power BI because it's free in a relative sense with your corporate subscription. How do you see this BI market playing out, If I may describe it as that. You can correct me if you don't want that label on it. This sort of business intelligence tool market playing out in the next 5 - 10 years?
Benn Stancil 16:54
Yeah, so certainly, there has been a lot of evolution from when we started, in a couple of ways. And so, One is: There was definitely a very big gap between, like, analytics space. Which is basically what we - you know, Mode's original name was ModeAnalytics.com. And the analytics part was very much, kind of, the space that we gravitated towards, which was - it's not BI. It's, kind of, true data science in the sense that we're building production models to recommend what you should buy next on Amazon. And so it was very much the why, not the what. But also not the predictive fancy stuff. There has been a lot of tools in this space now. Some in analytics. Some in BI. Selling, sort of, more data science stuff. Like, it's crowded basically by people, kind of, all over the place. My view is two things. One is: There will inevitably be some, like, sort of, consolidation around things that people generally believe to be the best. Like, it's just the market right now, it's, kind of, like, completely falling apart. Generally, Silicon Valley. That sort of thing puts a lot of pressure on, like, young companies and companies that don't have any real traction. And I think, like, it'll be difficult. Like, there's a explosion of companies that got started in the data space in the last few years, in part because it seemed easy. There was like: You could go out, you raise a bunch of money six months in. Every idea seemed like a great one. And they may all be great ideas. It's not to say those companies aren't gonna last. But certainly, like, it is not nearly as easy to do that anymore as it was. So I think there'll be some pressure to like, kind of ''Okay, let's pick and choose some winners''. Partly because that also comes from the customers themselves. When things are going great, you buy 10 tools, because why not? When things aren't going so great, you're like ''Well, we maybe gotta pick three''. And so I think there'll be some consolidation in that way. In terms of what I, like, moves towards, I mean, I am very biassed in this because of the perspective that we have at Mode. But my, like, belief, and in some ways hope, is that we don't go back to the way it was, where we have these like drag and drop BI tools designed very much for business users and, like, the CIO types. And then Analysts and Technical folks are off like having to do things in other places. I think that what we've developed over the last 10 years, really, as an industry is the need for these kinds of data teams that are focused on solving harder problems that aren't just doing sort of rote reporting. And so, the right tooling to me is one that allows those teams to work in the same place as the people who are trying to get, like, their BI needs met. There aren't like clean lines between what is BI, what is analytics and what is data science. It's all just different modes of operating with data. And the best thing is like figuring out ways to support all of those things in, kind of, a nice interoperable way. And so, I think it's like: Yes, there will be some consolidation. And yes, you know, it's a crowded space with a lot of different options. But where that eventually goes is ''Okay, how do we make it, so that we can like work on these things together?''. So Power BI also can support more technical users or whatever it is, you know. Tableau can support them potentially, or something like Jupyter Notebooks starts to support more of the BI use cases. I think we're more, like, ''Let's try to serve the whole spectrum of uses'' as opposed to ''Let's silo each of these things into their own buckets'' when the silos aren't particularly well defined, anyway.
Jonas Christensen 19:58
Yeah, it's really - Yeah, I compared this space a lot to IT in the 90s. And sort of, you know, we all got our personal computers. Really sort of at home. Not for the first time but it really started being commonplace. And I know as a kid, I had to learn DUS to actually, get into my PacMan games and all that stuff, right. And all of a sudden, you had windows 95 that almost any idiot could use. There's probably a space for something like that in analytics, where you just make it so much easier for things to flow. So that you save time and you don't have to necessarily learn new coding languages all the time to do some tasks that you haven't done before or whatever. You mentioned that there is this plethora of data analytics platforms available. And I'm thinking here: full spectrum. Not just what Mode and your direct competitors do. You know, if you start right from data ingestion, ETL, all the way to output: There are just so many of these guys, and a lot of them with pretty big valuations. Maybe half of what those valuations were six months ago, at this point in time, at the end of May 2022. We'll see what happens going forward to that share market. But there's so many tools here that are, kind of, doing really well. But often I sit here and think ''Do we really need all these tools to become these superpower data users or is it just a bit of overkill?'' Or maybe the right question is ''When do we need those tools?''.
Benn Stancil 21:29
So I have a somewhat cynical view of this, I guess. I think that no, we don't need all these tools. Not really. Like, do we need them?'' To me, it's like ''Do we need all the channels on DirecTV?''. Like, sure, there are people who are gonna watch some of them.
That's a great way to put it, actually. I haven't thought of that before.
We don't need them. The fact that there are thousands. Everyone would have somebody watching it? Yeah, probably. There's somebody out there who watches it and somebody who's probably really into the random reality show on Channel 912. I have no idea who it is. But it's probably not great, like, that we have all of them. And I think that it's probably economically not sustainable to have all of this. And this, I think gets a little bit to where the, sort of, funding market was. Part of it is, like, it was relatively easy to raise money. And there were a bunch of data people, like a lot of data people went out and started companies. And I think that is - I did this. I can't blame anybody. Part of my reason was, like, ''I'm a data person. I have an opportunity to do it. Why not?''. And so certainly, if it was 2019 instead of 2013 when I started, probably would have thought the same thing. It's, like, I don't regret any of that. But I think there are probably a number of folks who start these companies, less to solve the problem and more because being a founder in Silicon Valley is like: that's one, how you make money. It's, like, kind of where the cash is. There's just a thing that that's what you do. That's the next step. And I think there's been a like strong pull away from data folks being operational leaders, and like running data teams to go out and start a company. And part of that, too, I think is: Data as a field, in some ways encourages this, because what do you do when you hit the ceiling? Once you're a Data Director, what do you do next? We don't know yet. You go be director of data in another company. And like, you can kind of sort of hop around between those different things. But it's not easy to become an executive because data folks don't really have that career path yet. And so a lot of people are like 'Well, this is probably the next step''. Especially in the markets, it's easy for that. And again, like starting a company is a little bit of the thing that Silicon Valley tells you to do. Told me to do and I listened. I think that's, kind of, where people got published. And so I think as a result of that: Yeah, you probably do a lot of companies. Especially in a market where money's really easy. Get over funded. Get over valued. Maybe they all come out great. Maybe this market is just so huge, that doesn't matter. That's possible. But it seems unlikely that that'll happen again. It seems like ''Yeah, we've got 1,000 channels on Direct TV. Just not that many people watching.''. At some point, you know, these things can't sustain with 10 people watching.
Jonas Christensen 23:59
And I think the analogy of a TV channel is good. But it also is different in one way, which is: When you don't like something on a TV channel, and you can hit the channel button and you go to the next thing. But it's not so easy with tools. There's typically a learning curve for staff and people wanting to learn that tool. There's vendor locking. There is implementation costs and so on. So it actually is not that straightforward to just, sort of, paste together lots of tools in an organisation with the hope of multiplying your analytics efforts. And the whole lens of ''People are starting this because they hit a career ceiling'' as well. It's quite interesting. We might get back to that later. I want to explore your thinking on that. But with the fear of not coming back to this topic, I'll just hold off on that for a few minutes. Because I want to just dig a little bit deeper on this topic of analytics technology, data technology. I'm really thinking of the full end-to-end stack here. So my inbox is full of new tools every day. They want to present themselves to me. What do I mention? Cloudera, Alteryx, data bricks, the big cloud providers: All these guys. They say ''We can easily solve this whole pipeline for you and it's easy. Drag and drop and you combine some notes and your analytics dreams will come true. You will generate information with almost just the click of a button.'', which may or may not be true once it's set up. And then at the other end, you have more specialised tools, like some auto ML tools that are also trying to streamline, automate or augment the speed and quality at which we can produce a, sort of, very advanced machine learning models and so on. So it's really that whole stack. How do we bundle all these stuff together? Because you've described it a little bit, but you typically have both analysts and business users wanting to access and do some analysis on data to varying degrees. You're not going to get your marketing manager to do machine learning, or maybe some do by accident. But that's probably not a good idea yet. Maybe you can tell us whether that will be the case in the future. But it's really this sort of, you know,''We make it easy for everyone''. That's kind of the sales pitch, typically. Is that realistic? And what's the right bundling of all these things to sort of try and serve the full spectrum of users in an organisation?
Benn Stancil 26:23
So, I think that's tough. There's a lot of ways it could probably work. My general view, though, is: We will find ourselves wanting more bundling than we think. There was a post I wrote about this recently. It was basically on Microsoft and like Microsoft is - you mentioned Power BI - it's free, right? I mean, it's in-effect free. You can pay for, like, premium stuff. But it comes bundled with things that people already have. And a lot of people use it for that. And you can - there's certainly an argument to be made, sort of like: It's a monopoly. And it's, like, some version of them just undercutting the competition, because they have this distribution channel that nobody can compete with. And I think that's partly true. But I also think there is something, like, the people who are using it and buying it, find that convenience more valuable than finding like the exact specialised tool that they need. The fact that it is bundled: It's not just a pricing thing, or something that gives Microsoft, like, power to distribute it. It's also something where as customers, it's kind of nice to have everything together. And I think that there will be places where, kind of, to this point of like, they're being tons of JS startups now. All of these things are these, like, kind of, small point solutions for small issues. That, again, I think, are places where that matters. And it's not that those are bad solutions. It's just ''Can you build an entire product out of it?''. Which to me implies two things: Can you build enough product, like a top product enough to make money, to make the company sustainable? But also do customers care about that thing, being good enough to buy a separate thing to do exactly this? You know, not as an analogy of this is like: You got to choose which tools to stock your kitchen with. We don't have that much space in our kitchen. And you can't - Like, there are lots of specialised tools that are great. Like, an electric juicer is great. Those things are really convenient to use. Love using 'em. Way easier to handle. But it takes up a lot of space. And, like, when you're making some decision about what to use, it's like ''Do I really want to take up half my counter space for this juicer that I occasionally use?''.Like, probably not,
Jonas Christensen 28:14
Can you speak to my wife about this? We've got way too many things in our bench.
Benn Stancil 28:19
But like, exactly. I think that's at some point, it's like ''Well, I would rather have all of these specialised things, but I am giving up something for that. There is a sacrifice that I have to make.'' I suspect that the sacrifice most people would rather make is ''The tool doesn't work quite as well. But it's easy for me to, like, not have to do the frustrating things of putting everything away and glueing it together and managing this giant, like, overhead of 1000 things''. And so, I don't know where those lines then get drawn. Like, does that mean that we should bundle ETL and reverse ETL? Probably? Exactly the right boundary? Maybe not. Does that mean that we should bundle warehousing with your visualisation tool? Probably not. But like, maybe. I think those lines are hard to draw. But I do think probably will happen where we want to draw fewer of them than we currently do.
Jonas Christensen 29:07
Yeah. It's a really interesting conundrum. Because it is exactly that. Where do these tools that and stop? And the complexity underneath is, to some extent that - This is a comparison that I make, at least often, so this is my opinion. You can shoot it down if you think otherwise. But often data projects, whether they're implementing solutions in the business, or just that front end of getting your suggestion or your ETL layer stood up, they all feel, smell and seem a little bit like IT projects to the business. But they're actually very different because typically, historically, when we put in software solutions, they're meant to be producing and reproducing the same output given the input all the time. Part of it is an automation of sorts, right? It gives you other things, software of course. I'm sort of simplifying things a lot. But the nature of data is that it changes all the time. And typically, because someone changed something in one of those source systems, or the behaviour of what's in the data changes, and so on. So, part of the challenge is that it's very hard to ''automate and streamline'' something that changes all the time, because you've got to have someone looking at it all the time. Do you feel that that's a fair comment? Or am I being too simplistic in my view of that?
Benn Stancil 30:27
And how hard is it to automate this stuff. And this can be very challenging.
Jonas Christensen 30:31
Yeah, so my point is it often could get compared to, you know, IT projects that are more sort of generic IT project, as opposed to analytics projects, data projects, where we're trying to pick up this raw material and turn it into insightful decision making tools of sorts. But there is this variance in the raw material all the time that doesn't necessarily exist.
Benn Stancil 30:53
Yeah.
Jonas Christensen 30:54
In technology, in IT systems, or software.
Benn Stancil 31:00
Yeah, I think this is true. I think it's true in two ways. There is a layer of complexity to data things that doesn't exist in other domains. That is kind of what you're describing, which is: You have the data itself, and then you've got, like, the things that manage all of it, and ,sort of, the applications on top. You basically, like, - two very unstable things that are, kind of, having to run into each other. Where in most cases, you've only got one or the other. And like, you see this not just in, sort of, the administration of tools, or in, sort of, building systems. But even, like, at Mode, for instance, we build a - think of it as a BI tool, basically, - the types of ways you have to think of permissions in that tool is very different than what you would think of - Like, you look at it and it's kind of, like ''Just add, like Google Docs style permissions to it''. But it's very different because you're having to edit - edit access is also, sort of, write access, or query access to the database underneath it. There's a lot of levels to it, that aren't there. If I can edit a Google Doc, I can just see what's in the doc. If I can edit a query, I can, like, in theory, see anything behind that query. And so I think that exists kind of across the stack, where it's just like: Yeah, you have to deal with the complexity of the data itself, plus all of the application around it. So, like, I'm generally not terribly optimistic about our ability to, like, automate a lot of this. I think we basically have to get better at designing the system that we use and figure out what works best and just, kind of, have everybody follow those principles. There's another version of this too, which is: To what extent do we automate or, like, AI our way to actually making sense of this stuff. The point of all of this at the end of the day is, like, supposedly to learn something. At what point can we, like, have AI tell us what we need to learn. I used to be very bearish on that. I'm actually a little less now. I actually think at some point, we could get to a point where that could be possible, or at least like help us out a lot. I think we're ways away from it but that I actually think is, maybe, something we will get closer to in the next 5 to 10 years.
Jonas Christensen 32:51
And there's probably some sort of Pareto distribution called AI. It can help us maybe point out in what neck of the woods to look. Might not tell us the exact insight but, you know, ''Look over here. It looks like in this part of the data, there is something to look at''. And that's already coming in a lot of these BI tools as well.
Benn Stancil 33:11
Yeah, there's a lot of the, kind of, like, gesturing at things you should look at in tools now of like '' Hey, this looks kind of funny. Why don't you check it out?''. Which I still think is a little bit under-baked. But I think, like, we could get closer to. We can get closer to a point where like, the analyst job has actually helped out a lot by these sorts of things, I think.
Jonas Christensen 33:30
Yeah. I think that would be an analyst's dream. So, it might seem a bit threatening to some that actually the job gets automated. We all hear about that. But it really is almost always a blessing in my opinion, when that happens to jobs, that they do get this help from technology to speed things up. But we're never going to get fully automated or not in our lifetime. It's kind of what I'm hearing you say. So folks out there, don't forget your Python, you'll need it still. And that's also free. I think that's a big part of why that's so popular. Benn, when I started it was all SAS, all the way. Now that's not so much the case. But that's probably something for a different episode.
Hi there, dear listener. I just want to quickly let you know that I have recently published a book with six other authors, called ''Demystifying AI for the Enterprise. A Playbook for Digital Transformation''. If you'd like to learn more about the book, then head over to www.leadersofanalytics.com/ai. Now, back to the show.
Benn, I'm really interested in - Now going to that space of the people in this industry because this show is called Leaders of Analytics. It's really about the leadership that's required to drive all this forward. You made an interesting comment before around how there is almost this glass ceiling for data professionals in what they can achieve in organisations in that space because we're forging a path as we go in terms of selling all this stuff into the business, but also the executive presence or the lack thereof that we have now, and we need to sort of argue and push away. And if I may use those words, - maybe they're a bit too strong - but we need to sort of create our own executive roles as we go. We're not the first specialisation in history to have done that. But it does mean it's not the big challenge. Nevertheless, what are the challenges there for data professionals and how do you see that play out in different career paths and different career decisions that people make? You already mentioned, you know, some ended up as startup founders to become executives almost. But yeah, I'll let you put your own words on it.
Benn Stancil 35:40
So, I think there's a an industry question there, or, kind of, dynamics of the world question there and then a, kind of, question for ourselves. The dynamics of the world question to me is: People don't yet know what to do with senior data people. There are executive level, like C level data roles. The closest ones typically are things like Chief Data Officers, which actually are kind of more of, like, an IT and governance role in lot of cases and, kind of, like an analytical one. There's sometimes like Chief Strategy officers, which, sort of, can be shaped like what an analyst does. I think there's not really like a clear place to go as a senior data person. And also, there isn't a clear place to go as, like, a senior non-executive data person. So, if you're an engineer, you can become, kind of, like a principal or a staff engineer. You can become, you know, the sort of engineers that can work in places like Microsoft or Google and make a ton of money, have very good jobs and not be responsible for like executive decision making. But instead just be responsible for, like, the hardest of technical problems. You know what I'm saying? Like, sales reps. Very senior sales reps, sometimes just stay sales reps forever, because you make a lot of money as a sales rep, If you get the good customers, the good leads, the good territories, things like that. For data folks, like, once you pass the sort of - Okay, you move up to, sort of, easy career ladder. Not easy in the sense of moving up, but, like, easily to map one of. You know, a junior role to a mid level role to a senior role. It's kind of ''Where do you go next?'' and most people end up in roles of management. But a lot of data teams are capped at, like, mid management levels, because they report into a CFO or CIO or a CTO, and data folks aren't gonna get promoted to being the CFO. They're not going to get promoted to being the CTO. Like, that's not the path into that role. So what happens there? In some cases, people end up sitting around and waiting for like a VP role or something like a Chief Data Officer role. I think a lot of people probably try to basically hop to a different department, where they become a little bit more of a strategy thing, or like a Chief Operations Office, sort of deal, or whatever. Those sorts of things, I think are like: We haven't really figured out what that path is. So partly as an industry, I think it'll be and my belief is: If we think that, sort of, analysts provide value by beings rigorous thinkers, strategic thinkers, things like that, there's no reason why that can't apply to the executive team as well. In the same way, we have, like, embedded analysts in a product organisation or a sales organisation. It might be helpful to have embedded analysts. It's basically, like what a CTO often does, at the executive level. It's thinking about, like, the biggest and hardest problems. But it isn't doing it from a management and leadership perspective. It's doing it from a ''Okay, I'm in the rooms where all these important decisions are being made. And my contribution is being someone whose job it is to think about these things from their perspective''. I don't know, maybe. There is, I think, another aspect of this, which is that analysts have to evolve a little bit to be cut out for that. Where there is a tendency - and this is this is like a broad generalisation - but there is a tendency of analysts of wanting to say like 'Here are the facts. You decide'' and to speak a little bit in, like, sort of, couched language and caveat things. And when the, sort of, push comes to shove about what decision to make, it's often like ''Well, I need to do the analysis to figure that out''. And a lot of like, executive roles need to be: You don't know what to do. You got to make a decision anyway. And people are looking at you to, like, be a leader in those cases and say ''Data is inconclusive. I still gotta make a decision. I still got to live up to that decision. I still got to stand behind this. I got to fight for it. The buck stops with me if it doesn't work out''. And I think there is a desire among data people, because we've kind of been taught this way of, like, ''Do the research and the analysis and let the data guide us''. That is, in some ways, like a punting of responsibility. If you're an executive, no matter how much the data guides you: You are making the decision and you have to own that decision. And so I think there's, like, some of that, that - again speaking in very broad generalisations - that data folks would have to evolve on their side. Think about this as, like, ''No, my job is to make a decision, despite uncertainty. Not to remove the uncertainty from it''.
Jonas Christensen 39:46
Yeah. I think there's a couple of comments on that for me. We talk about data scientists. And another way of putting that is: Decision scientists. It's actually the one that makes the decision using data and if you can make that decision and help people do that quickly, then you're going to be invited into those boardrooms, to help with that. And that's kind of what you're describing, I think. There is something called the 40-70 rule, I think, which is made famous by Colin Powell. He said something along the lines of you need 40-70% of the information to make a decision, but don't wait for the rest. And that's kind of what you're describing, in the sence of the decision under uncertainty. Which is something that - if I join your generalisation - is inherently uncomfortable for people who like to get the facts straight with data analytics. So, that definitely is something that we need to be able to bend their own rules a bit in that sense, and meet in the middle with everyone else who want the answer now. It's an interesting development. Now one of the things you talked about was the path that you can take from starting your career to the top of the organisation. I'm not saying that it's the top of your career necessarily, but that hierarchical promotion that we're talking about here. In other functions, like, let's pick marketing, for instance, that also had this challenge for many years. That, you know, you can become a marketing manager, but typically not more. Then you started seeing the Chief Marketing Officers pop up, which is probably kind of where we're going. But then it sort of went to more like Chief Experience officers, the Chief Digital Officers that typically had marketing plus other things underneath. Is there a bundling of data analytics and something else that perhaps could make that successful? You may not have thought of this, so I'm sort of putting you on the spot a bit.
Benn Stancil 41:32
That's a good question. It certainly feels like it could be mixed in with other things. The thing is: I think a lot of those things are things that themselves already have a fairly strong executive presence. So, there's a lot of overlap between what data teams and finance teams do, for instance. People have a negative reaction to that because there is like a general aversion for a lot of data folks who report through finance. Like, they see finance has a very different role than what they do. And I think that's generally, like, a fair thought. But there is, like, a lot of, kind of reporting operational overlap there. But also from an operational perspective, like what data teams, like operations teams do. And operations teams that are focused on making businesses more efficient, figuring out ways to help everybody else get their job done more effectively. Like, data teams are kind of operations teams in that way. Where their improvements aren't, like, building processes in Salesforce, but are helping you understand what you need to do better. So, I can see those things kind of come together. But at the same time, like, a COO is very much a thing. A CFO is not going to go away. Like, those aren't roles that are sort of easy to subsumed by data folks. The other thing is: Bigger companies have business strategy teams that are typically, kind of, like a consultant type of deal that'll come in and sort of help people think through stuff or whatever. That's sort of the most obvious parallel to me of, like, an existing type of organisation. But that's, like, not so widespread enough that. I don't know if you can really get very far with that. I guess, there's, like, an IT pattern here, where data and IT have some overlap. And IT also, like, wasn't a very big thing 30 years ago. Now, it's obviously a very big, important department for a lot of companies. So data could just become another version of that. Though, certainly, it's worth pointing out that if everybody has a seat at the table, then nobody really has a seat at the table. Like you can't have an exec at everything, because at some point that exec team just is like ''Well, now we have the real exec team'', that's back to what the original exec team was.
Jonas Christensen 43:31
And there's a subtlety in all of this, too, that the CEO will not want too many direct reports, because it's more stuff to manage. Defeats the purpose of delegation to some extent. So there's only a number of whatever that number is. Eight to ten, typically, something like that. That's sort of the maximum for that. I can see that CIO analogy playing out in this space. Because I think now, no company will go ''We don't have a CIO''. That's sort of ''What? What do you mean? That's silly.'' It's our job then to make sure that in 10 years, 15 years from now, that's talked about in the same way, when we talk about Chief Analytics Officer, Chief Data Officer, Chief Data Analytics Officer. However you combine it. Yeah, we'll see what happens there, Benn. The future is still bright, despite this invisible ceiling. But something for us to think about. The more we're aware of it, the more we can solve it. Look, Benn, we're close to the end of this. I've got a couple of questions for you to round off. But is there anything that you would like to get across that we haven't discussed in this space that we've discussed today?
Benn Stancil 44:41
No, I think this was good. If you want more of the rants, they're on the internet. Won't overwhelm you with them audibly. If you want them on the internet in written form, have at it. There are plenty out there, as you've alluded to.
Jonas Christensen 44:55
Yeah, we'll comment on that now, actually. So Benn, where can people find out more about you? Get a hold of your content and connect with you?
Benn Stancil 45:04
Yeah, so the blog you referred to is on SubStack. It's just benn.substack.com. Other than that: Twitter. I'm not on Twitter a ton but occasionally. Twitter handle is just Benn Stancil. And then LinkedIn is the same. For the professional networkers: LinkedIn. For the trolls: I guess, it's Twitter. And then for the people who just want to read stuff: Substack. It's probably the best place.
Jonas Christensen 45:30
Please don't troll Benn. But please go and read his Substack. It's really brilliant. I enjoyed the thought provoking content that you have on there. But also, I'd say you're talented writer. You write beautifully. There's a lot of humour in there. So it's part, the thought provoking knowledge acquisition and part, just enjoyment. So that's a perfect place to go really, for your latest analytics thoughts, I'd say.
Benn Stancil 45:57
Appreciate that. You know, writing about, like, obscure tools for data people is pretty boring stuff. So, I got to do something to keep myself entertained through it. So we're not exactly talking about Hollywood Gossip here.
Jonas Christensen 46:10
No, and good on you for making it interesting for the rest of us. There are plenty of dry articles on those things out there. So the last question is one we always ask the guests on here, which is to pay it forward. So who would you like to see as the next guest on Leaders of Analytics and why?
Benn Stancil 46:27
So, there are two people come to mind for me, because they are people who do the part of the job that I think is really hard and I think they do it really well. So, like, writing a blog about tools and stuff to me is like ''Okay, you can say whatever you want. It's easy''. That's the easy stuff. The hard part is,like: building teams and helping people in their careers and, like, being a leader in, like, the proper sense of the word. Not somebody who just, like, yells in the void on the internet. I mostly yell on the void on the internet. But there are two people: Maura Church, who runs a data team at Patreon and Erica Louie, who runs a data team at DBT, are both two people who I think do a an exceptional job of, like, the actual hard part of the job. Which is building a team and making people's careers better and doing all those things that are, like, the actual difficult work of being a leader. And so, I think there are two folks who a lot of folks could learn a lot from that isn't just, like, ranting about Databricks, which anybody can do. But you know, building a team and inspiring a bunch of people to be better at their jobs. Not something I'm good at. Something they're good at. I would love to hear from them.
Jonas Christensen 46:27
Brilliant. They two great recommendations and I will be contacting them very shortly. Benn, thank you for your recommendation. Those two recommendations, I should say. Benn Stancil, It's been such a pleasure to learn from you and get insights into Silicon Valley and how it works, the future of analytics tech stacks in all our organisations and how we might go through the evolutions of leadership in data science analytics in general. All the best for you and for Mode in the future. And I hope to catch up with you again in the future on this show.
Benn Stancil 48:08
Thank you so much for having me. And yeah, I really appreciate this as well.
