The observatory
May 11, 2022

Benn Stancil, Mode

Benn and Kyle Kirwan, CEO and co-founder of Bigeye, converse on the rise of code-forward data analytics tools and why, despite better tools, the demand for data analytics is still unmet.


Read on for a lightly edited version of the transcript.

Kyle: Hello, everyone. Welcome back to The Observatory. Today, we're going to be chatting with Benn Stancil. Benn, welcome to the show.

Benn: Thanks for having me. Good to be here.

Kyle: So many of you may have heard of or seen Benn before. Benn co-founded an analytics company called Mode. He's also been pretty popular on Substack. And just a general voice in the modern data set conversation recently over the last year or so especially. So Benn, maybe we can get started with just “what is Mode.” How'd you decide to co-found that? How did you get started?

Benn: So Mode is basically a BI tool for analysts. Originally, Mode came from an internal product that we used at the company I was at before founding Mode. Me, and the other two folks that I started with, were on the data team at a company called Yammer. Yammer was a social network for companies—it was Facebook for Work before Facebook for Work existed.

We built an internal data product, basically, to help the data team that we were on to distribute our analysis around the organization. Essentially, we were providing BI-like services to the organization, but in a more modern way. We'd get a bunch of ad hoc questions from people trying to figure out which products we should build, who we market to, etc. We'd write a bunch of SQL queries, build a bunch of charts, ship them off to people, and then they’d want to update them and look at dashboards. That kind of stuff.

At the time, we didn't have any tools that could solve that problem. This was legacy BI, which was too constraining for what we wanted to do. Tableau was there, quite to the same degree that it is today. But also, we wanted something that was more code-first, and SQL a native part of the experience. And that didn't exist. My technical co-founder was leading the data engineering team at Yammer and built a couple of internal tools to help us do that. And one of them was effectively a SQL-like editor in a browser with charts attached to it.

After Yammer ended up getting acquired by Microsoft in 2012, we talked to a bunch of other people around Silicon Valley and realized that a lot of people either were building that same product—Facebook, Airbnb, Pinterest, and Spotify had a version. And all these tech companies were building these same kinds of like data tools for their data teams. Or the company who we talked to said, “This thing is great. Can we buy it?” And so once we saw that, we said maybe there's something to this and actually a trend that's changing the way that people think about data.

It's not just traditional BI, but this kind of analytics team-focused BI. And so we decided to go out and try to build that product. And that's basically what Mode has been ever since.

Kyle: You mentioned like building it around sequel building it around data-is-code. I feel like those have been sort of evergreen principles that are still very popular concepts, and the tools that we're seeing now are kind of just rising in popularity.

Benn: Mode started in 2013, which now seems like quite a long time ago. When we talked to people, there were some people who very much got it. They wanted a tool that was a SQL experience. They wanted a way to enable people to write code and weren't looking for a kind of drag-and-drop or choose a bunch of things from dropdowns kind of experience. Initially, that was kind of a hard sell, honestly. There were teams that really liked that perspective.

But there were a lot of folks who were coming out of traditional BI tools. They used things like Qlik or Business Objects and were realizing those tools were breaking down. But they just wanted a more flexible drag-and-drop tool, the same as Google Analytics, but could do more stuff in it. Then there was kind of a mentality shift. Instead of thinking, just give me a drag-and-drop tool with more features, actually, the right way to do this is to learn SQL and do this in code. And in 2017, they started to think that maybe SQL is the better way of doing this. Now we have data teams that are sort of more code-forward.

I think that that transition is sort of fully complete in a lot of ways now, where data teams are very much SQL-focused and comfortable in that. There still is very much a need for the kind of accessible interfaces for people who don't want to write SQL, but a tool that has SQL front-and-center is no longer alien to a lot of people.

Kyle: I was at Uber for quite a while in data there, and that was absolutely what we saw as well. There was a SQL training class, and a lot of people who came through Uber learned SQL on the way in the door. And I think that that's increasingly a common thing: SQL as that lingua franca.

Benn: Uber had a tool—the data Querybuilder, I think it was called. So Querybuilder was built by an engineer who left Yammer shortly after the acquisition to go to Uber and was inspired by the same internal tool that inspired Mode. So the tool we had at Yammer was literally called Querybuilder, and they just picked it up, sucked it into Uber, took it a bunch of different directions, and added a bunch of things to it. And obviously, things like Bigeye are inspired by a lot of that infrastructure. But yeah, the data tool that Uber had was more or less identical to the data tool that Yammer had, and was more or less identical to the inspiration for Mode.

Kyle: Yep, I know the person you're talking about, and a shoutout to them if they're watching. But, yeah, Querybuilder, a lot of mileage out of Querybuilder. I want to pivot a bit to another question.

So, Benn, you've been working on analytics since at least 2012 or so. And that does predate a lot of the major evolutions in the data landscape—Snowflake, Airflow, dbt all came on later than that time.

From your perspective, you mentioned a sort of lineage or some inside baseball in the evolution of data. How do you feel like data has evolved since then? And more specifically, sort of this analytics end of things? Do you feel like it's getting better? Or is it moving in the right direction? What's your overall perspective on it?

Benn: In some ways, for sure, but in other ways, kind of not. So prior to being in tech at all, I worked in DC doing a bunch of econ research for a while. So my background is econ. One of the sorts of like mysteries of econ is basically like, where did the productivity of the Information Age go? We have a bunch of computers. And in theory, that should make us way more productive. But in productivity statistics, the way people measure GDP and things like that, you don't see it. People aren't actually way more productive, despite the fact that they should have been given all the new technology that we have had over the last 30 years.

Data teams feel a bit similar to me, where we have a lot better tools. One of the things we want to try to do is we wanted to build dashboards of Salesforce data. We hired a full-time engineer to maintain pipelines between Salesforce and a vertical warehouse. We had basically a full-time DBA to maintain a vertical warehouse that we had to pay for the hardware to run. We built internal tools to build those dashboards. It was orders of millions of dollars a year to be able to put those dashboards together that now you could do with things that are mostly free. You pay $1,000 a month for a small Snowflake instance, using the lower tiers of Fivetran and dbt. And tools like Mode, you can get that for basically nothing.

So in that sense, the life of an analyst is way better than it used to be. But are analysts doing more interesting things? Every data person sort of is like, “I don't want to build dashboards. I don't want to build pipelines. I want to solve the interesting strategic problems.” And are we solving those problems? It's like, can you still just build dashboards? We've made the cost of producing dashboards and data pipelines and putting data in places so cheap. And we've increased the demand for what we want to do with it so much that we haven't actually gotten out of the trap of just kind of building the first floor of what a data library of tools would look like.

It used to be that support teams would not come to you and ask for dashboards of support stuff because they know that you're just trying to build dashboards that show how you spend your millions of dollars of marketing every month. And the support dashboard is way down the list. Now, it's really easy, too, and so you get more demand for these sorts of things. And so analysts are still kind of chasing a lot of other things to build before they get to these strategic questions that are the things that are supposedly motivating all of us to get in this career in the first place.

So to me, it is less of a problem of tooling now and more of a problem of analysts having to basically say no. Because we can, doesn't mean we should. Our job is still to answer these big, important questions. So we obviously had to prioritize the big ones, because we could only answer three a year. And now it's like, we can answer 1000 a year. We should still be answering the 1000 most important ones, not the1000 that come across our desk first because they all take a third of a day. So I think there's like some cultural stuff to change around to how we actually operate in a way where it's so cheap to produce stuff. And I'm not quite sure that we've really figured that out.

Kyle: The supply goes up, and so does the demand, right? There's this equilibrium point that stays right where it is. So maybe that's a great segue into another question that I had.

In the last year or so, you've gotten a lot more vocal than I think maybe you were previously. Maybe that just comes from being a founder, you know. A growing company keeps you pretty busy, I'm sure. You started a Substack about data that's got some readership, and you've been doing more speaking events.

Is there something specific that motivated you to start taking a bit more of a public role? There's a lot of conversation right now about exactly what you just talked about—the process on data teams and the philosophy behind how to do data correctly or efficiently. There's a lot more talk about the modern data stack. What prompted you to get more involved in that conversation recently?

Benn: Oh, I was bad at every other job. You know, this is as well as anyone. The job of a founder at a startup is often to do whatever somebody else isn't currently doing. And so when we first started Mode, actually, there were three of us. At some point, four or five of us. Most of those folks were engineers, and they were capable of building the product. Derek, our CEO, was out doing CEO things. You're familiar with those—talking to investors and being the smiling face of the company. I wasn't good at that one either. And so I didn't really have much to do.

So I basically wrote blog posts. And at that point, there was much less conversation about data ecosystem things. But I wrote blog posts that essentially scraped data from the internet and talked about stuff unrelated to technology but data-oriented. If you go to like Mode's blog, the very first blog post from four days after the company was founded, it was about Miley Cyrus and the VMAs. And like YouTube data that I scraped about the Miley Cyrus video that won a bunch of awards had like the worst YouTube reviews you've ever seen. And I was like, “What's going on with this?”

Data, people want to talk about data things. There's a kind of attraction to the 538-style of thinking and conversation among data people. And so, you know, it kind of resonated with the people that we wanted to connect with. At some point, it was like, the thing that you are best at is probably talking about these sorts of things. It's also something I enjoy—it's interesting to be a part of these conversations. There wasn't some sort of big motive behind it or grand plan, to be entirely frank. It's interesting to be involved in these conversations, and it’s certainly useful for Mode to be able to understand what people are saying, the direction of the industry, and what folks like you think about where things are headed. So, there is, unfortunately, no great story behind it.

Kyle: I feel like that's how a lot of things happen, right? They evolve sort of organically over time. There's plenty to talk about in this space. It's this interesting conjunction of it's rapidly evolving. The way things get done is still in this is like a Cambrian explosion—the era that we're in around the tooling and the practice is changing very quickly. So there's no shortage of interesting topics. And the people who've seen the evolution of the space through different stages have different perspectives.

Benn: There are also a lot of interesting things to me about Silicon Valley and the way that the whole ecosystem works. Even talking about the unbundling of Airflow is interesting. Or whatever bundling or unbundling we're currently in, to me, that's also kind of culturally interesting, too. And so, I don't know. I enjoy being a part of these conversations.

Kyle: Alright, then it's time for rapid-fire mode. I have three questions for you. The first one—I think this will be highly divisive—so think carefully. Pizza or bagels?

Benn: Pizza. Bagels have their place, but you can always eat pizza. I don't always want a bagel. If someone shows up with pizza, I'm like, yeah, that sounds great.

Kyle: Breakfast, lunch, dinner. The heavily loaded bagel sandwiches. Definitely a strong contender. But yeah, I think the universal applicability of pizza is tough to contend with. All right, Benn, question two. When you are coding, I assume you still do at least once in a while, what is your favorite thing to listen to?

Benn: When I write in like monospaced text editors… Sometimes it's code, and sometimes it’s God knows what. Like top 40, I don't know. I don't have some deep cut like this my house music from this 2008 trip to Malta. Olivia Rodriguez is pretty great. I don't know what to tell you. I see she's got a new show on Disney Plus. I'm excited.

Kyle: Encanto, I've heard, is the upcoming big hitter when it comes to the music.

Benn: “We Don’t Talk About Bruno,” number one. But the other song, the song from the sister, is better: “Surface Pressure.” It's the dark horse. Highly recommend.

Kyle: Last question. You have been in data visualization, charting, and analytics for quite a while. What is the most underrated data visualization that you think deserves more attention?

Benn: Animated ones, basically. The New York Times has its really fancy data stuff which is well beyond the capacity of what anybody in the world, other than The New York Times can do. But a lot of things they build where you'd basically like click through them, and they tell stories. Or you scroll through them, and it sort of evolves as you see it. And they seem sort of cheesy, and they're the kinds of things that you wouldn't necessarily expect in some McKinsey deck.

But intertwining a narrative with visualization to me is such a powerful way to tell stories and get people to understand what you're looking at. But it's the types of things that t you should be paying attention to, rather than saying, here are some charts.

There was this chart that The New York Times also did that was like COVID cases over time. And it was this circular chart, like a pinwheel. And it got panned. Why in the world would not just make a line chart? And like all data topics, Twitter lost its mind over it.  And to me, there is something to be said for the 3D bar chart that everybody hates, but can't look away from. So ultimately, if you have the world's most beautiful visualization, and nobody pays attention to it. So what?

And it's the things that creative people see unnecessarily ornamentation. And I think in a lot of cases, it's no, that ornamentation is to get you to pay attention to it and actually think about what it's telling you. And so I think there's something to be said for visualizations that are adorned in ways that don't seem like they should be, but in practice, makes you actually look at it.


It's like the train wreck of charts. You can't look away.


I've never had a chart that hundreds or thousands of people have talked about, so I can't tell them how to make them. They clearly have, and so good for them. They figured something out. The joke's on us.


All right, Benn, thanks for chatting with me today. This was a lot of fun. If you want to learn more about Mode or subscribe to Benn’s Substack, we're going to put the links in the description below. Encourage you to check both of those out.

And thanks for watching. I'll see you all next time.

share this episode