Ragnarokast Episode 12 

Steven

I’m Steven. 

Spencer

And I’m Spencer. Welcome to Ragnarokast, your podcast for all things marketing and MarTech. 

Steven

Hello everyone. 

Both

We’re the Co-CEOs of Ragnarok. 

Spencer

This is gonna be the worst podcast you’ve ever been on. 

Boris

Excellent. 

Spencer

I’m already getting that feeling that you’re, you’re a pro. You’ve probably done this more than we have. You know what 

Steven

The problem is? We’re just not paying enough money to do this podcast recording.

Spencer

Oh yeah. 

Steven

If we were to throw like 38 more dollars at it, Boris would be okay. 

Boris

I, yeah, I think it’s basically about 30. I think we’re talking about 38. 

Steven

We gotta get this guy a gift card. 

Boris

Yeah. Like, like the desk. Like you could sell this desk and then we could just, you could skip it. Yeah. 

Spencer

Welcome Boris. 

Boris

Thanks for having me. 

Spencer

This is Steven, who you already know. 

Steven

Who’s that? 

Spencer

We are the co-host of this podcast. This will be aired or, or posted, whatever terminology we want to use at an indeterminate date in the future. Okay. 

Boris

Do you think it’ll be in 2024? 

Steven

Oh, absolutely. 

Spencer

Yes. It’ll be this year, I think. Next month? Question mark. Two months… 

Boris

Wow. Maybe before the Super Bowl. Who knows? 

Spencer

So If you’re just tuning in now 

Steven

or on the Super Bowl,

Boris

Even better. Let’s, let’s aim for that. Let’s aim for the Super Bowl. Yeah, that’s a good Super Bowl release.

Spencer 

Yeah. Yeah. 

Boris

Whenever, whenever that is. 

Selin

It’s not February. February Insider. We have Insider for February. So maybe March. So March. 

Boris

It doesn’t matter. 

Spencer

Okay, so March. 

Boris

Fine. 

Steven

Maybe we should pay another $5 for that Asana project. 

Spencer

Hannah, where are you, Hannah? 

Steven

The person who actually does the schedule isn’t here. Yeah. Things like dates don’t matter to us Boris. 

Boris

Time. 

Spencer 

Time is. What is time? Time is a construct. Anyway, so this is releasing at some point, which if you’re listening to it right now, is now. 

Boris

Yeah. And so as far I’m concerned, this is the release, 

Spencer

This is the present. 

Steven

This content is timeless. 

Spencer

I will admit Boris, that I am personally newer to the Censusverse. The Censusverse. 

Boris

Censusverse is good. 

Spencer

Yeah. And so, or the 

Steven

Census, we call it the Consensus internally, 

Boris

I think. Yeah. I was gonna definitely call our first conference that, and that’s what I, we did a small virtual conference, and we called it Consensus. 

Steven

Oh, awesome. 

Boris

I mean, it makes sense for a conference, right? 

Steven

Yeah. 

Boris

There’s a crypto company—

Steven

Hey, that works out really well. 

Spencer

Oh, Census Con, Consensus, right? 

Boris

Consensus. Yeah. 

Steven

I like that. 

Spencer

That’s cute. 

Boris

I think it literally means to bring people together anyway. 

Steven

Oh that’s so cute. 

Boris

There’s a problem. Me problem. There’s a crypto company that named their conference the same thing. So I need that to end, then I can really own that name. 

Steven

Oh, you into crypto, bro? 

Spencer

You guys wanna talk about the blockchain? 

Boris

Yeah, no. We’re gonna put, we’re gonna put your customer data on the blockchain, and everything’s gonna work out. 

Steven

Yeah. 

Boris

It’ll just be real slow. But it’ll be so good. I guess who needs a clean room when you have, you know, the blockchain? 

Steven

I get NFTs, bro. You want, you wanna trade NFTs? 

Boris

I think I might still have one from like, just trying it out. 

Spencer

Yeah, I’ve, I’ve got an NFT of a painting of a cat that I bought for $2. 

Boris

Yeah, I think I have a JPEG of a cat. Yeah. I said let’s try this out. I think I’ve probably spent more in fees than than it’s worth. 

Steven

Yeah. 

Steven

Somewhere around $38 in fees. 

Boris

No, no. Come on man. Like less than that. 

Spencer

So NFTs and Bitcoin bros aside. 

Boris

Yeah. 

Spencer

Do you mind introducing us to yourself and to the Censusverse in general? 

Boris

Totally. Totally. So, my name is Boris Zerbe. I am the CEO of a company called Census, which was, I started about five years ago. My life is one of always building tools that make people more productive, more successful, having more leverage in life. She started my career at Microsoft on Visual Studio, which is a tool that I think a lot of people now have learned about, even in the data world, which is kind of cool. And the fun part when you work on developer tools is you work at the very bottom of the, very bottom of the software industry, right? You’re the tool that people use to make other tools that, other apps that people then use. And I worked on the lowest level. So Chrome was built using our tools, and Windows was built using our tools. And the fun part about that is that I learned very early that if you can make those people like 1% better, you have a massive effect on the world, right? Because you make Chrome 1% better, you make Windows 1% better than everything else gets benefit. And so leverage is a very powerful thing. I mean, Archimedes wrote that way before we were born, but, but I felt it very, very viscerally when I worked at Microsoft. And so about 11, 12 years ago I, or 12 years ago I guess I quit to start my first company. It was a company called Meldium that was an employee identity product, what you would call single sign-on, it was a, in plain English, it was a password manager, but designed for teams and for companies. That was because we were like, SaaS is everywhere or it will be everywhere. At the time, people had like two pieces of SaaS at every company, but now it’s 300, right? And so we felt like if you’re gonna use all this SaaS, which is great, everyone having their pane of glass, great tools for $19, then employee identity, your login would be everywhere and therefore it’d be nowhere, and you would have chaos. And so we built a tool that people could really use to, to log into their apps and that, as a company, you’d have a visibility over who your employees are and where they are and what tools they have access to. So that was like, everybody wins, and you can have more SaaS, which is more productive and you know, just gives people more freedom to, to use the tools they want. And we sold that business. And a few, you know, as I started working at the company we sold to called LogMeIn, just a very apt name for what we were doing. 

Spencer

I’ve heard of them. 

Boris

Yeah. Yeah. They were based in Boston. And what we found, what I found was we were a very product-led company, right? The company that we had sold Meldium was fully self-serve. Right? It was, it had a lot of usage analytics that were very useful to under, very helpful to understand how to talk to customers. And when we sold the business, and we started working with a much larger organization, we started working with their sales and marketing team. 

Steven 

Right 

Boris

And they tended to write and engage our users terribly. And I thought it was a cultural problem, but in reality, there was like a tool problem, and it’s ’cause they lived in their panes of glass, right? Salesforce, Zendesk, et cetera. 

Steven

Yeah. 

Boris

And they didn’t know, know what I knew about our users. ’cause it was sitting in the product and analytics side of the house. And so it felt like we should fix this, right? We should connect those two worlds. I didn’t want to connect them in a way that would just create more pipes for the world that, that are disparate. ’cause there’s already lots of ways to, to connect system A to system B, right? 

Steven

Right. 

Boris

There’s tools like Zapier, and they were around, what I wanted is to do the same thing we did for the, the single sign-on and password style, which is there should be a central place, and we should federate outta that. You know, we should try to find a way to get it out of one place. ’cause then it’s like a broker that, that you can trust. ’cause it’s, you can centralize, 

Steven

Right 

Boris

And so that’s how Census was born. That’s why the name is Census. ’cause the idea is there should be just one. And, and so we built it about five years ago, it was our first version and we started working with kind of high-scale software companies.  We built it out of the data warehouse, which was kind of nascent at the time. So products like Snowflake and Redshift and Databricks. And now we’ve got like 20 of them. But you know, we said, what if we use the data that you have, the data infrastructure that you have, and just give access to everybody else in the company to what the data and analytics teams already have? And that’s what Census does, right? It allows a marketing team to build audiences or campaigns directly against the same data that the analytics team uses to, to make dashboards for the CFO. So you get the same quality of data, but you actually get to use it for, you know, engaging with customers and doing operational stuff at your company where, you know, hopefully, money gets made and, and better brands, better brand experiences occur. 

Steven

Obviously this is a problem for a long time because things were kind of living in a, in a, a sort of marketing automation software. 

Boris

Yeah. 

Steven

You know, five, six years ago it would be something like, you know, this is before like the Brazes and the Iterables totally, really came into play. 

Boris

Totally. 

Steven

We’re talking about like, you know, exact target vis-a-vis Salesforce Marketing Cloud or Oracle responses. And you had kind of a crusty SQL Database. You had to kind of meld into that. But that’s also around the time that like Segment was starting and some of the kind of, you know, what would we become the big CDP players in the space. So what was sort of the biggest pain point that you felt like you were solving? Even like this stuff was available, it just wasn’t easy, I guess, to access. 

Boris

Yeah. You know, things change relatively fast in our industry. And then, of course I live and work in, in San Francisco, where we often are even like a couple years ahead of everybody else changing. So there’s always like, we’re just always on the next thing in some ways. But the, the pain I felt, the thing I saw was that there were tools for connecting data together that, that’s for sure, not new. And I think Segment, you’re right, was had been around at that point for a while, and what it did very, very well at the time was take your events from a website, right? Or from your app and connect them very seamlessly to marketing tools. And it did nothing to help, let’s say your CFO do analytics on your billing on your Stripe data, 

Steven

Right.

Boris

It did nothing to help the sales team figure out which accounts to upsell based on their usage. It didn’t really help the support team on kind of providing different levels of SLAs for different kinds of customers. 

Steven

Right 

Boris

So they were just like, we weren’t trying to say what is Segment not doing? In fact, all of our early customers, all of them had Segment, 

Steven

Right

Boris

And then they bought Census as well, and they also bought Snowflake as well. And I think they were just solving different problems and different sides of the house. 

People were buying Snowflake and databases like it. Because you need, ultimately, you, you, when you have more and more sources, more and more channels in which you engage with customers, you need to coalesce it somewhere. It is actually like a, you can’t just magically do that, right? There’s not like, you can’t just say, take my Stripe data and my Klaviyo data and just tell me who’s doing what across those two things. 

Steven

That’s right. This one’s object of one’s user. 

Boris

Yeah. And like they’re living in two different execution engines, two different websites, two different permissioning models. Like that doesn’t just, there’s no, you need a technological solution for that, right? And so data warehouses were literally designed to solve this problem

Steven

Yeah.

Boris 

To dump all your data, do batch processing in order to provide answers to the questions that, you know, leaders needed at a company. But those were very much around slow after-the-fact processes. Right? 

Steven

Right. 

Boris

They’re like, figure out my business in post. 

Steven

Right. 

Boris 

What happened last quarter? What happened last year? Tell Wall Street. And we just built it out of that because it, it had all the information, and it could be used in very flexible ways to, to make the scenarios that were not well done by, let’s say the, the singular pipe that existed to marketing tools we could solve it with, with that platform. And then it had all these fun side effects that, over the last five years, have become real shifts in our industry. 

Steven

Right.

Boris 

Well one is the thing I just said earlier, like the amount of SaaS just keeps increasing. Right. The amount of things that read and generate customer data or data as a whole, call it any widget information, like you said, user object, et cetera. 

Spencer

Yeah. 

Boris

Just keeps increasing. And so the need to, to centralize it and make sense of it across multiple tools just keeps going up as well because no single tool seems to be able to get a complete view. ’cause it’s, it’s actually getting harder, 

Steven

Right. 

Boris

Just like logging into apps, like, it’s like there’s more and more of them, right? 

Steven

Yeah. 

Boris

So you, you, you can’t just kind of shoehorn the solution Yeah. 

Steven

Until Okta supports it all, 

Boris

Until Okta supports it all. Which is, you know, why that kind of technology makes sense too, right? You, you need some kind of, in that world, you need an identity provider. And in my world, I felt like you needed you, you just increasingly need some broker for that data, and it needs to be flexible. And so if you’re only ever gonna do marketing out of two tools, you know, then maybe you can get away with the same solutions that we had five years ago. 

Steven

Right. 

Boris

But I think one thing that doesn’t seem to be stopping is more and more channels for more and more users across, you know, more time and you’re gonna need to serve them well, personalization matters, but also, like just reaching them where they are at the right time just becomes more subtle and more important as the years go by. 

Steven

So that’s an interesting thing to, I wanna hone in on that. 

Boris

Yeah. 

Steven

You know, people are gonna use more than two tools because there’s more use cases. 

Boris

Right. 

Steven

But yeah. You know, you’re kind of in this place in the market right now where, you know, the, the, we have this sort of best-in-breed stack where people were focusing on very much doing this kind of narrow use case that they solve across many industries, but they’re kind of building more and more into becoming cloud software again. You know? 

Boris

Yep. 

Steven

Vis-a-vis Klaviyo adding a CDP component 

Boris

totally 

Steven

bras around the corner from adding a CDP component. 

Boris

Totally 

Steven

this point, you know, it’s, it’s only a matter of time before we see that with a lot of other, I even think of like Intercom as an example, which 

Boris

100%. 

Steven

So in, in that case, right? Does it, do we have in the future still a sort of large democratized set of tools that we need to centralize? Or are we actually shortening the amount of tools like we did five, six years ago, you know, kind of going back into the cycle where Salesforce built into that or Oracle? 

Boris

Yeah. It’s a really good question of the forces in our industry that, cause you know, the usual way people describe this is like the bundling and unbundling, right? I actually think if everything is trying to be a system of record, which is what every SaaS product is trying to be in some way, right? It’s hard to imagine all of them becoming a system of record. Like yeah. They, they, they can all say they are, but it, it seems very hard to imagine any of those products that you just named doing a good job on every dimension, right? So you’re gonna use Salesforce, let’s say, let’s use the classic the ultimate, like in terms of unified platforms, right? 

Way more than Intercom or Braze or, or Klaviyo. Let’s take Salesforce 

Spencer

“Unified” with the, You know, heavy quotes 

Boris

Yeah. That it does everything in some form, right? But you want to provide a great customer support experience. Listen, the, the service cloud is not great, and you might decide to use Intercom instead. So, so it only takes a couple of exits Yeah. On any dimension that you wanna serve your customers for the SaaS product to suddenly no longer be the one truth of everything. Oh. And like, are you gonna do your billing in Klaviyo? I, I doubt it. 

Steven

Yeah. 

Boris

It’s very hard to see a universe where there isn’t disparate data in d disparate systems. And so to me, there is always gonna be pressure for people that need to get correct data, consistent data to get some kind of coalesced version of it somewhere. 

Steven

Right 

Boris

By the way, we might enter a world in 10 years where Snowflake and data warehouses are no longer the way people do it. 

Steven

Right. 

Boris

But you’re gonna need a technology that is very good at storing all the data, joining all the data, and letting you make insights on all the data so that you can take the right action on a user and a universe where all of the products you just named are good enough at doing that is the equivalent of all of them being also Snowflake. 

Steven

Right. 

Boris

How about that young? That’s fair. There. Which may be, hey, like maybe Braze is like, we’re gonna be as, we’re gonna be Snowflake. Alright, let’s, let’s, I like that. I like that ambition. 

Spencer

Pretty sure they’re built on Snowflake, so that would, that would cause Some 

Boris

Right. 

Spencer

Some issues, but, 

Boris

Right.

Spencer 

But so in terms of, you know, data accuracy, so you’re talking about bringing in data from all over the place. How do you know if it’s reliable? How do you know if it’s useful? Like is there a consensus help with something like that? Yeah, 

Steven

Well that’s actually one of the biggest problems that we see with people trying to jump in and, you know, and use data warehouses as the, as the true sources, it’s not necessarily clean. 

Boris

Right. 

Steven

And it’s not necessarily well documented for sure. And it’s reliable on a couple of key, you know, skill sets, let’s say of, of folks who rotate quite a bit through different companies. And so that knowledge doesn’t transfer very well typically. 

Boris

Yeah. No, I think you’re right. I don’t believe today there is yet a singular magical solution that will tell you this data is right. The closest I think we have to, that is probably the checks and balances and controls we have when we invented gap accounting. 

Steven

Yeah, That’s fair. 

Boris

That’s probably the closest thing we’ve invented as a society to say we have a system to determine that these numbers are correct. 

Spencer

I mean, in the email world, you know, there’s like, there’s tools like kickbox to to do email verification Yeah. That 

Steven

Says is as reliable as how many records are in their database. Right. Which I think is the same thing that we have across the industry. Right? It’s as reliable as as as their understanding of that interpretation 

Boris

Yeah. 

Steven

As opposed to an aligned process that we all have around this. 

Boris

I think that’s the key. I think the right answer is not to say, oh man, the data could be wrong. And it’s like yeah. Mistakes will happen. Yeah. I recommend most people at most companies to take an improv class. Yeah. They’re gonna be like, what, what’s he talking about? 

Steven

Yes. And?

Boris

And not for the yes and but for 

Spencer

No, but? 

Boris

for the embracing of failure. Right. One of the most fun exercises, in improv is that they’ll force you to come up with story and like you will fail. Right. At some point. It’s like, I, I don’t, the words don’t come out right. ’cause they’re pushing you to just be creative on the, on the spot. And the culture in improv when that happens is to, I consider that success. Literally everyone in the room will go, yeah, you failed. It’s a really weird, cathartic kind of religious experience. And, my point is, yeah, your data’s gonna be wrong at some point. Not every marketing campaign is going to be perfect. 

Steven

Yeah. 

Boris

Like, you are going to make mistakes. The key is I think to create a good set of incentives and a process by which it improves. And I think before Census came around, there were distinct groups in a company trying to solve this Right. Without helping each other. Yeah. Right. So there was a CFO working with an analytics team and the CEO to say, what did we do last quarter? 

Spencer

Yeah.

Boris 

In sales, what did we do? And like which user is the most active and which ones drive the most sales of the, the ray bands, whatever. Right. And that was almost wholly separate from a team of, let’s call it, marketers or growth marketers who were trying to say, I wanna build a campaign for who might like use this thing, and I’m gonna make an email. And both of them can make mistakes. But ideally you want those mistakes to lead both of them to get better. Yeah. And to not have waste across those two. And so that’s, that’s really what we tried to build with Census is to say, if we can make the one side that marketing and sales and support were not touching because it was of the world of product and finance, let’s call it. But we make it just as accessible as using the data in Braze. If we make it just as accessible, like the same number of clicks, the same user experience, but you’re using that data, then every improvement they make you get and every discovery of a mistake you find in marketing, you can improve on their side as well. 

Spencer

Yeah. And we, and since we work in, you know, primarily retention, personalization, we often end up working with both marketing and product teams. 

Boris

Yeah. 

Spencer

And we find this a lot where the marketing and the product teams aren’t talking to each other. They have completely different tools. And part of our like unspoken job is to ourselves at least, is to try to get them to slowly consolidate. Yeah. Because not for our sakes, but because it’ll actually make their lives easier. 

Boris

Look, when I started the company, good luck. I had a real hard time talking to my parents about what it’s because password manager, they can understand Right. Kind of. But CRM database marketing automation, none of those words mean anything to Mom and Dad. Right. And it took me about 45 minutes on that first call when they’re like, what are you doing? And I finally found a very simple example that everyone can appreciate, especially when you live in Canada where my parents live, which is everyone has called customer support of a major corporation. And it is the most common thing for you can ask, right? To ask your parents to be like, how many times do they ask you who you are and what the information is even within a singular call. And my parents were convinced that this was some kind of weird security thing. And I’m like, no, that’s what they tell you anyway. That’s what they tell you. Yeah. But it’s actually a data federation problem. They, they actually don’t know how your phone number’s connected to who you’re, who they’re talking to. And, and they don’t have pull up that information on the fly when they’re talking to you. And that’s something everyone feels forget, like driving retention in accurate numbers, which is the scientific part of the problem. But just general, like the experience you have as a customer with a company is can be so frustrating. 

Steven

Yeah. 

Boris

We got an email just the other day, my co-founder forwarded it to the engineering team to, to, because anyway, to get a laugh, but this is a, it was a marketing email, it was kind of a sales slash marketing email. Right. Very much going, using the pretext of us being a user to, to kind of pitch us something. And it says you’ve been a user since December, except it was December like four and a half years ago is when we became a customer. It’s like they just ignored the fact that we’ve been a customer for over four years. Which is almost like you don’t really care and and it could just be a singular data mistake and no big deal and you fix it. Or it could just be they, they’re not even trying and 

Steven

They forgot to add the why why why to that to that. 

Boris

Yeah. Yeah. Yeah. And so it’s like you can make it better across the board on all these things. 

Spencer

And with that in mind, you know, you’re talking about the the call center. They don’t know that, you know, you put in an email support ticket and now you’re calling. So that’s like identity resolution. 

Steven

You have to, you have to describe that email support ticket to them every time you’re on the phone with them. 

Spencer

Yeah. 

Steven

Did I get the reference code for that? Please? 

Boris

Yeah. There’s all those things. Yeah. 

Spencer

I love when I pick, when I pick up and they, they just, it just know. It’s like based on my phone number. 

Boris

That’s I’m saying. It’s magical when I love that. Yeah. Yeah. It makes it so much easier. 

Steven

It’s like, it’s like the Verizon assistant, hello Steven. 

Boris

It happens to so few brands and like we’re, that’s like breathing, I’m not even talking about getting fancier. When you check out at a store, let’s say they have a loyalty card, let’s say they’re actually getting more sophisticated than just the, the things we’ve been talking about. Are they going to bring up your last purchase that you made two months ago in those 10 seconds when you’re at the counter when they have the information technically ’cause they know who you are ’cause of the loyalty card? No. Like there’s so many ways in which we are not getting a delightful experience that we can make better with data, with like data that you can trust in more places. And by the way, I actually think my secret kind of discovery here is, remember what I said at the beginning, I think people should always seek to have maximum leverage, which is a, in other words have the most impact across the most number of people at your organization. And there are a lot of people working in data or product Right. That have tremendous skill at uncovering insights about the user base. And they would love for that to get used on the other side. And a lot of what you work on with retention, there’s basic things everyone should do that I’m sure many people are not, and you just bump them over the head with that. But sometimes it’s like, why don’t you bring the data science team that is eager to help to bear on the problem? Right. And everybody wins. Yep. 

Steven

I mean, have you ever worked with a data scientist before though? I’m just kidding. 

Spencer

All the data scientists listen, it’s like, come on man, I liked you now. Unsubscribe. Unsubscribe. Yeah. 

Boris

But how much of performance marketing have you seen been being run by people who don’t understand mathematics 

Steven

Or don’t understand how to run an A/B test or what a control group is? 

Boris

Well, which just statistics. Yeah. Just statistics. 

Spencer

All the marketers are out there like unsubscribe. 

Boris

Yeah. 

Boris

Podcast is telling me I’m not good. No, 

Spencer

No, it’s fine. But you know, on the, on, on identity resolution like Steven 

Boris

Yeah. 

Spencer

You know, for both of you guys, like what are some of the challenges? I guess this is a two-part. Yeah. What are some of the challenges you’ve seen? Maybe you could, you could bring some either test one or, or an anonymized case study. Yeah. And then something, oh, because we actually have mutual clients. Maybe you could use one of our mutual clients is he could talk about how Census actually solved the problem. 

Steven

I would say our mutual clients are pretty well-solved identity resolution. But I’ll give you a different scenario that I think is a little bit more interesting. Will, there’ll be an anonymized client. Okay. But so there’s always this problem between who you are anonymously on one application and when you identify on another application, even though it’s the same brand, so let’s say web and then mobile app, right? Yeah. And, and there’s a consolidation that happens Yeah. When you are, are kind of identified of both tools, but there’s no persistence in some cases Right. Because of things like the app gets updated and therefore refreshed or you know, your browser cookie expires or any number of things and it requires you to continually log in. So you’re constantly generating new users Yep. For yourself, right? Common problem in the industry. And when you kind of build an identity hierarchy, you’re always coming back to this point of like, okay, well what’s the most common denominator between all these systems share? And you always end up with, it’s an email address that I log in with or my interpretation of an email address that I created through database UID or something like that. 

Boris

Yes, indeed. 

Steven

But when you think about how much waste is created in that process, right? So I as an anonymous user generated a UID on the record and then I basically inherit that UID when I convert. That experience and that UID becomes the parent ID for all of my other login actions and those other UIDs more or less become reference points for that main one. 

Boris

Yeah. 

Steven

That problem gets extended when you have sources of data that don’t even use email address and they have to kind of connect into this, the UID through some other means. Let’s say you’re, you’re, you’re, you’re connecting with an external partner and you can’t share that PII level data. 

Boris

Correct. 

Steven

And so now you have to kind of roll back into a, just an ID record attached to maybe a, a cookie ID or, or some other type of database ID, they’re using it and this problem just compounds, it becomes more and more complex and ultimately you kind of nail down this point where you’re like, I’m just gonna have a lot of rows of the same user and I’m gonna have some type of, you know, join table 

Boris

Yep. 

Steven

Where I unify them up. But ultimately you still don’t have a kind of fully identified user ’cause there’s a bunch of other things that they still haven’t identified on that you’re still waiting for. And this sort of creates this disjointed experience to the person because they think they’re logged in when they’re not. And you don’t prompt them to log in into the very end of the product experience Yep. Because you don’t want to prevent their conversion. Right? 

Boris

Yep. 

Steven

So how do you kind of think strategically speaking right? In the mechanics of this aside, right? Yeah. How do you kind of think about like this, this problem of like, like I need to, like, as a brand, like I need to personalize more in order to personalize more. 

I need to know that hell I’m talking to 

Boris

Yeah. 

Steven

But order to do that, I have to do that in sort of a very, in a, in a, in a privacy concentric fashion where, you know, I can lean in on some of the providers out there that use device ideas kinda join this stuff. But, you know, how, how good is that in the future? You know, like I have to think about, you know, what is, how do I set the expectations with my users that until I know who you are, you can’t personalize this, but you want me to personalize this and you assume I know who you are. Does that make sense? Oh yeah. 

Boris

I, I was having dinner last night and this, you know, this came up a little bit where it’s like, oh, but it’s creepy. And I’m like, no, but it’s not that creepy up to a point. And then that’s really what you want. Yeah. A lot of people recently have like started turning on like, you know, all these iCloud relay type things and they’re going, the ads kind of suck. And I’m like, yeah. Turns out you actually kind of like some amount of personalization, right? You just don’t like two different companies I think sharing information about you, I think that’s the part that people found icky. 

Right. But I think within a brand you absolutely do wanna personalize. So, okay. There’s a lot to unpack I think in what you asked first, you implied right there something I think is super important, which is when people say the word identity resolution, which is a big and fancy concept, let’s split what you said, which is there’s, there’s one aspect which is taking an anonymous user and giving them some kind of telling you that it’s Joe

Steven

Like a, like a golden record type scenario. 

Boris

Well, hold on. No, I think there’s a, there’s a part that’s just de anonymizing, right? 

Steven

Right. 

Boris

Tracking an anonymous user and telling you once that person logged in, that that was Joe. 

Then there is what I think the much more interesting problem personally, like I at a technological level I find more, more interesting is just what you spent time talking about. Which is, well you have five Joes. Whether that’s, ’cause there’s Joe in the billing system, the Joe in the customer support system, and the Joe in the like partner system and they may be identified by any number of things and you need to figure out that the golden record that is Joe, right? That is like, that’s actually the same Joe four times. Let’s merge them, aggregate them, deduplicate them. That I actually think is a harder problem. 

The first one is important for a lot of companies. It’s actually completely unimportant to some. And I think as people start to embrace much more of a direct relationship with a customer, you will find them for like pushing towards registration earlier. Yeah. I think over time. And so that, that anonymous to non-anonymous will, will reduce over time in terms of how much time you spend time thinking about that. And there’s somewhat easy solutions to it, right? Right. You can, you can kind of point at a dot identify, call somewhere and, and, and get it done. I think the merging duplication is gnarly and interesting. 

And again, I’m not sure there’s a perfect endpoint. I don’t know that it’s like, well you apply some magical tool, and Right, you got it. I’m sure Amperity would say otherwise and they would say like, buy this Ferrari and like there’s a perfect, we will give you a perfect Joe at the end. Right. 

Steven

But still probabilistic at the end of the day. 

Boris

Of course at at the limit. Even at Facebook it’s probabilistic. Imagine I’ve met someone who owned this at Facebook and it’s like, even there it’s probabilistic. But I think you can make this an iterative process and it is easier now to do this with like tools like a warehouse is what I’m trying to say. 

Because joining is is and trying ways to compute connections between people is more doable. I actually think, I also think there’s other ways to personalize above, like you said, something very true, which is eventually you have to create this kind of identity that is a bag of identities. Right? 

Steven

Right. 

Boris

There’s two things I think that are, that go beyond that. One is there are sometimes bags that are not just a singular person. I’ll give you a non-anonymous example. I’ll talk about a real company Sonos. Okay. Sonos. Super interesting. They have multiple ways you buy products, right? You can buy them online, you can buy them at Costco, right? All that stuff. The way they map the resulting speaker into you is actually not one-to-one with you. It’s actually you register it in a household. You probably don’t even think about this if you use Sonos, but it’s, it’s actually the data model is the speaker belongs to a household and you belong to that household. 

Steven

I see. 

Boris

And so that’s actually really important for Sonos as when they, how they engage with customers. 

Steven

I guess multiple people could connect to that same speaker. 

Boris

Correct. Same household. Right? Correct. 

And all of this is important when you’re, you know, doing Google ads for the new speaker and someone’s like, I bought it at Sonos and I registered it. Why are you still advertising the thing to me? Right? It’s like, oh yeah. ’cause the pixel didn’t flow from that. Right. And so I actually think just modeling things in a way that makes sense to your business can go a long way. Now you asked about how do you do personalization with a world where privacy becomes more and more of a concern. If you think about who, who you are, like what you are, there’s like a nominal identity to Spencer and Steven, which is your name and, and your lineage or whatever. 

But there is a, what I would call to get really, really neutered. You are also like a, you have a structural identity, you have a, I am this tall, I am like my sports team is this team, right? You are basically a bag of attributes, right? Not just a name. And so I think one way we’ll get through the privacy kind of barrier is that there’s a lot you can do, whether that’s sharing across companies or within your company using tools where you’re like uncomfortable. Like you might buy an AI tool that’s gonna do stuff. You don’t want to literally say, here’s my user Steven, but you’re okay giving it a bag of attributes and engaging with that user as a bag of attributes. 

I think that’s how we’re gonna get through how do I maintain privacy but engage with the person in all the ways that matter? 

Steven

So it’s almost like a, like we typically call this concept like a micro-segmentation where there’s shared behaviors or shared attributes amongst these users. And so the messaging that would resonate would be equal or similar. 

Boris

Yeah. But even in the limit, in the very, very limit, does it matter that your name is Steven? Like for almost everything I will do for you, it doesn’t matter. Right? Right. Almost. So even within a company, we have some customers that are extremely privacy-conscious. They’re in other countries usually. And yeah, there’s a lot they do to mark things as like PII and so, so even though they have maximum flexibility in the data platform, the various users of our system get various levels of visibility Right. Into what’s in the column. Right. But they’re still able to do all the, all the campaigns they want to do. 

And so I think treating humans as bags of properties, bags of attributes may actually be better than saying, oh it’s, it’s Anna, it’s Joe. Yeah. 

Spencer

You know, there’s nothing that makes me feel warmer inside than being considered a bag of attributes. 

Boris

I mean, you’re a bag of mostly water, so 

Steven

That’s fair. 

Boris

Your primary attribute is water. 

Spencer

That’s true. 

Boris

I think you’re like 90% water, right? 

Spencer

A little bit of carbon. 

Boris

Yeah. 

Spencer

Yeah, yeah. Just a bit. 

Steven

So shake your case. Mostly beard. 

Spencer

Mostly beard and beers and coffee. That’s true. 

Boris

I mean, what else do you need? 

Spencer

A little bit of kombucha. 

Steven

Oh yeah. Gotta get the bouch. 

Spencer

Gotta even it out somehow. So shifting gears for a sec. There’s like a, there’s been a debate going on on LinkedIn about, you know, after Twilio’s lead got ousted. Yeah. 

Steven

Jeff Lawson. 

Boris

Yeah. Jeff Lawson. Very long run. Yeah. 

Steven

Yeah. Also just one of the coolest guys you’ll ever meet. Absolutely. Karaoke king. 

Boris

And, and like by the way, fun fact, like, you know, it’s like truly was always about developers, right? And so I feel like that’s, that says something. 

Spencer

Yeah. Yeah. Segments has always been a selling to developers 

Boris

Twilio long before Segment. Yeah. But yes, I think maybe 

Spencer

That was what made it a good fit then. Yeah. 

Boris

Yeah. I think segment segment’s superpower in terms of go-to market was that it as an engineer, it was like, yeah, I’m just gonna go use that analytics, Jess. Right. 

Spencer

So with, with that in mind though, there’s a lot of people are checking under the, the turtle shell, the rock and there, and you know, you have, you have people looking at the CDP as a rock and saying, oh, where’s the money under here? I need to shake the money out. So like there, there’s a lot of questioning of the ROI and how does A CDP make you money in the first place. And what are your thoughts on that? 

Steven

Just to piggyback off of that, I know you consider yourself a CDP, but do you really consider yourself A CDP? I don’t. 

Boris

Don’t really consider myself A CDP

Steven

Okay. 

Spencer

But let’s work into that. Yes. Sorry, let’s go back to the other one And then we’ll work into, 

Boris

I Think there’s a technological kind of set of conversations people are having about this, whether it’s on LinkedIn or elsewhere. And then there’s the business kind of questions. I’m 

Spencer

Thinking more of the business questions ’cause I’m, you know, part Steven and I are co-CEOs of one of my jobs is to make sure that we’re meeting our margin goals for the year and we’re not spending so much. And so I’m looking at stuff and I’m like, do I really need this tool? Do I really need, yeah, I 

Boris

Understand the $19 a month is like a real Yeah. 

Spencer

The $19 a month. I’m like, like, eh, you really got it. You know, you might’ve sold me there. ’cause you know, maybe, but 

Steven

I made Boris think about it. You can spend the money and then have to think about was it worth me spending it? Or you can think, is it worth me to spend it right now and then not spend it? And you know what? Yeah. 

Spencer

So that, those are the things that keep me up at night. But, but in terms of the CDP, like we know, like Steven and I know, but how would you to the, to the, the CEOs, the CFOs of the world who don’t know who, who aren’t experts in this space. 

Boris

I would go back to what we said like a little bit ago, which is there are very easy ways to measure the performance of a campaign, right? Like you, you, you talked about you work with, you work on retention as a problem, right? It’s like you can almost certainly say, we will help you do A, B, C, we will charge you X, and it will lift retention 2%, which has a tangible dollar value. Right? Right. So, so you’re not, I don’t think you’re saying you can’t assign a dollar value to, to various, let’s call it growth and marketing activities or sales activities. I think it’s fair to say that like one of the things, for example, we help companies do, like I told you earlier, is like improve the SLA on a customer support call. 

That might be harder to then assign a dollar value to that. But even there it might be a cost center optimization, right? It’s like, well we can’t deliver this SLA to everybody with prioritization. We can deliver the SLA and reduce, you know, kind of overall staffing. What I would say is attractive about our approach, which again, I genuinely don’t think of us as a CDP. We, we’re not a all in one box that does, you know, identity resolution and all these things we, we connect to your warehouse that you already have, that you’ve already bought from someone else and make it available to everyone else in the company. 

And, and in a way it, it’s plausible that we’re structurally going to be better able to deliver ROI because two for one, right? Everything the data team is doing for a completely different purpose, which the CDP used to not partake in, they will now be able to serve both purposes. I can only speak to the companies I’ve seen, but the, the need to do good good audiences when it comes to advertising is not going away. 

The need to figure out how to drive retention and growth like upsell both in individuals and in accounts when you’re dealing with a B2B context doesn’t seem like it’s gonna go away. Yep. So it’s, I think you should definitely not buy a very expensive all in one solution and have a data team. You should probably pick one of the two, right? So maybe you should have a data team and lighter solutions like what we sell or avoid the entire ENG team data and ENG team and f figure out if there’s a all-in-one you can buy that will do everything for you. But yeah, ultimately I think all of these teams, whether it’s a data science team that’s demonstrating its value or a all-in-one CDP demonstrating its value, you better drive lift or retention or, or something. 

Yeah. I think if none of that is true, if none of those things drive ROI then yeah. Not only is the CDPA bad business, but, but, but a maybe a good chunk of data science is a bad business, which seems unlikely to me right now. Right. Your 

Spencer

Way around this is you’re unlocking it for more teams beyond just marketing or one use case. 

Boris

Exactly. Yeah, yeah. Like remember the name of the company, right? Like my weird hope is that companies can actually have a census, like in the literal sense. 

Spencer

Yeah. 

Boris

And I do think if you leverage that work across multiple orgs rather than just marketing or just advertising, to even go back to the world of, you know, DMPs and all that crap. 

You at the very least have the same cost on our side. For us. The cogs are the same, but you get hopefully 2, 3, 4, 5 x the output, right? Yeah. And so then maybe the, the the, the margins start to make sense. I used to joke, but it’s true. Like the number one reason census is not a CDP is, well there’s, we’re not a database. We, we don’t have storage Census has no basically no storage, which is completely different than a CDP. Right? And that means it’s also a cost we don’t have and we don’t pass on to the customer. 

So maybe we’ll solve ROI on both ends. You 

Steven

Know, you don’t consider yourself A CDP, but then others would label you as a, well, if you’re not a CDP, then you’re a composable CDP. 

Boris

Sure. 

Steven

I think there’s benefits to composability for sure. I mean, being able to customize your own solution in any case is not such a bad outcome. It’s a packaging conversation, so to speak. What do you kind of think of that as, you know, as we, as we kind of talk about, or we talked about earlier that, you know, these like Klaviyo is adding on more of like cloud type software. Braze will, you know, it’s, it’s seemingly gonna be a CDP next week. 

You know, what is the, you know, when we talk about kinda like composability, is it just the optionality to pick the best pieces of that tool that you’re looking for, but then give people the optionality to not necessarily have to buy at all and get to that point where they’re at their customer service or Salesforce service Cloud Cloud, that’s it. Yeah. And then they’re like, well I need to move to Intercom because it’s not solving my problem. Like, does composability actually help with that problem or is it just more of a it’s just a markety pitch? 

Boris

So I’m a software engineer first and foremost. And composability has a, a real meaning, right? In software without composability and software, which goes all the way down to your programming language capabilities, right? You can’t build scalable software. It’s actually not possible because it’s an issue of, you know, if everything has to know about everything, if everything has to be deeply coupled, you just need extremely large teams and, and it’s very hard to scale components. So to me, composability first, like is, is valuable as a design precept, forget the business concept of composability. 

Just, just designing your software in such a way that it can be plugged in and into one side or another. And if every piece of SaaS decides, for example, to not let its data be exported, you know, like imagine the world like companies like Fivetran couldn’t operate, right? Which think about it, if you’re Braze, let’s say in that example, yeah, your desire if you’re gonna be a CDP is eventually you should be like, well I should also keep this data to myself, right? Keep you locked into this. 

So that would be the natural pressure. But I think the consumer is always gonna create the reverse pressure, right? 

Steven

Right. 

Boris

Which is to say just as like government forces certain larger companies into this, like I think the, the market forces medium-sized companies into this, which is you must allow your data in and out. And so that to me is the first thing. Like composability means you are interoperable with other data and other software. And I think that’s just good period. Right? There are some set of businesses for which one size fits all will work. I think it absolutely will be true. They should evaluate those kinds of solutions for the majority of businesses. 

You have something about your customer, your business view that is bespoke to you. A and B should span multiple channels. Like it goes back to what we said at the beginning. Like, if everything happens only in one org, one department and one channel, then composability might not matter to you at the business level, right? Right. I still think it’s a good software practice, but Right. But it might not matter to you. And, and, and I would not push you, but I have yet to see any company of trying to get to a certain scale, stay in that kind of world. I do think Klaviyo, I could see where it comes from, right? Because the Shopify world is tons of tiny businesses, right? 

And so yeah, there might be a world where you’re like, I actually don’t blame it. Like, you know, Klaviyo do a little bit more than the email, do a little bit of advertising and like now I’m a CDP and maybe that’s good enough for small scale type business, right? Yeah. And that’s fine. 

Spencer

We’re running up against the clock here, gentlemen, with the two of you mind giving us like a 30 to 60-second kind of wrap-up of everything we talk about. Maybe just like final insight and I’m actually going to, I’m going to here, 

Boris

It’s quite the ending. Just, just be insightful. Just, just, just 

Spencer

Yeah, if you could just be inspiring as, as Well. Okay. That would be great. All right. Here you go guys. 

Spencer

Steven, you’re up first go. 

Steven

Oh, right. So I think, you know, Boris, some of the things we talked about today that were really insightful was about just, you know, how to think about your sort of integration of your different tools from a, an operability across the vast majority of your organization. So kind of democratizing, so to speak, the access of that. And then thinking about some of the kind of current state of the market as we record this around CDPs or data warehouses. That there certainly is a lot that you can leverage there, but it’s, it’s only as good as the basically the, the effective business case that you’re solving outta the back of it. 

So what would you like to add to that? 

Boris

This is really good. Okay. I think what we said at the beginning, which bears repeating is there is an ever-increasing number of tools, applications, channels, et cetera. I think it’s not only going to increase, it might even dramatically increase because now everyone using AI is going to make an app as well, whether it’s internal or external, right? So we may actually see an explosion of what you and I would call an app, even though it might be much smaller and just built with ChatGPT. If that’s true, what matters to delivering a good experience for the customer is, or for your business anyway, is having trust in your data, in my opinion, like that, that that is a, a cornerstone of delivering a good experience. 

Trust is not an endpoint, it’s a process. It’s like you gotta create a, a virtuous cycle in which you have a feedback loop that helps you find the problems in your data. And I think doing that in one department across like in a singular tool is unlikely to succeed. That’s my take. And so building some form of collaboration across organizations with tools, composability and, and the right kind of infrastructure is a recipe for getting data you can trust into the hands of everyone in the business. And I think this only becomes more essential. We’re gonna have three pressures in the coming years. One is more and more automation, both ’cause it’s easier AI and also because you need to for like driving better workflows at scale two, we’re gonna have ever increase in kind of privacy expectations and finally, ah, hit the number. 

And finally, to use, you know, your favorite thing is like margins matter, like ROI matters. And so you’re gonna need to be scientific about how you spend your money, how you drive workflows, and do they work. And all of that means having a really good feedback loop between the actions that you take and the data that you have. 

Steven

Well said. 

Spencer

Yeah. I think we have a census here. Yeah. ao. Alright, well listeners and creepy watchers at home, please subscribe. 

Steven

And we, the data scientists don’t like us. The markers don’t like us and now the YouTube viewers don’t like either 

Spencer

Pretty much. We’ve got nobody now, but yeah, so Boris saying, but 

Boris

The blockchain people will just follow 

Spencer

You know what the blockchain bros. It’s okay. I like we we don’t need them. It’s okay. Sorry guys. We love you. But also you’re annoying. Boris, thank you for coming. 

Boris

It was a pleasure. 

Spencer

I know it’s a busy week with NRF and everything. Or actually, NRF was months ago, so I don’t, at the current time, not much is happening other than the things that you’re working on at the current time. Exactly. Everybody check out Census if you haven’t already. Here’s Boris’s personal cell phone number so you can tell. I’m just kidding. 

Boris

By the way, that’s one of those things that once it gets out, like my cell phone got out on something at some point and it’s like, oh my god. 

Steven

Well that’s, that’s part of his bag of attributes. Yeah. 

Spencer

Yeah. You’re just a bag of Attributes by way. 

Boris

I think that’s something that people will wanna protect a lot more and then we try right, the laws actually try to prevent tech spam. Yeah. And so it’s funny how like how much we focus on email and, and your Yeah, your phone number is now the wait till they get your Instagram and ah oh 

Steven

Be. It’s all over. 

Spencer

Steven doesn’t have Instagram so that’s not gonna be a problem. Alright, everybody like us subscribe us, watch us do whatever you do, TikTok things and I don’t know, wherever we are, do it. 

Spencer

Thanks. Adios.