EP 231 - Spenser Skates, Co-founder and CEO at Amplitude - So Much More Powerful if You Have the Data

Hi, this is Michael Waitz and welcome back to the Asia Tech Podcast. Today we are joined by Spenser Skates, a co-founder and the CEO at Amplitude. Spenser, thank you so much for coming on the show, particularly on this day. How are you doing today, by the way?

I'm doing, I'm doing fantastic. Thanks for having me. I think really huge fan of your work and the stories you get the entrepreneurs to tell. So I'm really excited to jump on and tell the story of myself as well as Amplitude.

That's really kind. Before we get into the central part of this conversation, let's just jump into some of your background.

Yeah, absolutely. So just to give folks who don't know me, so my name is Spenser Skates. I'm CEO and co-founder of Amplitude. Started the business back in 2012, right after a previous company worked on, and we had seen a huge opportunity to build the future of infrastructure for the internet, namely product data. And the web was undergoing a transition from what I thought of as a marketing-centric web to a product-centric web. And I was interested in building a company around it. Uh, we launched the company in 2014. We've since grown it. We took the company public last year. Now we do over $200 million a year in revenue and we do product analytics. For those of you who are not familiar with us, product analytics is a set of infrastructure that helps you run and build your product using data and insights from your users. And, you know, today we have over 700 people in offices around the world, including here in the US and San Francisco where I'm based, but also in Europe, Asia, and our main headquarters there is in Singapore.

Your headquarters is in Singapore. Super interesting. Can we back up for a second though?

Yeah.

Can you make a differentiation actually between and why you you thought this, the difference between sort of marketing-centric and product-centric. Like, why did you care about this? Because, look, you, you didn't mention this, but you went to MIT. You didn't study marketing, you didn't study product, you didn't do— you weren't involved in any of that stuff. Why did you care?

Yeah, well, we didn't. I mean, coming out, I graduated from school in 2010, and I didn't know anything about anything in building companies. And we just— we actually, the first thing that we did before Amplitude was this company named Sonalight, which was a voice recognition product that helped you send and receive text messages from your phone. And through doing that product, we got exposed to the full set of tools and infrastructure that was out there to help you build websites and products and things like that. And one of the things that was really clear to us as part of that process now, Sonalight was just okay, you know, didn't do that well. We had, I think, something like a few hundred thousand downloads. So, you know, not, not crazy explosive growth, but, you know, someone was using it. That was good. And the technology was just a little too early. But one of the things that we had done while doing it was we really cared about data on our users. We wanted to know what they were doing, how they were using the product, what they were succeeding at, what they weren't. And we surveyed so many of the tools out there at the time. There was Google Analytics, that was kind of the standard that you started out on. There was this company named Flurry at the time, there was Mixpanel, there's Kissmetrics, there's Localytics, there was App Annie, there, there was, there were so many, I remember just so many different analytics products. One of the interesting things that really stuck out to us at the time was that a lot of these products had been built for the previous generation of the internet. Yeah. And what I mean by that was that they were very focused on this idea of a page view and which of your pages had the most views and where that traffic was coming from. And one of the things we realized is these are all very marketing-centric ways to look at your data.

Do you think this page view thing— and I always thought it was funny, right? Larry Page wrote the original, um, or some of the original algorithms for Google and called it page view, right? But do you think this is why that Google—

oh, that's so But Google is such—

it's true though, right? So Google is such a large part of what happens on the internet, particularly when it comes to analytics and, and search, obviously, right? But do you think that's why there was such a big focus on page views? And if it's not pages that people should care about, what is it?

Yeah, so I, I don't know about the connection to his name, but that comes from the idea that you have a webpage and that you want to track views on that, on that page. So absolutely, the fact that Google is, you know, the dominant company in terms of internet traffic, and that's where everyone will get their traffic from means that a lot of the infrastructure they've built to help you manage and think about that comes through that lens. And they're ultimately a marketing company at the heart, right? They sell advertisements to help you generate more traffic on your, on your pages. Now, the problem with that way of looking in the world is the internet had undergone a really big set of really big change when we started, which was it was going from this idea that you just, it would be a marketing resource for your company to the actual product itself. So let me, let me explain what I mean by that. First, you know, You look at, for example, B2B products. It used to be most B2B businesses would have a web page online where they'd capture leads and then you talk to them and then sell your product that way. Now it's actually the product delivery is happening online. This is the whole rise of software as a service. And the idea is you're going to just be using the web, the website as the actual product itself. And so when you're using the website as the actual product, that's a whole different way of using and tracking and thinking about your users. That's very distinct from just capturing leads. You actually want to know what features they're using. You want to know what parts of they're getting stuck in. You want to know what keeps them coming back. You want to know all sorts of different patterns on your user base that you don't really care about if you just have marketing as you just have your website as a marketing channel.

So this is, this is really subtle. Like it seems like it's pretty obvious in black and white. I can't remember the year, but one year Apple just changed its entire website and it went from being a marketing site to being the store itself. In other words, they no longer just said, hey, we're releasing the new, you know, MacBook or whatever it was. They were like, here it is and you can buy it right here. Now, that's not exactly what you're saying because their products aren't digital, But I feel like that was a tipping point for me. We're transformed from here's all the information, go to the store, go somewhere else and buy it, as opposed to here's the product, how are you using it kind of thing? Yeah. Cause that's what you're doing.

Totally. And that Apple's a great example of it where now a lot of way you're interacting with the brand, you're not going to the physical retail stores to engage with them. You're buying the products. You know, I just bought a whole set of new chargers for myself online. You know, I wasn't going to the store to do that. And that's the case with so many companies, right? So you look at retail, retail companies. Walmart is now thinking of how do I sell more products through online? They actually had this thing in one of their earnings releases where they said that shoppers who buy both in-store and online spend 3 times as more with them as those who just buy in-store. And so it's a huge deal that they want to unlock this channel. Same with media companies. You look at Disney or NBC or, you know, any of— or HBO, any of these companies, they want to— they're not delivering through the old school distribution channels. They're actually delivering through the web now, their product. And so this is happening across every single industry out there. There where you're delivering your value and your product through the internet. And the challenge that we saw was that the infrastructure that had been built to help you track it was still very, very marketing-centric, and it was not focused on helping you track that product journey. And so that's— that was the genesis of Amplitude. And we ended up building an internal tool with Sonalight that when we showed it to a lot of other companies, they're like, wow, this is amazing. I really want to get these same sort of insights. We actually found out that one of the questions we had was how important is the accuracy of the voice recognition to whether you come back in the product. As it turns out, as you might imagine, it's incredibly important. If you have a first— if you were successful in your first voice recognition attempt, you were twice as likely to be a long-term user than if not. And that was a huge, huge deal for us because that told us what part of the product to focus on. And that's what kind of everyone else in digital wanted to know.

Right. So I'm in the digital product business. And this is why when we were prepping for this, I asked you, what kind of microphone do you have? Because it's the same thing. You're smiling, right? But it's the same thing. If in the first 3 or— I think it's in like the first 45 seconds, Your sound is like really bad. It doesn't matter what I sound like.

Someone's, someone's going to drop out of the podcast. You don't want to hear me anymore.

No, they don't want to. They don't want to hear you at all. But is this one of these situations where almost like Slack, which if again, I'm just trying to remember here, it was a game company and they started building this tool internally to be able to communicate and share stuff because everything else out there was terrible, at least at the time where you were building this for Sonalight. And then you realize, wait a second, this is the better product.

Yeah, that's, that's exactly what happens. The exact same situation.

I love it.

As Slack where we ended up We spent so much time internally building this tool because we knew what it had to look like and we knew what sort of insights we wanted out of it. And it was just, it was crazy because there were only two of us at the time to spend, spend so much time just doing, trying to do data and analytics. But it was so important for us to really understand how people were using the product and where they were getting stuck. And the problem that all the other companies we talked to had was they didn't have a team capable of doing that. I mean, my co-founder, we were, I was, I'm an MIT guy and my co-founder, he was the top algorithms engineer in our graduating class at MIT. And So we had the chops to build infrastructure like this. And in fact, analytics is a much better problem than voice recognition because it has like a defined, you know, that's where you can use algorithms and data structures and distributed systems to do a much better job than, you know, and that's where kind of our skillset, it matched really well to it as well. And so, you know, that's, we set off on that path in 2012 and haven't looked back since.

Why did you start with voice recognition into texting, right? Because this, it looks to me, if I understand your history properly, like something, you had a job after you graduated from MIT, you stopped doing that. And then you said, hey, let's build a company that does voice recognition for texting. Was this just a thing where like it impacted your life and it was just like, Jesus, I wish I didn't have to text this. I wish I could just say it. And then you realized, oh, that's not going to work.

Yeah. The reason we built it is because we thought the technology was at the edge of what was possible with technology. This was actually before Siri. We had developed it and we said, hey, a lot of these voice recognition may take off as a new interface and we want to be at the forefront of that. So let's try it. Let's try this out. And in fact, you know, one of my learnings from it was how much I hated voice recognition as an interface. It's just, it's bad. Technology just wasn't quite there yet. It wasn't accurate enough for you to control a computer with. You know, if you look at what has been used to control computers, keyboards was the very first, and then you had the addition of the mouse, and then now you have touchscreens. And those, they have a bunch of attributes about them, but highly accurate inputs, you know, is one, and instant responsiveness. So the second you touch something, you get a reaction back from the interface. And voice isn't the same way. You don't have the same level of, you know, first you don't have the same level of accuracy. And then second, somehow humans have developed voice or sound and audio as the way to have high bandwidth conversation. And so our brains have evolved for that. But in terms of interfacing with computers, it's super clunky because you have to say your statement and you have to wait a bunch of time for the computer to respond to see if it even understands you and you're kind of not really sure.

So here's one of the things you don't know. In 1998, probably 1999, I was part of a tech team that was working with the trading desk at Morgan Stanley in Tokyo. And we were literally digitally transforming that entire trading desk. So we would go into the— you'll love this— we took the wire that connected to the printer from an exchange for the futures, and you'll see where this is going in a second. We split the wire and took all the data off that wire and then could get real-time price information. Yeah, from the, from the exchange. Nobody else was doing this back then, at least in Japan. The other thing we were testing was, hey, instead of making the trader type in what the order is, just have them say buy 30 Nikkei, pay 50. It just didn't work. And we just tried this.

Yeah, trying to— oh my, that, that's trying to do a voice interface for trading is like the worst ever because then it's like you can't— you can't— there you cannot make any mistakes. Accuracy is super important. And so if you're never— if there isn't going to be a question in your head about it, then no way, forget it.

No way.

That's so funny. That's so funny that you tried that.

Oh my God. And but so to me, this, this experience was almost like— do you eat sushi, by the way?

I do.

So have you ever had uni, sea urchin?

Uh, my wife is very into it, but I can't. It's, it's not— okay, that's a little too far outside my comfort zone.

Yeah, but I want to go through this with you because it's the same experience that I had the first time I had uni. It was terrible. It took 10 years to go back and have good uni and then realize, oh my God, it's like a gift from nature. It's so good if it's good, but it took 10 years to go back to it. And for me, voice recognition was the same thing. I'm like, this is terrible. And I've never gone back to it anyway. Sorry, I interrupted.

Yeah, it's not clear. Well, we'll see. You know, one of the other lessons is it was hard to build something successful as a 2 or 3 person team. You really have to be a scaled team to improve the accuracy of the algorithms and fit it for use case. And that's why you see, you know, it's still not clear. Like maybe Siri will be successful. Maybe it won't. Maybe Alexa will be successful. Maybe it won't. Won't, you know, we don't know yet if those are still going to be around 10 years from now. It's going to look very different.

I think so. Do you remember when you and your co-founder at Sonalight were building this analysis, probably tool at the beginning, but sort of turns into a platform, and you looked at each other and realized, hey man, isn't this the right product? You know what I mean?

Yeah, we, um, when we decided to shut down Sonalight, we actually explored a number of different ideas, but we kept coming back to what we had learned in product analytics, and we were so clear in our head that that was the future. One of our hesitations was that there weren't that many successes in the space, Seemed like there was tons of competition and there wasn't really anyone had broken out and it wasn't going to be, you know, we weren't clear. It's like, okay, are we going to be able to build something that really breaks out and gets to the next level? You know, the one big success was Omniture that had sold to Adobe in 2009 for $1.8 billion. Yeah. But then the next generation of tools, you know, there's kind of like a highly competitive commoditized market. Now, since then, we've been able to do very well and break out in a number of respects. And, you know, it's really clear that the market and the need for this is huge. You know, you talk about wanting to run your digital business online and track what's going on in it, you know, people are going to spend infinite amount of dollars on that. But at the time, you know, very much wasn't clear. The funny thing was, I remember a lot of venture capitalists were talking about big data and they were talking about the rise of mobile. And it was so frustrating because I was like, okay, if you take those two things together, what do you get? Well, you get application analytics, so you get analytics for mobile. And that's what we, you know, before we came on the product analytics, one of the versions of Amplitude, we talked about it being for mobile apps. And I remember trying to pitch all these VCs on it and none of them really really understood it. And it was so frustrating to me. But now, you know, you kind of look back and you're like, wow, you know, it's a lot of ways it's obvious that there's going to be a third-party vertically integrated tool that provides all of these insights into what's going on in your product for you.

I sometimes feel like, God, I hesitate to say this, but I sometimes feel like venture capitalists are like Henry Ford's customers. You know, they just want faster horses. And until they see something like what you're building, they don't know it's going to work until it works. And then everybody wants to pile into Series A and Series B. And again, it's not a— it's not— it's not to be sort of detrimental about all investors, it's just interesting to me that there's a big herd mentality around, well, this hasn't succeeded before, we haven't seen it. So it's like, as a founder myself, I've stopped trying to pitch ideas to people that don't get it, but that's probably a story for another time. So what does growth look like to you? Yeah, you're, you're smirking a little bit. In other words, big business, $200 million of revenue in the United States. There's a whole world out there. You have offices in Singapore, you have offices in Europe. Where do you go and what, what has to change? Do you know what I mean? Because you can't just cut and paste from one market into another market and get the same insights, right? No.

You have to be very specific. So I think, let me say a few different things. One is that Asia is very clearly the market of the future for software. No question. Japan is the single biggest market for software outside of the United States, next to the United States. And then you have a lot of places that are growing really fast. I mean, we have, we do a lot of business in Korea today. For example, Korea is actually our biggest business in all of Asia. And so, and then, you know, you obviously have Southeast Asia, India, you know, Australia, New Zealand. And so, so huge, huge future potential in terms of the software industry. Now it's really early for us in all of those. We've managed to work with a number of great companies. But, you know, one thing we're really excited about is going into Japan next year. And one of our realizations is you have to be very, very deliberate about how you enter these markets and that the way they do business is very, very different depending on, you know, Japan is very different from Korea, is different from Southeast Asia, is different from India, is different from Australia and New Zealand. And your approach has to be to partner with great local talent to bring your business there. Now, the core value proposition in the software is the same in that no matter what, you're still, you You still have the same problems. You still really want to understand how your users are using your product, want to get infrastructure that helps you do that. You want to run all these applications on top of it. So the need for amplitude is that fundamental need is the same, but the way you build and scale a business is quite different. And particularly now, you know, now that where that a lot of the economy more broadly is looks like it may enter into recession, you have to be very, very thoughtful about your approach in entering new markets.

Do you think that this digital, this digital transformation, I keep using the terminology, right? I want to get back to Walmart just for a second and then come back to Asia, just because you said that they're now also figuring this out, right? But they're also physical products too. Do you see that some of these big retailers will then turn into retailers of digital products as well? Because they have all these touch points, right, where they can sell digital products, not just physical, physical products. Is there a risk to some companies, like I would even say Facebook, for becoming kind of the IBM of the internet, where they had this incredible power at some point, but they kind of took their eye off the ball a little bit. And then when other companies were innovating and changing and iterating, they kind of get lost in the sauce, if that makes sense.

[Speaker:SPENCER SKATES] Yeah. I think this next decade, from my standpoint, is going to be the key moment for who owns the commerce channels of the internet of the future. And so to your point about Walmart, Walmart, very, very aggressive about selling what they do online. And so you could easily imagine them bundling a whole bunch of other sort of products. And so this is what every company's trying to do. You have media companies trying to bundle and rebundle products as part of its suite, like Amazon, you know, with Amazon Prime including all sorts of crazy stuff as part of its offering. Um, you know, you— and then you, you have Facebook trying out all sorts of different things as well. And so it definitely is a time of fast change. And so whoever plays it right is going to own a lot of, you know, there's a lot of money to be made where so much of the economy is shifting from an offline economy to an online economy. Whoever sets up right within that will end up doing very well.

What, um, can we get a little bit geeky if you don't mind? Yeah, you've built this, you've built this really kind of high-powered platform. What kind of data is getting analyzed and what kind of insights come out of it?

Yeah, so we help people understand customer journey so that they can make better product decisions. Let me give you a really simple example. So Peloton's been a customer of ours for a number of years.

Interesting.

And you're familiar with the Peloton bikes. One of the things they wanted to know was how do you keep users coming back to work out? Because, you know, if you're anything like me, you know, you know, working out is one of these these things you should do, but are you really going to go do it all the time? You know, maybe, maybe not. It's hard to, you know, it's hard to— you need that little bit of extra motivation. And so they looked across for the users that kept coming back. What do they have in common? What learnings could they take from those users and apply it to the rest of their customer base? Actually, before I say what it was, I'm gonna— Michael, I'm gonna have you guess. What do you think, uh, the most common thing that users kept coming back— like, what did the users who kept coming back to Peloton's workout class have in common?

So not what— not what they were doing, but what did they have in common?

Yeah, what What, what, what were they doing in the product?

What were they doing in the product that they were probably watching something or listening to something like they were probably using it as a way to put on some headphones and get away from whatever else was going on in the environment where they were riding that bike. Yeah.

Close. That's not quite what it was. What they actually figured out was that the most important thing was a social interaction. If you had a social interaction while doing the workout. So if an instructor called you out, if you did a workout with a friend, if you had some sort of virtual interaction with someone else while doing that workout, you were so much more likely to come back to the product. And that was a huge insight for them because what they then did was they made it much easier to have social interactions in the app. They had, uh, while you're doing a workout, they added all sorts of things. They did things like high five, so you could high five other riders. Uh, they did things like leaderboards so you could track the progress and how well you're doing on your workout versus your friends. They added all these little social features to make it much more engaging. And as a As a result, the, uh, the rate of people coming back to do workouts skyrocketed, and it was a huge, huge deal for them. And that's why they have one of the most addicting fitness products out there, is because they've really cracked this with their users. They figured out how to make it a social experience, and that wasn't something they had figured out before they had, uh, they had Amplitude data.

This is super interesting, and I want to figure out how they figured that out in a second, but I need a little bit more information. I have never seen a Peloton, and I don't even know anybody who owns one. I don't know what it's okay. No, I know what it is. And I know what it does. And that's great. But I just don't know anybody who's had an experience with it. So I'm going to ask some questions that may seem a little bit stupid. But that's just because I'm ignorant. Not, not, not Tom, right? You know, I did this thing. I wish I could remember what it was called. But I went through this phase where I did that bicycle riding thing. You know what I mean? Where you're in a room with like 30 or 40 other people, they've got a leaderboard up on, they've got an instructor out there just like urging you on. I can't—

SoulCycle.

Yeah, that's something like there was a competitor to SoulCycle, but something like that. Can you do that on Peloton? So I'm sitting in my living room or sitting in my, you know, my garage with my Peloton, it's connected to the internet. And then is there an instructor that then has classes for 30 people that are in different places, but again, a leaderboard and urging it on? And is the screen on the bike itself, or is it on your mobile phone or on an iPad or something like that? Like, how does this work actually?

Yeah, so the screen is on the bike, and what they'll do is they'll schedule classes at certain times with instructors who are live, and then you'll be able to see yourself in the class alongside lots of other people. And then the instructors will give people in the class shoutouts, so you will— you could actually hear your name get called out. And one of the things that they they saw was that if you had one of those interactions where it wasn't— didn't feel like you're just watching a pre-recorded video, where it's— there's actually some sort of social element, uh, to the workout, that that was a huge predictor of whether you came back and did a second workout. And the kind of macro learning for them as a company was that they weren't just selling a bike experience to help you get fit, an exercise bike, uh, they were actually selling you going to bike ride with your buddies. Are you bike riding with some friends? Are you bike riding in a fun way? And that little bit of social energy was one of the biggest things to get you over the hump of actually deciding to do a workout.

So how does Amplitude know that? In other words, you're somehow connected through software, through the platform, getting this data from the bicycles, right? But how do you know that that leads into people want to be more social, or the people that are coming back know that their buddy Lisa is jumping on the bike as well? And like, how do you know all that stuff?

So what they— what the Peloton team did is they track a bunch of the interactions people have with the bike. So when do you start a workout? When's your next second workout? And then what are you doing during the workout? And what they looked at was what are the things that are most correlated with a second and third workout? What are the things that you would have to do in your first workout that are most correlated? And one of the things that stood out is if you had a social interaction during that workout. So if you did a high five or you called out by instructor, or you were working out with a friend, you were much, much more likely to come back for those subsequent workouts. And so, and they looked at a lot of different things. They wanted to know, okay, well, maybe certain instructors are better than others, but there's certain types of workouts better, maybe your demographic, you know, where your base location-wise is better. But it was actually that social behavior that they figured out was key.

This could not be a surprise. I mean, we knew this from the mall. I mean, the only reason why people went to the mall wasn't just to go to, you know, to Foot Locker, it was to go there with their girlfriends and their guy friends to just kind of walk around and chat when their parents weren't around. Like, the social aspect of shopping was more important than the shopping aspect. No, this should not be totally—

it, it makes so much sense to us now when you think about it. You think about, man, I really hate working out, but I really hate exercise, surprising, but oh wow, when I do it with a group of people or with friends, I'm much more likely to go. You know that intuitively as humans, but it's so hard. There's— it's so confusing when you're building a product and there's so many different signals coming in that how can you pick out what the thing is that's most important? I want to give you one more example really quickly. It's something that may sound obvious, but it's not obvious to you. Look at the data. So Facebook had this exact same problem where they had hit a wall in terms of user growth and they weren't growing, and they wanted to figure out what was it that led get people to engage more and use the Facebook app more. And they looked at a lot of different things. They looked at, you know, maybe if you fill out your profile more, that'll make it more attractive. Or maybe if you upload more photos, you're more likely to come back. Do you know what they ended up finding, Michael?

If they could just chat with their friends.

So it was, it was how many friends you added. Actually, it was just, just adding a friend on Facebook. And you needed to— threshold was 7 friends. If you got at least 7 friends added in Facebook, you are much more likely to come back to the product. Now you might say, well, duh, the whole category name is social networking, that's obvious. But what's funny is no social network before Facebook had figured that out. MySpace didn't figure that out. Friendster didn't figure that out. You know, Orkut didn't— like, none of them figured out how important it was to make it easy to add friends.

Yeah.

And so it was, you know, it was just a little bit too hard on all these other platforms, and they capped out their growth and they didn't become this giant behemoth. But because Facebook was the first company to really look deeply at the data and its users, they ended up figuring out that the, the friend aspect— and that's why the whole category is now even called social networking, right? You know, is because that part of the product is so key. You, you know, and that as a result of that insight, they made it, you know, they did all sorts of things. They made it suggested friends for you. They had your friends suggest friends when you first signed up for the service. They, you know, just would constantly bombard you with different ways to make it easy to add friends. And so that was just a really key moment for them as a company. And they broke out and, you know, now they have billions of users on their product and they're one of the big five in tech and, you know, they're this massive business. But that insight all came from the product data. And again, it was not obvious to the people at Facebook at the time.

Yeah. I want to make this point. I said like it should not be a big surprise, but I don't want be the person who says it's super obvious. We don't need data analysts to figure this out. The whole point is that it's buried inside all of this stuff that like you can't see the forest for the trees kind of thing. Right.

And, and, and vice versa.

Exactly.

Exactly.

And we're more inundated with data today than we ever were before. I wanna ask you this too though.

Look, look, look, go ahead. There's one more famous one that I wanted to share where Apple, when they launched the first iPhone, do you know what the key interface insight they had was versus all smartphones before was that the Keyboard was a total waste of space. And they did that by looking at how people use these phones and looking at data. You might not think, well, Steve Jobs, they just came up with this insight because he's a genius. It wasn't the case at all. It's because they've really looked at how people are using these things and understood that deeply. And so this idea that like, you know, you can just build a great product in a vacuum just by, you know, coming up with great ideas, totally, totally misguided. You are so much more powerful as a builder if you have the data.

Yeah, I want to also make this point that we're so overwhelmed with data now, and once you figure something out about one company with all the sort of data privacy and data protections in there, can you then take those learnings for other companies that are in similar businesses and maybe even in different businesses, right? So if you're looking at all this data at scale, right, Peloton figured out that there was a social aspect to this that maybe they didn't understand, right? And again, we can go back to SoulCycle and just say very few people probably went to SoulCycle alone, but nobody was writing that down. When they signed in, they didn't say like, hey, I came with my friend Bob. Like that, that just didn't happen. But they probably did do that. So they didn't have the data to understand that necessarily. But on a digital platform, I think this gets back to what Amplitude does. All the data's there. You know this from, from going to school at MIT, an engineer's job at scale is to minimize noise, right? And find signals. And that's what you're doing. You're just trying to find these signals.

Yeah, exactly. And it's, it's hard in products because products are big, complicated freaking things. An average product has thousands of different touchpoints you can do. And the question is, out of those thousands, which are the ones that matter?

Yeah, that's the, that's the other point too, right? Is that what does matter? You can have all these signals going off and just going like, oh, let's go down, oh, let's not go down that road cuz that road's a dead end and it doesn't mean anything. But you don't know it really until you go down it. And physically you can go down every single road. It takes time. Digitally you can actually go down a lot of roads because you have all this compute power that allows you to do that and do this kind of analysis, right?

Yes, exactly.

Do you see, so again, I said I want to get geeky. I want to get a little bit more geeky. Do you see this change in the way software as a platform is getting developed by breaking things into microservices and then also being cloud native as opposed to transferring stuff to the cloud and being API first so that every microservice that you build— I'm presuming you're building microservices unless things have changed— and because they're cloud native, then they can be everywhere at the same time. It allows you to test and iterate and do all this stuff faster, and then that connectivity through the APIs is so much easier.

Does all this make sense? Yes. So you, you've hit on the key of why this is possible to do. Why hasn't Amplitude or a company like Amplitude been built before? Well, Frankly, because the architecture of how products and the internet has been built has not allowed us to track and understand all these things. If you're doing all this stuff offline, you can't— it's much harder. It's impossible to know, you know, you can't really track. But if you're doing all this in the context of a web page or a web app or a mobile app, you can actually track every single action that's happening. And so it's crazy. For the first time in the history of product development, you can actually see people using your product as they're using it. Yeah. And that's just enormously powerful as a product creator. You know, it's like, imagine if you're, you know, it's like you can see people using your art while you're creating your art and see how they're reacting to it. And there's no longer this big barrier between, okay, you create it and then you put it out there and then you wait for the reactions. No, you're creating it while it's being used. And this is what the best product teams are doing. They're continually getting feedback from how it's being being used and using that to iterate and adjust and change how they approach things and make a better experience for their users. And so products are no longer these static things that are just shipped out there. They're constantly adapting and evolving to you as a user. And so that's a really cool future. The other thing I want to point out is the part that's crazy to me. So, uh, one of the big trends in the last few decades has been the rise of agile development. And so this idea that we're going to constantly iterate on a product and ship new versions all the time. And products are so much better as a result. It's crazy to like, you'd be, you know, if you starting a tech company today and you're saying, oh, we're gonna use a waterfall to build a product, you'd be thought of as crazy. You would be crazy actually, but go ahead. It's standard to iterate, continually iterate in an agile manner on your products. The part that's missing though, is it's great that you can ship new versions of your product all the time. But how do you know if those versions are better and they're working and what's going wrong or what's going well on them? Well, that's exactly where the second half of this revolution, which is product data and you get feedback back is coming in. And that's, that's the core. That's at the core of what we do. And we're in such the early days of this, like, I'll tell you, you would be shocked at how little the average tech company knows about how customers use its product. It's crazy bad. Now, the very best, you look at the Facebooks or the Netflix or the Intuits or the Atlassians, they know it. But the average tech company, really, really poor on this axis.

But this is why I want to get a little bit geeky, right? Because I wanted to point out to people that they don't see— most users or most clients don't see all these changes in the backend architecture, right? I think when people think about a CTO, right, like a chief technology officer, they just think about a guy or a gal who super knows how to program, as opposed to somebody who's architecturing the infrastructure around into which that programming is going to get done. And that's a big, that's a big difference, right? Because you can be a great coder and really bad as a CTO, but because the architecture changes, it's hard for the people who are using the product to understand what type of information is actually now being generated created to help, and then how that product can get developed. I like to think about when Tiger Woods signed on with Nike before they had a golf business, right? So he's swinging these clubs every single day, and the people who are making the clubs get no feedback except when he goes back into their laboratory and says, "Can you change the angle on this a little bit? Can you do these kinds of things?" Obviously within the rules of how clubs can get made based on what the rules are in the golf industry. But think about the way digital products are used today. It's a, it's a massive paradigm change to be able to literally sit there and watch people use the product and see where they're drilling down and what they're using and look at the stuff that they're not using at all. Right. I use digital products every single day.

Sorry, go ahead. Yes, yes. That's exactly right. It's, it's this total, like if you think of it, this even is before digital products, you know, you, you go back to the physical products like a car or cookware, it's like like you're just— they're being used in a vacuum. It's like there's this total separation. You build them and you ship them and you have no idea how they're used in the wild. And so we have this opportunity for the very first time to actually look at how people use these products as you're building them. And that's, that's transformative to the way products are going to be built. And that's, that's the really exciting thing.

So are there companies that get this and companies that don't? I always like to talk about sales as well. You know, like when you rock up to a potential client for the first time and say, you know, again, BlackBerry would be the perfect example of this, right? Even after the iPhone came out, they're selling, selling, selling more and more. Their profits went up, their sales went up. There was a little bit of momentum and inertia there, right? That just kept it going. But if they had the data that you have, they would've known that this is going to die. Like not even a slow death.

Totally, totally.

That this is a much more— Sorry, go ahead. Yeah, because no one wants this anymore, but they're doing it because the people that are buying this thing aren't the people that are using— like they're all— there's all this data that's out there that they would've had where they would've said even maybe years in advance, I think we need to change the way we're building this product because the lower third that phone is getting used in a way that makes no sense in the context of how phones should be used.

Is that fair? Yes, that's absolutely right. The companies that do this really well are the companies that are winning in the tech industry today, and everyone else is scrambling to catch up and is deathly afraid of being left behind.

And again, this gets back to something I said before, right? I grew up in the '70s and '80s, so make any age joke you want, but IBM was like a behemoth and it was like a monster. And if they didn't do it, then it didn't happen, right? And if they wanted to do it or if you wanted to do it, they would just come in and do it. Like, I remember when OS/2 2 came out, even Microsoft was like, oh no, a graphical user interface from the company that's making the computers. Like, people were actually scared, but they were just so far behind. Again, when you go into companies where the infrastructure isn't even set up to be able to use your product, right? Because the presumption is that if they're selling all these products, there has to be an infrastructure to feed the data. You can't just plug something into something that doesn't have data pumping out of it, right, as well. Yeah. So how does that work for you guys?

So that's the number one challenge that we work with customers on where their biggest challenge is you have these products that are already built, highly complex, thousands of different things you can do on them, and then you have to instrument them for the first time. And that's, that's part of that's, you know, I skipped over it in talking about Amplitude, but that is the hardest part of getting set up with us or doing this really well, which is how do you actually collect the data, make sure you're collecting the right data, make sure it's tagged properly. The other big thing is like no No one can keep thousands of data points in their head at the same time. Like zero people, you know, nobody can do that. And so the question is, how do you actually manage the collection and organization and analysis of that? And so we have a number of things that, you know, best practices based on how people use Amplitude to do that. But it's a hard problem in that it's not a given unless you invest in that capability and being very thoughtful and deliberate about how and where you collect data in the product. It's not a given that, you know, you can just get all these insights out. Peloton, you know, was only able to do that because because they had worked really hard to think about what data they're going to track. Same with, you know, a lot of the other companies that I had mentioned.

Did you watch the news on Peloton? Do you know? Because they went through this whole transformation, right? And this whole— I don't want to say restructuring because it sounds like a bad thing, but in the background, all this stuff is happening with Amplitude and you're watching the news going, yeah, that's going to be fine because we know where this is going. Like, do you watch the news for other companies where you're working and just think it's going to be okay? You know what I mean?

Yeah. You know, it's, it's funny. I think one of the things I wish we had done earlier on, maybe we'll do it in the future, started an Amplitude investment fund and then invest in companies that are big Amplitude users. I mean, like, look at some of our biggest customers: Intuit, Atlassian, PayPal, Square, you know, or now they're called Block. Like, they have been some of the best at really understanding their— they have an edge over everyone else because they understand their customers so deeply. And if we had invested in them when we started Amplitude, we'd be doing super— we'd be doing even better than we are now. So maybe, maybe one day, but it really is a competitive edge. It's, it's crazy to think about, you know, just knowing your customers and having your team use that data in the right way. The other mistake I see a lot of other companies making is they think they build all this stuff out themselves, they try to get their data team to build out themselves. And it's incredibly painful to do. Like I talked to data leaders out there. And I'll ask them, hey, what percent of the way are you building out your data stack vision? And the response is never above 50%. It's always something like, oh, 5% or 10%, you know, it's like really short. It's like, okay, well, at some point this thing has got to start producing returns and you guys started, you guys showed them that. And, you know, by the way, if we, you know, we can help you do that. But it's so funny that, you know, they're all trying to reinvent the wheel on this thing.

Yeah. Which is impossible. It'd be like you going out and building a Peloton bike and trying to compete. Yeah.

No way.

No way. But there's no way. That's not where your core competence is. Totally. Right. So your core competence is obviously building this data analytics product, but theirs is building bicycles and having that social aspect to it. What is, what does, um, sorry, I lost my train of thought. Work with me for a second.

No worries, Michael. One of the other things I actually want to call out I think is important. Go ahead. So a lot of people right now, economy's heading in a downturn, resources are becoming scarce, you know, should you really care about this idea that you can use product data? I would say this is now is the time where it's incredibly important. One of the things that one of our venture capitalists from Sequoia told us was that this is from Pat Grady, Spencer Skates, he said that a downturn is the perfect time to cut fat and build muscle. And if you think about what is fat and what is muscle, well, muscle is stuff that compounds over a long period of time, a.k.a. your product. And if you use some, that's where using something like us to build a better product really comes into play. Whereas what's fat? Well, stuff like, you know, marketing or sales programs, that's really only going to get you a one-time return on the investment. And so it, you know, we see a lot of companies that are really looking to make cost savings or really look to— looking to drive their growth more efficiently turn to Amplitude because this is such a good time to invest in your product because that's what will allow you to outlast everyone else. A product, you know, if you look at companies that have driven their growth through product-led motions, they've done by far the best. You know, I talked about Intuit and Atlassian, our examples. HubSpot is another great example that's been a big customer of ours for a number of years. So there's so much there.

Coming out of my trading background, We think market dislocations are always opportunities, and the companies that actually build into those dislocations are the companies that continue to succeed. Are there other applications for this that where you haven't used it yet? I mean, e-commerce seems to me to be like a massive opportunity. Do you deal with e-commerce companies and try to tell them you have all the— because, and I think Amazon is a perfect example of this too. At some level, I feel like you strip away AWS and the Amazon e-commerce business could again, just go away at some level. I know that's controversial, but, and I, and I I know they, they do a lot of data analysis, right? But product discovery is almost important as product availability to me, right? When I go on to a site, you know what I mean though, right?

In other words, I totally— so tons of— yes, absolutely. So tons of, you know, I already talked about Walmart, you know, they're, they're an Amplitude customer. There's a whole bunch of others, Under Armour, Gap. One of the actual big ones is quick service restaurants. So Chick-fil-A recently became an Amplitude customer. Interesting. And one of the things they wanted to figure figure out is like, it's literally like the 'do you want fries with that' question. So what's the thing that you can offer? And so they actually figured out what items are good to buy if you buy certain other items. Turns out fries are actually a really, really incredible upsell to do, which, you know, again, sounds obvious, but not obvious if you're kind of in the trenches building the app. And so we've been able to help them do customized product recommendations in the context of their product on stuff that we think is more likely to sell. And that increased a bunch of their average order value, which was fantastic.

What kind of feedback do you get from your existing clients that then helps you build on top of the product more, if that makes sense? Right. Because again, you know what you know, but you're not dealing with all of their products, right? But do they come back to you now that they use your product? Do they come back and say, hey, you know, we think this is a thing, can you do some analysis on that? Or if we could look at it this way, it would be better. And then you can build that product out and everybody else can then get the benefit of it, if that makes sense.

Yes. Sorry, say it again, Michael.

In other words, how are the big partnerships that you already have driving the product growth that you, that you have as well? In other words, they come back to you and now you have all these insights, all these ideas, and then you think, oh, actually, we could productize that as well.

Oh, yeah, absolutely. All of our best features have come through things our customers are already trying to do on their own with Amplitude. I mean, this is a core part. One of the things I actually love about the analytics business is that there's a million ways to slice and dice your data and look at it through different lenses, and you can combine it and do all sorts of, you know, all sorts of ways to view your data in all sorts of ways. I think, and so the core muscle we have is take something someone asks for, productize it, and then give that to the rest of our user base. Like that initial Facebook example, we now have a product called Compass that will look across all of the different things your users can do in your application and pick out which are the most correlated with that. And this is the core engine of Amplitude. This is what makes us able to succeed versus everyone else in the space is we're constantly— we— it's not like we have any crazy, you know, here's the one really cool feature and the one really cool insight that we have that no one else have— has. It's more that we can continually get new versions and new features and new types of analyses for you to do with us out there. And now we have this phenomenal library of analyses that, you know, depending on what industry you're in, might apply to you and then help you build a better product as a result and give you the knowledge of all the past prior people that have been using Amplitude. Interesting.

What are the implications of things like GDPR and data privacy on your ability to continually analyze this stuff? You see a lot of stuff going on in Europe where even parts of Google Analytics are getting shut down. Does this impact you guys at all?

So this is actually a huge opportunity for us. So I want to just talk about, I think people kind of paint data all with the same sort of brush and people are like, oh, anti-data, you know, data is anti-privacy. That's actually not the case. I think of a spectrum. On the one hand, third-party data that is sold and resold versus advertising networks and used to create a profile on you. People hate that stuff. Like they don't want companies to track them around the internet. They don't want Facebook or Google or any of these other companies to build a profile of who you are and what you're doing. Doing on the internet. On the other hand, and so that stuff is basically being legislated out of existence. On the other hand, you have first-party data, which is data that you're using within the context of an application. Now, this is like Netflix recommending shows to you based on what you watched before, right? Everyone expects that, you know, it's like, duh, like Netflix has that data. Where people have a problem is that data is resold or shared with other third parties, right? So everyone expects products to better and for products to track what you're doing in them to help deliver a better experience. And that's what we do. What people hate is when that gets shared or built up as a profile for third parties or other people in the ecosystem. And so GDPR is actually this huge opportunity for us. We've seen like there's so much regulatory pressure on Google right now. There was an Austrian court last year that, that declared Google Analytics illegal. There was an Italian court earlier this year that also declared Google Analytics illegal. Yep. There's this new regulator in France called CNIL, CNIL, that is saying, okay, here's who is approved and who's not approved in terms of tracking data. And of course, Google Analytics, not approved. And so because Google's from a regulatory standpoint, honestly, they're kind of just walking away from the analytics business at this point.

They are.

They are. They forced their users to upgrade from Universal Analytics to GA4 as part of this, but GA4 is a really crappy product and migration's incredibly painful. And so it's this kind of— there's this vacuum in the market where people are like, hey, what is it? I need to figure out what I can do next. And so we've had a number of companies come to us, they're like, hey, we gotta find an alternative that's privacy-friendly from Google Analytics. And that's a— that's been a huge opportunity for us so far. And so I think it's really important when it comes to these debates around data and privacy to be clear about what sort of data you're tracking and what are you using that data for, because that's what people really care about. Yeah. You know, again, everyone expects a product— you to track data on how people are using the product so you can make that experience better. But when you use that and resell that to third parties or use it to advertise, that's when people— that's when the regulators crack down and people get really worked up.

Can I make an offline analogy just to— for people to understand this? And again, tell me if I'm wrong here. If I go into a restaurant frequently and I order a Brunello because it matches whatever I'm and the waiter remembers. And when I come back, they say, the last time you were here, you had a Brunello with this steak, but now you're having fish, you should have this Pinot Grigio instead, kind of thing.

Yeah.

Yes. Nobody minds that at all. As a matter of fact, that's—

you love that.

Yeah, I die for that. That's where I keep going back. But if two things happen, one, if they say, do you mind if we write it down as a 37-year-old male that you had this?

That's starting to get weird.

Yeah. And then resell this to the restaurant in New Mexico. And then when somebody goes there, they're like, well, Spenser had this, and he's in the same cohort as you, we rec— you're like, get away from my table.

That's so weird. Yeah, exactly. That's exactly the right analogy because the idea that, hey, you're gonna get better service from working with us cuz we get to know who you are. Right. Great. But the idea that, you know, hey, this data's starting to get reshared and sent all over the place. It's like, get me outta that world.

Yeah. But that's the difference between what you were talking about with Google Analytics and with what Amplitude is doing, right? I don't think people understand this idea of third-party data cuz it doesn't sound like they don't know what that means. And first-party data stuff, you just know I'm I'm the waiter. I served you for the last 10 years. I know what you want. Oh, that's your aunt. I know what she wants too, because she's been at this table. That's okay. And nobody complains about it, right? But the flip side of it is this third-party data, it's like, it's kind of creepy. And that's not what this is anyway.

That's exactly right.

That's exactly right. Okay. I want to end here because I feel like I've taken up way too much of your time and I could keep going on. Spenser Skates, the co-founder and the CEO at Amplitude. I hope you had as much fun as I did.

Michael. I really did. If any of the listeners here want help with learning about how to do a great job with product data, come check us out at Amplitude and we'd be happy to help.