How I Built A $1.4B Software Giant Called Amplitude

In this interview, Spenser Skates reveals how a failed voice-recognition app called Sonalight led him to found Amplitude, why he went a year and a half before charging a single customer, and how a $1,000-a-month bet from a company called Super Lucky Casino became Amplitude's first sale. He also opens up about the personal trade-offs of building a company that took over every part of his life.

Hi, my name is Spenser Skates, I'm CEO and co-founder of Amplitude. Amplitude is a digital analytics platform: we help you understand what your users are doing in your digital product. We've grown enormously fast. We launched it in 2014, and since that time we've grown from $0 to $270 million in annual recurring revenue over the course of just under 10 years.

We have 2,700 companies on us, including a lot of the Fortune 100 in the United States. We took the company public via a direct listing in 2021; we're valued in the billion-dollar-plus range. Now we're thinking about how we continue to grow the company and take over the world.

At Microsoft, at Google, at consulting firms, or in finance, except for one person: my current co-founder, Curtis. Moving to California, moving in with Curtis, I got a mattress on his bedroom floor, and we did that for a few months. The first thing we worked on was this app called Sonalight. It was a voice recognition application that allowed you to talk to your phone to send and receive text messages. This was actually before Siri. We launched it, we had a good amount of usage, we had a few hundred thousand people download the app, and so we knew somebody was using it, that there was real usage for this thing. We got up on stage at demo day, and we had the most amazing demo day presentation.

The problem was, a few months after demo day, we couldn't get any of our users to stick around. They would try out the app, they'd be like, wow, this is kind of cool, but then they'd churn out and they wouldn't continue to get value out of it. So we wanted to understand why that was: we built our own analytics platform to help us understand. We couldn't figure out why we couldn't get their attention, because the key factor was how accurate the voice recognition was.

To figure that out though took all this munging and customization of the data, trying to understand how that correlated to whether someone comes back two weeks later, and whether they got a first recognition event and it was accurate. We just spent all this time on this internal infrastructure. When it came time to wind down Sonalight, we said, wow, actually a lot of the other companies we talk to say hey, we can't get these same insights about the product experience, we really want to know what drives engagement and retention, so can we get what you built on the analytics side.

First we said no, that's just an internal tool we have. But then, after we decided to wind down Sonalight, we said hey, maybe there's something here. I remember in June of 2012 we ended up making the decision to switch over full time into building out this analytics platform, and that was the start of Amplitude. We launched the company in 2014, and it's been growing since.

To this day, in terms of folks who I know have my interests as a founder deeply at heart, Paul Graham and Y Combinator are at the top of the list. Most of the time within YC while we were working on Sonalight was spent talking about how do we grow it, how do we launch it, how do we get more users on it, and we'd meet with them weekly about it.

I remember this one conversation: Sam Altman was starting to get involved at the time because he had just wound down Loopt. I remember Sam really challenged us about how passionate we were about this problem and whether we thought we were going to be successful. I remember saying to him, yeah, we're not particularly passionate users of voice recognition as a technology, we just think it's a cool technology. He said, yeah, you guys should work on something else. That was the genesis of us deciding to wind down Sonalight and ending up working on Amplitude.

One of the biggest mistakes we made with Sonalight was not spending enough time talking with our customers. We would build product, because as engineers that's what you know how to do: you know how to build product, but how do you talk to customers? It's weird, it's awkward, it's uncomfortable. One of the things we did with Amplitude very purposely was saying I'm going to make sure I spend half my time talking to customers no matter what.

Before we even built product, we just reached out to 30 other companies to do research, to ask them, is this a pain point, would you use this, what are your pain points, how do you think about this in terms of how it fits your business. Through that process we actually found about seven companies that were interested in using Amplitude; turns out only one of them actually ended up using it after that, but we felt we had enough validation and enough understanding of the problems they wanted solved.

That's my number one thing: as a founder, how do you spend half your time interfacing and really understanding customers? It's so easy to do other things. It's so easy, if you're an engineer, to build product; it's so easy if you're a designer to spend time on that. But it's very, very easy to not spend time talking to your customers. So as a founding team you have to be very deliberate about spending 50% of your time engaged with users.

Perfect the enemy of good: my point there is that I think one of the mistakes people get into is thinking, I don't want to let go of this idea. We were really in love with the potential of voice recognition; it was this new technology at the bleeding edge of what's possible, it was hard to imagine a future without it, it was a highly technical problem. It was what I'd call a 95th percentile idea, in the top 5% of ideas.

After doing it for a year, we realized: one, we weren't passionate about the space, and two, the technology wasn't quite there yet, so it wouldn't be possible to create a breakthrough. So while it was a good attempt, we knew we could do better in terms of what we built, and we got incredibly lucky. We searched around at a lot of different ideas and we ended up zeroing in on analytics, and in retrospect that was a 99th percentile idea for us, top 1% of ideas.

First, there was a big pain out there: we knew the web was shifting from a marketing-centric web to a product-centric web and would require a new set of infrastructure, so what Google Analytics and Adobe had done in the previous generation would need to be redone. There was an opportunity to build something massive. Second, we were the right founders for it: we were hardcore algorithms engineers, and figuring out how to create a system that successfully processes massive volumes of data and spits out a result was perfectly in our wheelhouse.

The opportunity was there, and it was a great match for who we were as founders. I think of it as a 99th percentile idea, and I'm so glad that we made the decision to wind down Sonalight as a result and go after something great. You often want to just go build something, and building stuff is great, but you also have to ask yourself continually: am I building the right thing?

I regret it in retrospect: we should have been able to launch in six months instead of taking a year and a half. I made such a big mistake: I stopped talking to users in the middle of it, and we just focused on building. As it turns out, there were a lot of people who would try it out and use it for free, but none of them were actually interested in paying us. It wasn't until July of 2013 that I actually got our first customer to agree to pay us, and I should have asked for money much earlier, because then I would have realized we weren't talking to the right companies.

I remember I was in a meeting with this company called Super Lucky Casino, our very first Amplitude customer, and I went through the demo and the pitch with this guy named Brett Taylor. We got to the end of it, and he asked me a question I had never been asked before up to that point: how much does the product cost? I was like, what, I don't even know what to answer, I've always just given this away for free and assumed people would use it for free.

Then I remembered the advice of Patrick McKenzie, who says always charge more for your product. I thought I was going to charge maybe $50 a month for this thing, so I thought, let me double that and ask him for $100 a month. $100 a month is enough, let me 10x that, think of the biggest number I can think of: $1,000 a month. So I said $1,000 a month is how much it costs, with as much confidence as I could muster, and to my surprise he said okay, we'll pay for it.

I was like, oh my God, holy smoke, someone actually wants to buy the software we had built. All my dreams of building software that we could charge for had been validated in that moment. I thought, wow, I can actually build a piece of value that someone wants to buy, this is incredible. We launched the product six months later, in 2014.

In retrospect I should have been asking for money way earlier. For all those customers I would have gotten a lot of noes, but that would have been okay: I would have gotten much faster to someone like Brett, and realized I should have been out there trying to keep finding someone until I found, found, someone with enough pain that I could get them to pay us money. I was afraid because I knew they would say no, or I had a suspicion they would say no, and I wasn't confident enough in my offering yet.

I thought, well, I need to make the product better before I can ask for money. In retrospect that was backwards: I should have just said, look, I know the product's early, but I want you to commit some money so that as we build it and make it better, you're committing to it too, with the investment you're making. Frankly it was a lack of confidence in the product.

I should have known: it's okay to have an early-stage, underdeveloped product, and it's still okay to ask for money for it, even in that state. You're going to get a lot of noes, but it doesn't matter how many noes you get, you only need one customer to say yes for you to actually build a business around it. Understanding that dynamic earlier, and just focusing on companies who are willing to pay you money, would have allowed us to much more quickly get to the state where we focused on customers who actually were interested and had enough pain to pay.

After that first customer, Super Lucky Casino, they introduced us to other companies. Brett was an ex-product manager from Zynga, and he had a lot of other former Zynga product managers who were used to having this sort of data and insights and wanted it now that they were off at other companies. He ended up introducing us to Sei Chen, who was running this company called Hey, and then a bunch of other ones, and we started to grow within that community.

Finally, in February of 2014, we launched the product for the first time. I don't know how the heck they found us, but somehow they saw this TechCrunch article and reached out to us. They ended up paying us $3,000 a month, and it was like, okay, wow, a legit company wants to buy us. Then it just started snowballing from there, with customers referring other customers, and we would spread as a product.

At the beginning, what was funny was the landscape was very crowded: there were tons of analytics companies that all did similar-sounding things. Google Analytics was the biggest and the default thing you put in there, but there were lots of others too, the list just went on and on. There were probably 20 or 30 companies in this space, it was a very crowded space.

The thing we saw when we looked at it is none of them went deep on the problem: they all did surface-level metrics and stopped there. We said, if we go much deeper we think there's a ton of value, and you don't have to scale off as you grow either, because you can't get to the data volumes, you can't get to a large enough scale, you can't get to the depth of insight that you want. Most companies don't use this software yet, most companies don't understand what their users are doing in their product, it's like a black box they cannot see through. Our goal is to open up that box and help everyone see what's going on so they can understand that experience and create a better one.

We grew from $0 to $1 million in the first year, then from $1 million to $4.5 million, then to $14 million, then to $31 million, then to $53 million, then to $80 million, then to more than $100 million. We started in 2014 and got to $100 million by the start of 2020, so it was about a six-year journey in total, and I feel incredibly lucky to have hit that.

In terms of when you're going from zero to one, you're just begging anyone to use the product, because you don't have much of a product or a brand or a customer base, or that much advanced knowledge of the problem. You're trying to consistently pull off miracles, so I'd spend my time going out and talking to customers directly, full time on the sales side, just trying to find opportunities for us to have as customers. We would commit to new parts of the product, we'd say, if you buy us, we're going to build this, this, and this; that was the core part of the motion, because there was nothing there, so you kind of had to just agree to whatever a customer wanted in order to get them on.

Once you get to $1 million in ARR and go from 1 to 10, you have a kernel that's working, and it's all about trying to do the same thing; you'll still have to do some of it, but then it starts to pick up momentum of its own, it's not something you're constantly pushing anymore, there's some momentum to it. Once you get to 10, stuff starts to break down a lot, because what worked, doing heroic efforts and single deals, does not work anymore. You need to build an organization, you need to build repeatable processes and scale, so you need to start bringing in outside executives who work on taking the thing you had and building scalable programs.

You need another salesperson on this: how do you enable them, how do you track and forecast so you can have good forecast accuracy, how do you manage the creation of pipeline regularly, so you're always taking something that was a miracle you pulled off and figuring out how to replicate it in a much more scalable fashion. It's crazy, it's still surreal, I don't believe it, I'm way more successful than I ever thought we'd be. I feel so lucky that we were at the right place at the right time and tried to work on this in the right way, so I feel incredibly lucky to be here.

Now what I'm focused on is making sure we don't waste the opportunity: we're the leader in the market that we have, and we have this promise of how we can go take over things. I'll tell you one bit of insight, a secret about it, which is that no matter what size you're at, you're always expected to grow more. Everyone wants to know how you're going to be 10 times bigger than you are today; even Google gets that question. Even if you're as successful as Google or Amazon or Microsoft, they want to know how you're going to be 10 times bigger. It's both a blessing and a curse.

It's a blessing in that there's always an opportunity to do more, but it's a curse: just like when I was starting out, I felt unsuccessful, I feel unsuccessful today, because there's so much opportunity we have to go realize. Now there are parts that get easier: you get a lot more help, you get to hire great and talented people, you get the knowledge of being in the market, you have a brand, a product, customers. But there are also parts that get harder: it's hard to run a large organization and be agile, it's hard to make sure you don't get too full of yourself and that you stay ahead of what's next. So in a lot of ways, you still feel like that entrepreneur from the early days.

You can start a company at any age. You can start it in your 20s, your 30s, your 40s or 50s, any age is possible. The key thing to understand is how hard it is and how much of a life commitment it is. I've been very deliberate about sculpting my entire life to allow me to build this company, and that extends to my personal life, to where I live, to what I think about. I just obsess about what I'm doing here day in, day out, so you have to be in a spot in your life where you don't have anything else going on.

My number one advice would be: understand that it takes over every single aspect of your life. If you're willing to do that, I think most people are capable of starting a company, but if you're not willing to do that, if you have other constraints or other things happening in your life, very candidly, look, I've fallen out of touch with so many friends because I've been so focused and dedicated on building this, and part of me misses out on that, but part of me realizes that's the choice I made in life, and I'm okay with that trade-off, because I knew that's what it takes to build it.

There's this famous Harvard Business School study that looked at the elements of world-class performance across many fields, and there were three big ones: one was 10,000 hours of practice, the second is expert coaching, and the third is enthusiastic family support. I thought a lot about how you set up all of those: you need to spend a lot of time on it, you need to go get help from other people who are really good at this so you can learn much faster, and then you have to set up your personal and family life in a way that supports this.

What's tough is a lot of people are not in a place in life where they can do that, and just recognizing that is a trade-off of building a company. But if you can set those things up, where you can spend lots of time, where you can get coaching from experts, and where you can get the support of your family, then I think it's very possible to do.