What's up, everybody? This is Cortland Allen from ndhackers.com, where I talk to the founders
of profitable internet businesses and ask them how they got to where they are today
so that the rest of us can learn from their mistakes and their successes. Today, I'm
doing a first for the podcast and actually talking to two guests in the same episode.
They are Peter Reinhardt and Calvin French Owen, the CEO and CTO of Segment, respectively.
Segment is one of the most popular platforms for collecting all of your company's analytical
data in one place and then sort of rerouting it wherever it needs to go. At first, I thought
I was flirting with disaster by having two different guests on the podcast at the same
time. But it turns out that hearing the perspectives of two different founders of the same company
is somewhere close to twice as insightful as just hearing from one. During our conversation,
Peter and Calvin begin by sharing the story behind Segment, which really starts with numerous failed
attempts to get their startup off the ground. They really had zero success early on and very
little to lean on besides themselves, their third co-founder Ilya, and a rapidly declining bank
account. After that, we get into how exactly they found a product market fit and created something
that people actually wanted to pay for. And we also spend some time trying to put a finger on
just why it's so difficult to understand and internalize and consistently apply broad advice
like make something people want until you've both failed and succeeded at doing so. I think the
contrast between the way their business grows and the types of challenges they have to overcome
before and after finding product market fit really illuminates why it's so important to
make that the first milestone you need to go after. All of these growth hacking techniques
and sales and marketing skills really don't matter until people actually want what you've made.
Anyway, I find what these guys have to say very illuminating and I hope you guys will too.
So without further ado, Peter Reinhardt and Calvin French-Owen.
Okay, we're trying something different today. Joining me on the podcast is not just one but
two founders, Peter Reinhardt and Calvin French-Owen. They are the CEO and the CTO respectively
of a very successful and popular analytics company called Segment. How's it going, guys?
Good. It's really good to be here. Yeah, thanks, Corwin.
Can you guys identify yourselves and kind of do like a personal intro so everybody can know
whose voice belongs to who? Yeah, sure. This is Peter, one of the co-founders and the CEO.
I studied aerospace engineering at MIT and then dropped out with Calvin to get started on Segment.
I'm Calvin, another one of the co-founders and CTO here at Segment. I work primarily on
kind of the backend infrastructure and a lot of the data pipelines here. As an odd story,
I actually knew Peter way back in seventh grade from a summer math camp.
Oh, cool. So you guys go pretty far back. Yeah.
Peter, I had no idea you went to MIT, by the way. You were a course 16?
I was, yeah. Calvin was there too.
Oh, cool. Calvin, what course were you?
Six.
Okay, me too. It's very typical for people listening in. MIT's course numbers go by
numbers. So course six is computer science and 16 is aerospace engineering.
Yeah, we have what, 21 departments and all of the humanities are in one?
Exactly. It's all condensed. But anyway, can you guys give me a description of what Segment is,
just to provide listeners with some context and I guess also a description of how you guys
would measure your products and your success. Yeah, so Segment is basically, you can think of
it as a living record of your customers and every interaction that you're having with your customers.
So we help companies collect data from their websites and their mobile apps and their payment
systems and their help desks and pull all that data about interactions with their customers into
a single spot. And then we fan that data out to the analytics tools, the email marketing tools,
the data warehouses, etc., where they actually use that data. So we're sort of like a data sewer
system, if you will. We basically sort of invisibly help the data go wherever it needs to go.
On the flip side, if Segment ever breaks, it's kind of a mess.
Yeah, I think one of the interesting things about that being in that domain is that that's not an
idea that you could imagine someone just coming up with out of the blue. And I'm sure you guys
have a zonky story for how you got there. But maybe let's start at the beginning, since most
of the people listening in are in a situation where they haven't really started a company yet,
or they're just now contemplating getting started. So they'd like to hear about how you
got to where you are today. You guys met in seventh grade. Did you know back then that
you wanted to start a company together? Or was it something you talked about?
No, Calvin, you want to take that?
Yeah, at that point, not at all. I think all we were focused on was learning trigonometry.
But yeah, I think the first time we kind of seriously considered starting a company together
was kind of after freshman year at college, we met our one of our other co-founders Ilya.
And kind of throughout that year, we would spend times on kind of these hair brain schemes,
whether it was to build a digital surround sound set up so that everyone's laptops would be playing
the same music, or Peter and I were trying to game this one like online auction site for a
little while, basically just like building lots of little hacks around different parts of our life.
And even though none of them really panned out or became kind of big projects,
we got a sense for what it was like to work together. And then fast forward to junior year,
and Peter can probably take this over from here, but Peter and Ilya took this one class
called Founder's Journey at MIT.
Yeah, and what happened was, it was actually basically the structure of the class was every
week they would bring in a founder who would talk and just tell their story, not this similar to
indie hackers. And the first week, this guy Adam Smith came in and just talked about how he had
started this company called Zabni and how they had raised to like $45 million and so forth.
And I was just like floored. I was like, wow, this is just incredible. And I was just like,
I was like idolizing sort of everything, everything about him. And there is this funny moment at the
end of class where he was sort of milling around talking about a few people. And Ilya, our third
co-founder was like, you know what? I'm going to ask if he wants to come back to the dorm for beers.
And I was like, dude, don't do that. This is just going to be embarrassing. I don't want to be around
when you're going to ask that. And he's like, screw it. I'm just going to do it, which if you
know Ilya, it's a very Ilya thing to do. He goes over and he's like, hey, you want to come back to
the dorm for beers? And Adam's like, yeah, sure. And in like that two second interaction, the
like accessibility of like starting a company, I was like, this is just a normal dude. You know,
I can like, I can go to a dorm and drink a beer. You know, it was like so humanizing
that it sort of made the entire enterprise feel accessible, if that makes sense. And then there
was a one of the one of the professors of the class was Scott Hamontineja, who is a partner
of General Catalyst. And at the end of the course, he was like, you know, if anyone wants to come
pitch me at, you know, General Catalyst headquarters in Cambridge after the class, you're welcome to
come do so. So I was like, well, I definitely want to do that. So why don't I pitch some of the sort
of like hairbrained idea for a tool that would let professors monitor whether students were
actually reading their course notes. So weird idea. Anyways, it's a bad idea. But pitched him
on it. And at the end of the conversation, he's like, well, this is cool. If it's going well in
the spring, we'd be happy to put in 100k. I was like, what? I have never seen that many digits.
And we're like off by two digits from the amount of money I've ever seen before.
And so I was just like, this is crazy. So maybe this is like a lot easier than it seems. So we
had just like a sequence of these sorts of experiences that led up to us applying to Y
Combinator in the spring of 2011. That just made everything feel like, okay, this is doable. And,
you know, frankly, we were in sort of a good unusual, like time and location to actually
get exposed to these things. It's pretty lucky. So they didn't have the course that you took when
I was at MIT. Or if they did, at least I didn't know about it. But interestingly enough, I also
was really inspired by having Adam Smith come give a talk. I don't know what the event was exactly.
I think it was something put on by Y Combinator. They're going from campus to campus. But Adam
Smith came and talked about his experience. I did not think to ask him back to my dorm for drinks.
But I talked to him a lot at the event. And it was just incredible to me as well.
I think one of the things that you said that stuck out to me was how you're inspired by the
fact that he was just a normal dude. And it made it seem like it was possible for you to do something
that was similarly incredible. But at the same time, all three of you guys are far from normal
dudes and the common way that people might think of it. I mean, all of you went to MIT. What do you
think, you know, sets you apart? Or do you think anything sets you apart? You know, do you think
anybody could have done what you guys have done? I think a lot more people could do this than think
they can. I think there's like a pretty significant like idolization of founding things of CEOs at
later stage companies, etc. It's like, I think it's certainly it makes a better story to dramatize
it, which which is fun. But I think it I think in many ways, it ends up creating a barrier to doing
it. I think it's a lot more accessible than people realize. I'd say if there's anything that helped
us kind of succeed, or kind of break out of this pattern, where for a long time, we were just trying
to find product market fit, and we're building all these different things that didn't work out.
It's actually the closeness and kind of the relationship of the four of us at the time,
how we kind of built things out as a team. Basically, when we started, we started with
this classroom idea, it was myself, Peter, Ilya and Ian. I think it took about a year and a half
of just failure, various failed ideas, where we're all living in the same apartment, we were all
constantly thinking and brainstorming and figuring out what next idea we take.
We had whiteboards, like lining the walls, we basically be hanging out 24 seven, Peter and I
shared a room where there is a foot and a half between the two beds, and there's really nothing
else in the room. I think that kind of closeness, and the strength of the relationship that you get
from just working with other people for that period of time, is really what allowed us to then,
when we came to the current idea, just execute relentlessly against it and actually build out
segment, kind of the primordial pieces of what exists today.
Yeah, I think having co-founders to rely on is such a humongous advantage. If only because it
helps you not quit. I mean, when you talk about spending a year, year and a half working on an
idea that ultimately didn't pan out, it's so much harder to endure that and keep going as a solo
founder, where nobody else understands what you've been going through, no one has seen the journey,
nobody can really talk to you about it as an equal, versus having people who are doing it
with you and you just say, okay, this might not work out, but at least we're all together,
we can figure something else out. Was there ever a time where you guys were on the verge
of quitting before you ended up getting into Y Combinator and dropping out?
Yeah, I would say 100% that the most important part of the relationship was the quitting
prevention mechanism. I don't think in that year and a half of struggling for product market fit
between when we got into Y Combinator and when we launched Analytics.js on Hacker News, I don't
think there was a moment that anybody seriously considered actually leaving or quitting or going
and getting a job. We were all deeply committed to each other as co-founders and I think that
the strength of that personal relationship and a few other smaller things that we did to
make it a fun or environment that we wanted to be in made a pretty huge difference in actually
allowing us to ride out that year and a half of bad times. An example of another thing was we
slightly overspent on having a nice office apartment so that it would be a place that we
wanted to be rather than a place that over time grew to be a place that we wouldn't want to be.
So it was like an apartment with a view, for example, which I think makes it some place that
you want to be. Yeah, it's like little things like that and just paying attention to your
psychology and your mindset to extend your ability to actually give your all to what you're working
on. I think another thing that you guys did that is challenging for a lot of people is that you
were able to juggle, at least to a certain degree, multiple responsibilities because MIT is not an
easy school to be at. In fact, having any amount of coursework or a job or family to take care of
really takes away from the time that you can devote to your startup.
How much time are you guys spending on your first ideas while you were still in school?
So I guess when we first applied to YC, we are honestly just building prototypes.
And so we applied just after or kind of at the tail end of junior year. And at that point,
we were going around to all of the professors that we knew. We even talked to Robert Morris,
who was a Y Combinator partner. And we said, hey, what would you like to exist
for your students and your lectures? And it was kind of out of those conversations that we spent
maybe the last 90% of the last four weeks of school just focused on that piece as we were
applying and then subsequently got in and then subsequently said, okay, at this point,
we're going to decide to take some time off from school and actually focus on this chance full-time,
because we feel like it's one of those kind of unique opportunities that might not come around again.
So tell me about the idea that you guys got into YC with and what those first months in YC were like.
Yeah. So the idea was that we would build a classroom lecture tool that would give students
this button to push to say, I'm confused. And the professor would see this graph over time of how
confused the students were. And we were pretty excited about it because we were students and we
felt like these classes could be live in depth by a little bit of feedback or interaction between
the students and the professor. And some professors were excited about it. Robert Morris had
basically said that this was what he wanted. He wanted a button that students could push.
And so we applied to YC with that idea. And I remember very distinctly in the YC interview,
Robert Morris is one of the YC partners. And he was in there with Paul Graham and Jessica Livingston
and so forth. And PG is asking us a bunch of questions about this. He's like, oh, I was one
of this as a student, because the professors would say something dumb. And then all of a sudden,
he turns to Robert Morris and he says, would you use this? And Robert Morris, I assure you not,
says no. Just a flat no. What a betrayal. Yeah, I actually asked him two weeks earlier.
Two weeks earlier. Were you guys mortified?
Briefly mortified and then tried to recover. Like, well, we talked to these other 20 professors and
blah, blah, blah. Anyway, so they let us in anyways. Yeah, that was the first time. That was the idea
for all of Y Combinator. We raised about 500K a demo day on that idea. And then that fall semester,
we put this into the classroom for the first time. And it was like an unmitigated disaster.
Basically, all the students opened their laptops and they all just went straight to Facebook,
which is probably what we should have expected to begin with. Turns out, putting laptops in the
classroom is incredibly distracting. We actually saw it. We saw a study up here. Someone did a
rigorous scientific study and determined that when students have their laptops out in class,
they spend about 60% to 70% of their time not paying attention,
which was pretty much what we found in our less rigorous product market fit experiments as well.
So it seems like you guys really didn't spend very much time before launching,
actually testing this out. You were doing a lot of customer interviews with professors and perhaps
students. Is that like a lesson that you guys took away from that that changed how you worked
on your future products? Yeah, it definitely is. That's not to say that we didn't do any testing
of the product. I remember we'd frequently be biking over to Stanford and kind of like a math
dash and like testing in summer classes there or take the train up to Berkeley from our apartment
in South Bay and be meeting professors after class or almost ambushing them and saying like,
hey, I just saw your lecture. Would you like to try out this new product? But I think in terms of
how we kind of took their feedback and how we looked at student use across the classrooms,
we didn't really understand the lecture dynamics where for the best professors,
it's almost more of a performance art and they already understand where students confused.
Instead, we were trying to kind of push our view of the world onto these professors rather than
really understanding and listening to what their concerns were. And I think if we'd been further
along, we would have probably taken a step back and said, okay, what's really working about this
product? And there were some things that were certainly working like students would take pages
and pages of notes in the product, which in retrospect seems crazy for a startup that's
three months old and has no visible signs that it's going to persist any longer than that.
Would you actually trust all of your course notes inside that product? But interestingly enough,
people did. And so as we were transitioning kind of through the next idea in the next year,
we started paying much closer attention to the problems that we were actually trying to solve
and not going in with any pre-baked assumptions about how things should be,
instead figuring out what the customer actually needed.
So what did your next product end up being? Because a note-taking tool or a tool for
students to signal to professors when they don't understand what's going on is so far removed from
what you've ultimately ended up with, which is segment. I'm curious what the
trail of breadcrumbs was to get you to where you are now.
Yeah. I think the big failure in that first product was that we felt like it was how we
wanted the world to work as opposed to what problems the world had. As we shifted out of
that product, we were trying to think of like, well, what problems did we encounter when we were
trying to build this? And we felt like one of the problems that we had encountered
was trying to use an analytics tool to actually properly segment the data. So the existing
analytics tools at the time wouldn't allow us to analyze how one classroom had used the tool
differently than another classroom. So we decided, okay, it's important that we build a better
analytics tool. This is a bad idea for a bunch of reasons. It's very difficult to sell the value
of insights because the value is fairly unclear and very spiky. So it's difficult to sell an
analytics tool. It's a really crowded market. There's a lot of companies that build analytics
tools, which now, of course, is to the advantage of segments since we integrate with all of these
different companies. And there are hundreds of them, if not thousands. And at any rate,
we decided to build an analytics tool that would do advanced segmentation. And this is
actually where the name segment came from. So we spent about a year building out the
infrastructure and so forth to make this advanced segmentation possible. Frankly,
really struggled to get product market... continued to struggle to get product market fit
for the reasons that I mentioned before. It was crowded, difficult to sell the value of insights,
etc. So we spent basically the next year sort of rumbling along with that and never really
got any customers to really adopt it and depend on it. So we got to December 2012. So we're now
a year and a half in. And we were starting to run low on cash from having raised a demo day.
And we realized we should probably go back and talk to Paul Graham and sort of get his thoughts
on how we should proceed. So we had office hours with him and we were walking around a little
cold to sack by YC and bring him up to speed. And he sort of stops and looks us straight in
the eye and says, so just to be clear, you've spent half a million dollars and you have nothing
to show for it. How'd you guys feel about that? That was pretty much our low point, I think.
This is not a great feeling. I told him this story later and he looks at me and he's like,
well, yeah. That was the situation. He was totally right. He was like, yeah, I was right.
That was exactly what was going on. Did he give you guys any constructive feedback after that?
Or was it sort of just a pep talk? It was sort of just a pep talk. And then we were like, okay,
well, we need to figure out a new idea here. So pause there, rewind all the way back to that
first week of YC. And in that first week of YC, we've been like, well, we should have analytics
on our classroom lecture tool. And so we had Googled them all and we found Google Analytics,
Kismetrics, and Mixpanel. And we're trying to look at these three tools and figure out which one we
should use. And we're like, okay, well, Google Analytics seems like it's more marketing-y,
Kismetrics is sort of more revenue-y, and Mixpanel is more product-y and sort of how they do think
about their analysis, but we don't know which one. And then we looked at the APIs for collecting data
and we're like, okay, the APIs are kind of similar for collecting data. But actually, if you squint
your eyes, it's the same data flowing into all three systems. So rather than decide which one we
want to integrate or which one we want to use, which is a business problem, we're just going to
solve an engineering problem, which are the sorts of problems we know how to solve. And we're just
going to send the same data into all three tools by building this abstraction that can send data
anywhere. Like, okay, cool. So we built that little library, it was like 100 lines of code
in hundreds of thousands that we wrote for that classroom lecture tool, and we like set it aside
and forgot about it. And then like four months later, we needed to use that sort of thing again.
So we like cleaned it up a little bit, improved it, and then set it aside. Four months later,
cleaned it up a little bit more, maybe pulled it out as a separate library, and then forgot about
it again. And then we were trying to sell our own analytics tool at this point. And we kept
encountering this objection, which was, well, I already have Mixpanel installed. And so I don't
really want to use your analytics tool. Like, it's too much of a burden to install it. We're
trying to figure out how to like growth hack our way around this. And our co-founder earlier was
like, well, what if we took this library added ourselves as the fourth service that it can send
data to? And then every time someone has that objection, we give them this little library so
that they can install us alongside and give us like a fair shot. And we're like, Oh, that's a
cool idea. And so we did that we took this little piece of code, open sourced it and started sending
it out to people. And they started replying like, Oh, this library is awesome. I'll definitely use
it. And then we'd follow up a couple weeks later. And they would be like, Hey, we saw you're using
the library based on all that sending data to our analytic service, here's the API key, all you have
to do is drop it in. And they're like, well, the library is fantastic. But like, I don't really
want to use your analytic service. Okay. So a few more months go by. And we start noticing that this
little library is getting a few stars on GitHub. And we don't really take that much notice of it.
But it's definitely the first time that something has had sort of this organic pull. And so we
definitely took a little bit of notice of it. And then we had this conversation with PG.
We sort of sat down and Calvin and Ian had put together this manifesto, which was, we think
there's a big business behind Analytics.js, which is this open source library that routes data to
different analytics tools. And I was like, guys, this is the worst idea I've ever seen. This is
four or 500 lines of code. It's already open source. Like I in no way see how this can turn into a
like significant business opportunity. And we argued about it all day. And I went home and I
was like racking my brains. Like how do we how do I kill this idea? Like this is this is the worst.
I finally figured out I came in the next day. And I was like, all right, guys, here's what we're
gonna do. We're gonna build it like a really amazing landing page. It's gonna be really
beautiful. I was like really trying to sell them on this. And then we're gonna put up this landing
page on Hacker News. And we're gonna see what happens. And I was thinking like, this will be
this will kill it off. And we can like go find something. So we built it, I think it maybe took
us a day or something. And we put it up on Hacker News. This is on December 12, 2012. 12,
12, 12. Still remember the date. Yeah, it went straight to the top.
It's very humbling experience for me. Peter, come up with the perfect way to validate it.
Yeah. And it was like thousands of email signups, thousands of stars on GitHub. People are reaching
out to us on LinkedIn demanding access to this thing. So the whole thing blew up in like 36
hours basically. We sat down and we were like, well, I guess we found product market fit. So
let's build a hosted version of this library because it's kind of a pain in the ass to use
as an open source library. And so we built that over the next five days and launched that on
December 17. And then, you know, we had this giant waiting list. So by December 30, we had
about 70 companies sending maybe 20 events per second through the system.
That's amazing. And I really like how you you mentioned earlier that you're going to people to
get them to use the product and they were complaining, you know, they were saying,
we're already using mixed panel, we don't want to use it. And your instant reaction was, okay,
well, how do we growth hack our way around the problem? And I think that's the reaction that
a lot of founders have, which is, I think somewhat contrary to like the YC mantra of
make something that people want, right? Where ultimately, if you make something that people
want, you don't have to growth hack your way around any sort of objections, right? You put
it up on Hacker News, or your target audience, wherever they are. And they love it because it
clearly solves some problem or it does whatever it is that they really want to do. What was your
next step after that? Did you end up raising money? So then we spent about a month just sort
of tidying up the product, we built some like dead end things, like we tried building a dashboard
for PG for YC companies to share their growth metrics with PG. That was a bad idea. He never
looked at it. And so we had some like dead end product experiments. But eventually, we ended up
building a bunch more ways to collect data. So like iOS and Android SDK is ways to collect data
from the server side, we only launched with JavaScript data collection that could send data
to like seven tools. So within a few months, we had maybe 10 different places that you could send
that you could collect data from like libraries. And we could probably send data to like 30 tools
or so. So we sort of expanded both the sources of data and the destinations of data. And then
we really we launched on TechCrunch. And then we started focusing on marketing. So we built
Analytics Academy, which was like a pretty heavy lift in terms of writing good content about
analytics and how to get started with analytics. That did well and was like pretty important,
I think in terms of building the brand and starting to drive early adoption. And there
was a lot of organic adoption of just like word of mouth starting to happen as well. So by June
or so of 2013, we had maybe 1000 companies sending data or 1000 customers sending data
through segment, we still weren't charging for it. And then at that point, we had maybe like 60k
left in the bank or 50k left in the bank. So we were getting we were pretty much running on empty.
How big was your team at this point?
It was still just the four of us.
Okay. And who's writing all this code? Were all of you guys coding? Because you guys are
releasing product after product? Yes.
Yeah, at that point, yeah, pretty much all of us were coding. Peter was doing a bit more of the
like analytics Academy content. We're all kind of writing blogs as well to figure out what things
can we deliver that we've learned that are actually really useful for our audience to understand,
like what sorts of internal code can we put out there as open source, kind of to help make more
and more people aware of segment. Okay, so you guys are basically four of you guys are all
contributing. You've got 60k in the bank, your product seems to be catching on and you've got
1000 companies using you for free. And at this point, I think, you know, you've probably got a
few things running through your mind, which is, thank God, it's something is starting to work.
But also we're running out of money, you know, should we try to charge people or should we try
to raise more money? Which decision did you guys make? And what led you to that decision?
So we were pretty solid in the camp of like, let's try charging people. But we were also
terrified of asking for money. So we just started asking people for like ridiculously small amounts
of money in retrospect, we're like, okay, we should charge like $10 a month for the service,
which is just like, respect. But we were just like terrified of asking for money.
It took us about I was gonna say that we we put up this like public pricing page with the $10
a month for analytics routing. And we were really nervous at the time that was going to set
expectations there. And I remember, Peter and I actually had a manual stripe script where we'd
go through each one of our customers and we'd like, we hadn't set up any form of subscription
building. So at that point, we just had to run this script once a month and charge them and just
hope that no one got upset at this $10 a month charge. Wow. I mean, we had one customer that
messaged us. I'll never forget this guy. I think Eduardo is a customer from Brazil. He actually
messaged me a few months ago, he's still a customer. And he messaged me and he's like,
Peter, I'm very concerned about your pricing. And we're like, oh, shit, we're charging too much.
And he's like, it's far too low for this to be a sustainable business. And I really want to use
segment for a long time. You really need to find a way to charge me more. So it took us maybe like
a year to really like ask people for reasonable amounts of money for the value that was being
offered. Was this the first point in your entire process where you guys had charged for any of the
stuff that you made? Or had you charged for the classroom tool as well at some point?
Yeah, no, no, we had never charged for that. Okay, so you guys put up this pricing page.
How did people respond? Was it enough to kind of pull you guys out of back from the brink of
death? Or did you also have to raise money in addition to that?
It definitely was not enough, because we weren't asking for enough. So I think we got to maybe
like $1,000 a month on revenue or something like that. And then, which I guess was a big deal at
the time. But then we're like, the only way this works is if we raise around. So we raise $2 million
in seed funding, we're like, okay, we're gonna build out a team, we're gonna hire someone to do
sales, and we're gonna figure out how to actually sell to real companies here. Because up to that
point, we had been mostly working with small startups, but we were starting to see larger
companies sign up. And we were like, okay, well, actually, maybe this is a problem that gets worse
as you get into larger companies, and larger companies are more willing to pay. So we should
probably start thinking about building out a sales effort. So we did, we hired, started building a
team. First salesperson, Raf, joined and I was sort of his sales engineer. And over the course of
that first six months after he started, so the first part of 2014, we basically went from zero
to a million in annual revenue, the first half of that year and the second half from one to 2.5.
So that 2014 from a revenue perspective was sort of the breakout year.
That's humongous. Were there were there any lessons that you guys learned,
you know, going to making like your first hundred thousand dollars in revenue that,
you know, stopped applying when you were trying to grow from one million to two and a half million?
Because I'm aware of I talked to a lot of founders at various stages. And the things
that are important in one phase are very often not important in another phase and vice versa.
Was there sort of a stair step approach that you guys had to take? Or was it kind of smooth
sailing all the way up to that amount? Yeah, I think I think Jason Lemkin does
or is also known as SAS. There does like a good breakdown of sort of like the stages of revenue, at least
from what we saw. So like zero to a million in revenue was just sort of like a mad scramble
where what we really had to figure out was how to ask for a reasonable contract size given the
value that we were providing. And frankly, it's like, it's a little bit of a ridiculous process,
I think, to like train someone with an engineering background like myself to ask for that.
So we had the sales advisor, Mitch, who basically we went to one of these sales meetings, and he's
like, you have to ask for 120k a year. Keep in mind, like the most we had been thinking before,
this was $120 a year. So this is 1000x more. And I was like, dude, I don't know if I like,
that's crazy. Like, I don't know if I can do that. And he's like, Peter, if you don't ask for
$120,000 a year, then I quit as your advisor. Wow. And I was like, well, I guess I'm asking
for $120,000 a year. From one customer? From one customer, yeah. So I asked, and he negotiated
me down to $18,000 a year, which was pretty embarrassing, but was also two orders of magnitude
higher than what I expected. So there was that. And, you know, basically, it was like a six-month
period of us asking for increasingly larger amounts of money until we stopped turning beat
right when we would ask for it. And then people were like, yeah, it is that valuable to me. I'm
happy to pay it. That's insane. Do you remember their reaction when you first said $120,000?
Were they taken aback? Well, I turned beat red, so then they just sort of gave me a comical look.
I think one of the biggest challenges for a lot of people listening is that a lot of us are
developers, and our primary skill set is writing code, right? So we know how to build a product,
we know how to keep it running, we know how to write tests. But once we start getting into these
areas of sales and marketing and these other things that seem like intangible dark arts,
they're scary, you know? And we don't really know how to take the first step
into getting good at those things. What did you do besides having advisors that taught you these
skills? Yeah, I think it's actually not just that they're sort of mysteries. I think there are some
strange idealisms in the developer community around whether things should be free or how
much they should cost. And the costs are often like... I think there's a lot of just sacred cows
that often come along that really end up being a pretty significant hindrance in actually building
a business. When you are setting pricing in motion, it doesn't matter how much it costs
to provide the service. All that matters is that you bill significantly less than the value that
you're providing, because then it's a totally reasonable transaction for the customer. It's
super helpful to them. It's like massive net positive. So your costs are kind of irrelevant,
but I think because on the engineering side, you're mostly thinking about cost and you're
mostly thinking about how the thing itself functions. It's a shift in perspective to take
the perspective of the customer and what value you're providing them and sort of their whole
view on the transaction. So I think there are just like some strange sacred cows that end up being a
hindrance to building the rest of the business that are sort of superpowers on the engineering
side, but a hindrance on the other parts of building business.
Yeah, I can agree more. And I like what you said about taking the perspective of the customer,
just because that's so crucial in pretty much every phase of running a startup. When you're
doing your classroom tool, if you guys had known somehow magically what every person wanted and
you were able to get into their mindset, then you would have built a much better product.
The same thing applies to not only building product, but sales, how much to charge, et cetera.
Yeah, I think the major perspective here that we had or the perspective shift,
at least in my mind, was we started out thinking of ourselves as primarily a developer tool,
where it's something that you start using segment and it's much easier to hook these things up or
you don't have to worry about integrating new tools for your data any longer. But when we
talked to companies out on the market who are using segments, they saw it as saving
their developers time. Internally, they have a team of maybe 10 or 15 people who are maintaining
this giant data pipeline. And they're spending a bunch of cycles on that. And those 10 to 15
people end up being really expensive, particularly if they're highly trained software engineers.
And so to most of our customers, they don't think of us as a developer tool, but just a service that
we provide, such that they can now take those 10 to 15 people and start them on something that's
much more ROI positive for the business. It's such an amazing insight to start thinking that way.
I mean, I talked to Patrick McKenzie, better known as patio 11, and another guy Brendan Dunn
on the podcast months and months ago. And they were both working as contractors or consultants
at some point in their career, which I have also done. And I charged like a pretty decent rate or
so I thought at the time I was charging like companies 125 an hour to work with me. And these
guys were charging like multiples of that just because they understood that they will look at
things from the customer's perspective. And they could calculate exactly how much value they are
providing how much money they were making these companies that they worked with how much money
they're saving them. Whereas I was looking at myself as just a good developer. So I priced
myself relative towards other developers. And so I wasn't able to charge nearly as much.
How did you guys come to understand the value of segment? And how did you end up having that switch
in your mind from, okay, this is just a developer tool to, okay, here's how much money we're going
to save customers. I think the first part is, you sort of have to test the limits of what the
customer is willing to pay. Customers aren't stupid, right? Like they're doing the analysis
on their side to decide whether it's worth it or not. Like you can you can sort of test with your
with your pricing, especially if you're doing more, like mid market or enterprise sales,
where you really are sitting down with a with a customer and talking talking it through with them,
and coming to a reasonable price. Every conversation we would try double the price that we had tried
previously until we like hit a ceiling. And it was very obvious that like, okay, we hit a ceiling,
like, we have exceeded the actual value or a reasonable return on the value that we're
providing. So that gives you some numerical cap. And then you can also just talk with customers
and understand like, when you use segment, what value does that provide you, which sounds weird,
it's so bizarre to like, flip the conversation and like, ask a customer or a prospect, like,
what do you find? Like, what do you value in the product? Because usually you're like pitching,
pitching, pitching, but to flip the gears and like, have them tell you why they find it valuable,
you then hear in their words, like, well, the thing I find valuable about this is that like,
these three engineers are going to do this other thing for these four months, and then I won't need
this other I won't need to hire someone to maintain this ongoing. And you're like, okay, well, that
just gave you your ROI calculation, right? Those three engineers are going to do something else for
four months, what's the value of that and this ongoing maintenance person. So you, by sort of
understanding how they think about the value, and then testing the limits, I think you can,
you can pretty accurately narrow in on what the price should reasonably be.
That's really cool to see like the difference in what you guys were doing,
kind of this like pre and post product market fit phases of your company or beforehand,
you're so focused on, you know, iterating and changing your product and trying to find something
that people like. But once you hit on that, then it turned into kind of a process of doubling your
prices and talking to customers and estimating how much your product was worth. And you're kind
of sitting on this rocket ship that just needed to be fine tuned in a way. But also, I mean,
you guys obviously haven't stopped developing segments of product, how did you guys balance
product changes and actually improving what what it was you guys are building with efforts like
sales and marketing, which don't really affect your product and just affect your positioning in
customers minds. Yeah, I think we we never really thought about those two as a trade off. So we
never really, I don't think we really thought about it as like, well, we could either develop a new
product, or we could invest in sales and marketing. I think once you have product market fit,
you can sort of look at sales, sales and marketing as a unit and say like, well, if we invest more
heavily here, then what, like how much revenue will that generate? And then separately on the
product development side, you can say like, well, do we have the resources today to just sort of
solidify our existing product? Or do we have the resources to sort of like pursue building a new
product? So I sort of see them as independent tracks. And I think, as a result of that, we
actually started building out of sales and marketing team much earlier than, or at least a
sales team much earlier than most companies do. Raph was our basically first hire, and he led
sales. So, you know, from the fifth person in the company was a salesperson, which I think is a
little bit unusual, at least for companies founded by young engineers. So I think that team, that
sales team just sort of grew with customer demand. And then meanwhile, we sort of separately built
out the product development and an engineering org at sort of the pace that it could grow. We
never really saw this as a trade off. We have built a number of sort of significant expansions
to the product every time. The most significant one was probably warehouses initially, where we
sort of realized that one of the most important destinations for people's data was their data
warehouse. People wanted that data in Redshift and BigQuery and so forth. So we launched that
maybe like a year and a half or two years later. And that was big in that it like pulled us up
market and really sort of expanded the value that we were providing to customers in a significant
way. And then also sources, so massively expanding the number of places where we could pull data.
We could pull data outside of people's products and starting to pull data from the cloud services
that they were using like payment systems like Stripe or Zendesk or Salesforce or so forth.
So the product has expanded in a number of different ways as we felt that we have the
sort of bandwidth to expand it. But it's always been two sort of separate investment engines,
if you will. It sounds like you guys were really spending a lot of time learning from your
customers. Like your success was feeding into the ideas that you're able to come up with
for what products to provide that your customers would actually like. But a lot of people aren't
at that phase yet where they're so successful that they have paying customers who said,
yeah, I'll pay you more for this feature. It is dramatically easier to find your second
product market fit than your first one. Because your second product market fit, you have a defined
customer base. And you go to them and you've solved the problem for them. As soon as you
solve a problem for someone, they will tell you all their problems. Because they're like,
well, maybe you can solve this or this or this or this. Apparently you solved something. So
they're excited to tell you about everything else anyways. So it's much easier to find product
market fit repeatedly after you find it once. Were there any mistakes that you guys made or
things that after this point where things started to take off that if you go back,
you would do it completely differently? Really, it's hard to say if we do anything
completely differently. The one thing that I wish we had done a lot more of a lot earlier was kind
of written everything down. I think we have a decent culture of writing it segment now,
whether that's on the blog or on our internal wiki, or just kind of recording things as we
learn them. But earlier on, we were just sort of slap dash like moving really quickly, kind of
throwing stuff together. And we didn't have enough time to write down all of our assumptions,
why we were building the things that we built and in what way it made sense to build them.
And I think that hurt us over the long term because we didn't keep in mind which assumptions
would change. Like in the case, suddenly there's this new product Redshift that's out on the market.
For instance, we built initially the ability for customers to just instantly spin up a Redshift
instance, which is this data warehouse. And we would manage it all for them. Over time, as more
and more customers started adopting the Redshift instances, we realized that simultaneously,
it wasn't that great for us, because we would continually have to tweak the sizing of these
Redshift instances, we had to view into all of their monitoring, but we'd have to kind of
babysit them. And on the customer's end, it wasn't good either, because they had no window into what
that Redshift instance was doing. They couldn't see any of the metrics, they couldn't see the
sizing, they couldn't see the data loads. So eventually, we said, Oh, our assumption that
what customers really want is just sort of some like turnkey Redshift instance here,
isn't actually the reality, they'd rather be able to have the customization of managing it
themselves. So in that case, we kind of flipped our assumptions, but it would have been much easier
to do had we written that all down and made them kind of explicit from the start.
Is there any sort of startup philosophies or schools of thought like the lean startup or
things that Peter Thiel's written in zero to one, or crossing the chasm? I mean, there's all sorts
of startup movements and books and things that people preach for how you should run a startup.
Is there any one in particular that guided how you guys grew your company?
I think the synthesis business books where they have like a point of view,
where they've sort of synthesized across a bunch of research that they've done or whatever.
I think those are helpful. I think there's like a bottomless stack of those books to read.
I found a lot of the biographies of some of the like, great founders to be possibly more helpful,
because the story helps it sort of stick and see how these principles really play out.
Made in America, the story of Walmart's founding, I think is pretty fascinating.
Elon Musk's biography, Who Says Elephants Can't Dance about IBM sort of getting reinvented,
soft war about the rise of Larry Ellison. These sorts of books really like those stories help
solidify what things really look like in practice, maybe a little bit more than
just focusing on like the ideals from sort of pure business books that are
a little bit more synthesized, but a little bit more abstract.
I feel like startups are this interesting class of problem where you can get the same advice from
10 different places. And like with YC, they plaster it all over the shirts that you get
the first day, there are stickers, they all say just makes something people want.
You can get that advice and it sounds so simple. But until you've screwed it up enough,
and done enough failures yourself, it's really hard to internalize what that means until you've
both failed, and then subsequently succeeded at it. And so I think it's sometimes tough to get
kind of traditional advice from these books without kind of going through those motions yourself,
and actually failing at them once and then succeeding at them after the fact.
But I do think they kind of help sprinkle in ideas, particularly now as we've grown into
now as we've grown into a little bit later stage, just following what other companies are doing and
how they're operating and how they're running. And at least in our case, I find we try certain ideas
from those and implement them. Peter recently was talking to a few of their founders and had this
idea for product pitches, the general idea that people kind of pitch what products they want to
work on every six weeks or every two months. And so we've adopted some of those sorts of ideas
internally based upon what we hear. But I think it's really hard to take those sorts of lessons in
the early stage when really the whole world is kind of your unknown universe, as it were.
I saw someone, someone told me recently, they sent a tweet, actually, and I think they said
that if startup advice is short enough to fit inside of a tweet, then you can guarantee it
will be misunderstood and misinterpreted. And I think make something people want is like a perfect
example. Because I went through YC too. And I don't know how many times I heard that and then
went off and made stuff that people did not want and didn't really catch it until months and months,
maybe even years later. We did exactly the same thing. I think it's funny how you realize what
kind of all the YC partners were saying only, or at least in our case, it was only after the fact.
It was like a year and a half after we'd been through YC, we're like, oh, that's what they meant.
What is it you think is so hard to internalize about this advice? And how could,
I'm kind of putting you on spot here, but how would you try to explain it to a new founder
or somebody who's just starting out and might not fully understand it?
I tried to explain this whole concept at the YC Stanford class with a lecture on product
market fit. But I think the problem for me personally was I didn't actually know what
product market fit felt like. So I didn't know when we have it or didn't have it. And basically,
I only had, if you imagine you're trying to train a machine learning algorithm in your head,
and I only had data on the negative side. This is not, and this is not, and this is not.
And you're like, well, it doesn't really give you a boundary. If you're trying to train a
classifier, this is a really bad data set. And so once we had one success, and once we sort of
knew what that felt like with launching analytics.js and sort of feeling the product market fit there,
it was like, oh, and all of a sudden the classifier got much stronger. In that presentation at Stanford,
I was trying to help people just understand what it feels like to train product market fit,
so they could maybe have at least by proxy one data point on the other side of that classifier.
I think that's the hard part, is it's a very sparse data set, and it's almost entirely one-sided.
Yeah, so I just want to talk about the future of Segment 2, because I think it's kind of a clue
into your psychology and where you guys are going with this. What are your goals as founders,
and how have they changed over time? Because I'm sure there are things you guys are thinking about
now that you probably didn't even consider four or five years ago.
Yeah, I think we have a reasonably well-developed sort of vision and product strategy for where we
want the product to go in the next, or we think the product should go in the next, let's call it
five, maybe 10 years, based on all the conversations and learnings that we've had with customers about
what problems they have and how we think we can solve them with Segment. So we're doing a major
product launch in October, so I can't sort of totally talk about what's coming. But what matters
the most to me is executing against that product strategy and solving a lot of the problems that
customers have with data, with customer data, with actually being able to market and support
with their customers effectively, and sort of solving those problems for them with the products
that we're building, and basically building out the vision that we set out over the last few years.
That's what matters the most to me, and I think there's a certain size of company and funding
structure and so forth that people probably talk about more in the press that supports
that sort of impact that we're trying to have with that vision. But I think what ultimately
matters to us is less like, oh, we're going to IPO or whatever, and it's more like what really
matters is building out that vision and bringing it to all the companies that really need it.
What about your personal goals as founders?
I think for me at least, what I've appreciated most about our journey at Segment is just having
kind of a front row seat into the life of a startup. At least I know Peter has mentioned
this on a couple of different occasions, but he likes to say that every startup has their own
totally unique journey and no two are really alike. So I think that's part of the reason
that these business books are so fascinating is because when you're figuring things out and
defining a new company structure and defining what the org looks like or what the priorities
look like, there's no real right answer. And so that discovery has been at least incredibly
rewarding for me, and I'd like to continue seeing it forward kind of over the coming years.
What about you, Peter? What are your personal goals?
I'm not sure that my personal goals differ very much from the company goals.
You've become Segment.
Yeah, I think founders get pretty deeply tied up in terms of their identity with the company.
For better or worse, I've seen companies get acquired where this basically...
Well, it has a very destructive impact on a founder, I think.
When I've seen founders have companies get acquired.
Yeah, I mean, I totally understand that. And I think, in a way, having your goals
so aligned with that of the company is a tremendous advantage. And it's why you
should probably do something that's at least somewhat interesting to you.
Otherwise, five, 10 years down the road when you're still working on this business,
you might hate your life.
Totally.
It's kind of funny because you initially wanted to kill the idea and now you're the CEO and your
identity is all wrapped up in that. What changed besides just the success of Segment?
I think I didn't understand what it actually was. So the initial idea was,
well, there's this open source library and it's a really good abstraction for sending data to
different analytics tools. And I was like, that's just not that exciting. How many analytics tools
are there? How many analytics tools is the company actually going to use? Who cares?
And then it turns out that, well, the data that we're collecting is not just useful for
analytics tools. It's also useful for email marketing tools and push notifications and
help desks and CRMs and security and fraud tools and data warehouses and advertising
conversion pixel. I could go on and on. This is just like this data about what
interactions a customer is having with a company are deeply important to every aspect of the
business. And that's a totally different proposition. So I think we launched this
thing not really knowing what it was, to be honest. And then as we saw people start
using it and learn from them, from the customers, like what it actually represented to them and
what it could become, that became far more exciting. I think it took us, I'd say actually
we had so much downward momentum in our morale. I think our morale as a founding team bottomed
out like a month after we hit product market fit. And then it started coming back up as we realized
sort of what was actually happening. And yeah, I think the year or two after that,
of sort of the journey of actually understanding what problem we were really solving for people
gave us a ton of motivation to actually go like really invest a lot of time in solving it.
Yeah. And that morale is extremely important. Anyway, I think that's a really good point to end
the interview on. I know you guys got places to go and things to do, but I really appreciate you
guys coming on the podcast. You had some really awesome insights. And why don't you tell people
where they can go to learn more about you guys personally and about Segment as a company?
Yeah, you can learn more about Segment at Segment.com. And my personal blog is REIN.pk.
Yep. And my personal blog is C-A-L-V dot I-N-F-O. But I write mostly for the Segment
blog these days. So go check that one out. All right. Thanks, guys.
Thanks for having us. Thank you.
If you enjoyed listening to this conversation and you're looking for a way to help support
the Andy Hackers podcast, then you should subscribe on iTunes and leave a quick rating
and a review. It only takes about 30 seconds, but it actually really helps get the word out.
And I would personally appreciate it very much. In addition, if you are running an internet
business or if it's something that you'd like to do in the future, you should join me and a whole
bunch of other internet entrepreneurs on the AndyHackers.com forum. It's basically a community
of like-minded individuals, just giving each other feedback and helping out with ideas and
landing pages and marketing and growth and other internet business related topics.
That's www.indyhackers.com slash forum. Hope to see you guys there.