Peter Reinhardt, CEO, Segment: Finding Product Market Fit
Peter Reinhardt, CEO and Co-Founder of Segment tells the story of Segment’s journey to finding product-market fit, a tumultuous ride at best. On several occasions along the way, he had convinced himself that they had found it only to realise they hadn't quite yet. When they finally reached the sweet spot, they just knew. How did that product-market fit look and feel that convinced them? Watch the video or read the edited transcript below.
Many of you here today are SaaS founders, in the earliest stages of building a company. But the reality is that 80% of startups don’t find product market fit. That means that 4 out of 5 founders are not going to find product market fit and it means that the other 1 out of 5 is going to struggle through it and barely find it.
But that’s just the stats. The reality is that searching for product market fit is this crazy emotional free fall. You can check yourself into it; you can see mirages of it, it can make you sick.
In 2012 at the peak of our search for product-market fit I lost 10 pounds in three weeks, and I was hospitalised twice for panic attacks. I haven't had panic attacks before, or since, so just to give you a sense of the sort of emotional roller coaster that at least we went through in finding product market fit. It can be quite intense.
And we mislead ourselves in the search for product-market fit. This is a picture of us in 2012. That’s me and my co-founders Ilya, Calvin, and Ian. We convinced ourselves that we were always on the cusp. We were always on the cusp of figuring out what the detail was, what the thing was that would suddenly trigger us and flip us over into finding product-market fit.
But the reality is that we really struggled to find it between June of 2011 and December of 2012. So a year and a half of misery, which many of you maybe are going through today. I hope not.
Of course it got better from there. I wouldn’t be here today if it didn’t. We're now in San Francisco, a little over 100 people growing very rapidly. But back then, it was just the four of us in an apartment, coding a lot, had no customers. It was pretty sad.
The reality was that the only thing that got us through was that we were a bunch of cockroaches. What I mean by cockroaches is that we just figured out how to survive. We kept our burn very, very low. And by keeping it low, we just powered through. It gave us the time to test not just one idea but two product ideas, three product ideas. Some companies do four or five. It's very important to keep your burn super, super low.
We're going to break down this product market fit process into three stages. The first is search. You're building, testing different ideas. You're putting them out there. And during that time you want to keep your burn super, super low so you have time to test the one, two, three, maybe four, maybe five ideas.
Then something magical happens. All of a sudden there's this magic poof, and you discover product market fit, which we'll go into quite a bit in a minute. And then you're suddenly transformed into a sort of uni-roach or soonicorn, and everything gets easier. Not easy but easier because all of a sudden you're selling a problem that people care about. It means that people want to join your team, it means you have customers that show up and want to buy your product. And so for the first time, you feel like you have a momentum. You're not just pushing this boulder uphill.
Today we're going to talk about finding product market fit, and we're going to break that down into three things. First, we're going to talk about building category-leading businesses.
What is category leadership and how can we deconstruct that into something a little bit more achievable.
Then I want to share some stories from the early days of Segment around product market failure or product market fit, failing to find it, what that felt like.
And then finally what sort of the magic of product market fit felt like to us, at least in December of 2012. My goal is, in highlighting the sort of feeling of these two things, to give you a sense of how you can winnow out bad ideas and find some of the good ideas earlier on.
Let's start by deconstructing category leaders. Why do we even care about that category? Like who are they? Why do we bother? Well, the reality is that category leading companies are 10x larger than the other companies in their space.
Salesforce is a good example. They're a $50 billion market cap today. $8 billion in annual revenue. Their next closest competitor is something like SugarCRM or Zoho, substantially smaller companies.
The reason that we care and the reason that we're talking about category leading businesses is because they're huge. They've somehow created this massive moat around themselves. It's worth digging in and understanding how do these companies somehow do that? What do they do along the way that allows them to create that big, sustainable, long-term advantage?
If you look at most of them, they've taken their initial SaaS product like Salesforce’s CRM, and all their customer data that's being generated in there, and they've turned it into a platform. And that platform allows you to build an ecosystem of partners and customers that create this sort of moat ecosystem around you that makes it possible to grow to that scale.
But what's required to build out the platform? Peter Thiel has said that to build a platform you need to get to $100 million in revenue. Bullshit. That's a lot of revenue. And why? Before $100m in revenue, you don’t have a big enough market for someone to build exclusively on your product.
If, as a partner, building on top of a platform, if you managed to get a couple of percentage points of the platform’s customers and it's under $100 million in revenue, you can't build a sustainable business that way. But by the time you get to $100 million in revenue and you have line of sight to a billion, your partner will start to truly build businesses based purely on your platform.
We know that to be a category leader, you need to build a platform. To build a platform, you need to go to $100 million in revenue. $100 million in revenue is still a long way away. So how do you get there? If you aren’t reading SaaStr, you should go check that out online.
But the way that he breaks it down is he says from $0-1 million in revenue, is impossible. From $1-10 million is improbable. But from $10-100 million is inevitable. And the reason that $10-100 million is inevitable, at least in his mind, is that by then you have too much momentum not to. You have a mini-brand, many customers who are hopefully happy and loving the product that is growing from $10-100 million is going to happen eventually unless you screw it up. It might take a long time.
$1-10 million is improbable because for the first time, you have real scale. You have real customers; you have real money exchanging hands, you have realistic expectations on your business. But it's still the early team. You can't build a great executive team, a great management team until you get to $10 million in ARR.
For that $1-10 million journey, it's a real grind for the founders and the early team. You're pushing through. You're a skeleton crew holding it together through these early stages of scale.
And $0-1 million is impossible because this is the stage where almost everyone fails, where 80% of companies fail to find product market fit. So today I want to zoom in on this $0-1 million stage. This is the impossible stage to finding product-market fit because it's the foundation of this entire long journey that eventually leads to building, if you succeed, a category-leading company. So we're going to zoom in on that.
It's crazy, right? 4 out of 5 companies not finding it. 1 out of 5 barely struggling through. I mean, that's like a horrific failure rate. It feels like we ought to do better, right?
But the reality is it's even worse than that, which is as founders and investors, we often talk about how we learn from failure. But the reality is the studies that have been done on product market fit and failure; it turns out that for founders that fail the first time to find product market fit they're actually no more likely to find product market fit the second time. It goes from 22% to 23%. That's pretty horrific.
If you succeed in finding product market fit the first time, then the odds go up. The odds jump from 22% to 34%, which is still pretty bad but it's a 50% increase. So what that means is that there's no learning encoded in failing to find product market fit. But there is learning encoded in actually succeeding and finding it.
I know for ourselves we felt this problem acutely. We felt like we could just mislead ourselves based on the smallest amount of positive information. We would see a customer give us some small comment or some little bit of feedback, some vague interest and we'd convince ourselves that this little positive feedback was the beginnings, the beginnings of hopes or something that could grow into real true product-market fit.
What we missed were examples of what product market fit felt like. Some positive example where we could say yeah, that's nice, but that's still not product market fit. We needed some positive example to train the machine learning algorithm in our head as to what it feels like. Because that's what I want to give you today. I want to walk through some of the early history of Segment and hopefully give you a feeling of what these two things are like so that you can differentiate between glimmers of false hope and true product-market fit.
This is a picture of us in 2011. We had just gotten into Y Combinator. Today, Segment is a marketing infrastructure company, but back then we were building a lecture tool. And we had just left MIT and RISD, and we wanted to help out our professors and students. The idea was that we would give students this interface where they could say “I'm confused” and the professors would get this graph over time of how confused the students were. It might be kind of nice right now. I can see how or if I'm leaving any of you behind.
We went around, and we pitched hundreds of professors. We hustled. Most of those professors were disinterested. We dismissed that as technophobia, the ones that get it they know what's up. So we felt like these professors they're being cautiously optimistic that theyreally represented sort of the early signs of product market fit. We convinced a couple to try it out in their classrooms. We'd come up to them right after the lecture, and we would pitch them on the product right as they finished.
But the reality was that in the most cases they were just doing us a favor. They were like, “Yeah, I'll help these students out. I will give them a chance.” That's not what product market fit feels like. That was us just purely misleading ourselves. And to make matters worse, as soon as the students opened their laptops they just went straight to Facebook. That's also not product market fit, in case you were wondering.
This was horrifically embarrassing. And it was super obvious from the way that students used their laptops in class that we didn’t have product market fit. But the professors actually should have been an early warning sign too. Because the professors were just vaguely interested and the few positive interactions that we had among the hundreds of professors that we talked to, that's not what product market fit feels like. No one was in dire need of the solution that we were offering.
We thought we learned from that. We went back to the whiteboard, and we said, all right, we're going to build an analytics tool like Kissmetrics or Mixpanel or Google Analytics, and we're going to launch that. We think that there's something amazing that we can build there. So we started coding.
And true to the lean startup principle, we went out, and we started, got out of the building to try to validate it. We had a few conversations with people who expressed sort of like, yeah, you know, I have problems in analytics - I have this problem with Google Analytics and I have this problem with Mixpanel. So we convinced ourselves that those little discrepancies between what people wanted and what we thought we could offer in analytics was exciting enough that we should go build this product.
In particular, a few people said, “Yeah, I'd love to stay in touch. I'd love to hear how this progresses.” And so we said awesome. They're beta users, right? All our beta list. That's not what product market fit feels like.
Six months later we moved to San Francisco, and we were still coding. I was occasionally talking to sales prospects, and I would just try to exploit those little differences to convince ourselves that we were really onto something big. It's really easy to do because you want to be succeeding. Right? Just convince yourself that, hey, these little positive interactions are really good.
The other examples of these little tiny positive interactions, this is an Olark Live Chat with a visitor to our website at 3:00 a.m. some sad, sad evening when I was probably up late coding. He or she asked a few questions about how our product works or what it is. Like this is great. This is a little positive interaction, right? Someone is really curious and wants to learn about our product. But the reality is that it's at 3:00 a.m. it's a 6-minute conversation, and it ends with me asking a question and them just disappearing.
When we got to December 2012, we've been working on two broken product ideas for a year and a half, and we realized, all right, this is not working. We do not know what we're looking for.
We decided to go back to YC, and this is us outside the YC office in Mountain View. We decided to ask for help. And we're walking around a little cul-de-sac here in around Y Combinator and bringing PG up to date on everything that we had tried to do over the last year and a half. And when we were done explaining it, he stopped, and he looked at us, and he says, “So just to be clear, you've burnt half a million dollars, and you have nothing to show for it?”
That was the absolute pit of our search for product-market fit. It was probably the most brutally honest moment I've ever had in my life. But it was true. It was true we had burned half a million dollars, and we didn’t have product market fit. In fact, we still had $100,000-120,000 left, and that was enough. We said, okay. We get one more shot on goal here. This is our third shot. We got one more chance.
Let's pause there and let's rewind all the way back to the beginning in June 2011, the first week of Y Combinator. In that first week we had said, hey, we should have analytics on our classroom lecture tool. We Googled it, and we found Kissmetrics, and we found Google Analytics, and we found Mixpanel. And we looked at the APIs for those tools, and we saw that they're collecting all the same data, but then they gave us different graphs out the other side. So we said you know, we don’t know which of these tools we're supposed to use. So rather than solving the business problem let's just solve the engineering problem. We're a bunch of engineers. So we just built this little abstraction that could send data, pull data in in one format and then fan it out to all three tools downstream. This was just a little tiny code like a little bit of code in this massive lecture tool that we built. Then a year later, when we were trying to sell our analytics product, we kept encountering this objection which was someone was already using Mixpanel or Kissmetrics or Google Analytics and they'd say, “Yeah, I just don’t really want to install another analytics tool. It's like a lot of work.”
Then my co-founder Ilya had this great idea. He said, “You know, remember that analytics wrapper we wrote a year ago? What if we pulled that out added ourselves as a service, a fourth service that they could send data to and then every time someone comes to us with this little objection, we hit them back with this? We open source, and we hit them back with this open source library.” And so we did that.
And people started replying to our objection handling emails, and they'd say, “Oh, that library is awesome. I'll give that a shot.” So they would use the library and then a few weeks later we'd follow-up and say, “Hey, now that you've installed the library, we'd love it if you would use our analytics service. Here’s an API kit if you want to turn it on.” They're like, “Yeah, whatever, man. I don’t care about your analytics service.”
We got to this fork in the road where, back in December 2012, we had just talked with PG, and he’d been brutally honest with us. We realized we're going to shut down our analytics tool. But what are we going to do next?
And my co-founder Ian at the time was like, “You know, I think there's a really big business behind Analytics.js.” And I said, “That is the worst idea I've ever heard. It's 580 lines of code. It's already open source on GitHub. Like just explain to me for one second how you plan to build a sustainable business around that?” And I couldn’t convince the other guys it was a bad idea.
The next day I came in, and I said, “Alright, guys. Here’s how we're going to test it out. We're going to build a landing page. We're going to put a little email signup form at the bottom and then we're going to put it up on Hacker News, and we're going to see what happens.” And I was thinking to myself like, tee-hee-hee, this will prove them wrong. This is like going to show this is total trash.
I couldn’t have been more wrong. It went straight to the top of Hacker News, got a couple of hundred uploads, we got thousands of stars and followers on GitHub in that first day or two, and we started getting these crazy emails from customers. This is Outreach via LinkedIn. This guy says, “What does a brother have to do to get access to the Analytics.js beta? And I'll give you feedback and tolerate bugs like you wouldn’t believe.”
Full stop. That is product market fit. There is like you can't differentiate these two things. It's not one metric starting to go up. It's this sort of holistic thing of every single metric in your business is exploding.
And when we were on our lecture tool and analytics tool we kept thinking like what are the next features that we should build? What do we build next? But once we had product market fit it was obvious everyone was screaming at us like, “You need to build more integrations,” “You need to build server side libraries.” It was super obvious.
When we were working on our analytics tool and lecture tool, we had written hundreds of thousands of lines of code that no one gave a crap about. Like we deleted it long ago. I don’t think anyone noticed. But now we had built this 580-line code thing that was an elegant solution to a real problem that people had.
And when we had been building the analytics tool and lecture tool we had this grand visions about what we could build and what we could solve for people, what it could become down the road. But with this, we just solved this tiny problem. The tiny problem of I don’t want to install multiple analytics tools. And just solving that teeny-tiny problem in an elegant way ended up being compelling to people.
If you walk away with anything from the presentation today, I want you to walk away with an understanding that product market fit is not vague, positive conversations with customers. It's not glimmers of false hope around some random positive interaction. What it feels like is a landmine going off. That's what the Dropbox founders described it. It feels like stepping on a landmine.
To summarize how difficult this is, I thought that ClassMetric was going to be this amazing product market fit. It turns out the world didn’t give a crap. I thought that Segment.io would be this amazing product market fit, but it turns out that the world also didn’t care. And then for the Analytics.js, I thought that it was a really bad idea, and it turns out that the world had a need for it.
It either goes to show just how hard finding product market fit is, or I guess maybe how obtuse I am. But I think it's really important that if you do want to find a product-market fit if you want to become one 1 of the five founders that do find it, that you do not fall into the same traps that we did. Glimmers of false hope are not the same thing as customers just ripping it out of your hands. All that takes is being honest with yourself.
If you liked this, you will love the 101 speakers we have for you at Saastock 17. https://ti.to/saastock/saastock-2017Get your ticket now.