Stewart Butterfield on the famous Slack pivot

In a conversation with Guy Raz (transcript), Stewart Butterfield tells the story of how Slack was born. Besides the great insights on product-market fit, it is also an amazing plot — full of spectacular ups and heartbreaking downs.

They started out around 2009 and the intent was to create a multiplayer online game. The conditions seemed perfect:

Guy: So what did you do when you walked out of Yahoo? What was your plan?

Stewart: There wasn’t an immediate short-term plan. But it didn’t take long for a lot of the same people who worked on Flickr and had worked on the previous game to decide that we wanted to work on the game again.

Guy: You wanted to go back to gaming? You didn’t learn your lesson the first time around?

Stewart: Yeah, we apparently didn’t. I mean, so the world looked pretty different at that point. So now it’s the beginning of 2009. There was at least an order of magnitude more people online, whereas, in 2002, the majority of people didn’t have Internet access at home. Now the majority did. Computers are way faster. All the hardware was way cheaper. Online games were something that was really popular. And so we figured, like, oh, this time we can’t fail.

There is — you know, all of the conditions are perfect, so we should do this.

There was me and three of the engineers, and what we decided to do it, which was called the Glitch. It was completely different than anything that anyone had ever seen before.

This game — it was a bizarre, fantastical world — really tried to encourage individual creativity. People could create stuff inside the world. You would milk butterflies in order to get butterfly milk, and eggs grew on trees. And the look was kind of Dr. Seuss meets Monty Python meets modern-day graphic novels. As you moved around the world, the look changed dramatically.

Guy: So you had a track record with Flickr. Like, you already had a reputation. You’d been on the cover of Newsweek. You’d worked for Yahoo. So when it came time to raise money, was it pretty easy?

Stewart: Yeah, absolutely. It was very easy for us to raise money. And so we started off with you know, a million and a half dollars. We were able to hire someone. We were able to afford all the technology we wanted.

As they began developing the Glitch, they faced a crucial point: there was a niche of people very interested in the product, but, despite a lot of effort, they never managed to improve their metrics with a broader audience.

Stewart: And as we started developing the game, we had a bunch of really positive early indications. So we charged people money, and they were — they paid a lot, you know? Like, the average person who paid was paying $70 a year.

Guy: And this is, like 2010, 2011, something like that. And people are already paying for it. It was that good?

Stewart: Well, it was that good for a very small population. In fact, it was really hard for us to get people even to go through the first few minutes of the game because it was just so different and so weird. Most people who tried it were — what the hell is this? And just pass out in the first three minutes of gameplay.

Guy: All right. So you guys launch this thing. There’s some early success. You raise lots of cash. And then November of 2012, you shut it down. What happens?

Stewart: Well, being in business, and I think especially being a CEO requires a lot of unnatural optimism.

And at some point, that optimism was exhausted. We had what’s called a leaky bucket. People would come in the top of the funnel. And the funnel is kind of describing: first people hear about your thing, then they go to the website, then they sign up, then they in our case, play the game a little bit, then they end up paying you. And in each of those stages, some people fall out of the process. So, that’s why it’s called the funnel. The leaky bucket is when you get people in the top, but they just fall out. They fall out the process too early. Not enough of them make it all the way through.

Guy: They didn’t stick with it.

Stewart: Yeah, they didn’t stick with it. And it was always the next thing that was going to fix it. Like, the next game dynamic we added, the next bit of customization, the next thing, the next thing, the next thing. But as we continued to try those things, we just never found that magic formula that would make it work economically.

It would have been a fine, what people call, lifestyle business. But it was never going to become the kind of business that would justify $17.5 million of venture capital investment.

This is a very important learning: fixing leaky buckets is extremely hard. This matches with my own experience building software products for consumers. There is also the good reminder that, in the context of venture-backed business, you aim big and try to get there as fast as you can.

Back to Raz and Stewart:

Guy: So at what point did you come to the decision to shut it down?

Stewart: I was losing a lot of sleep in those days. It was like 2 or 3 in the morning. And I was in bed. And I hadn’t been sleeping for hours. And I just realized, like, I don’t believe this can work. Like, I don’t believe it anymore.

There’s a saying — if you’re thinking about firing someone a lot, you should just fire them. That intuition — if that keeps coming up, it’s almost certainly correct. And you wouldn’t be thinking that all the time if there was a real shot at making the relationship work.

I think it’s exactly the same thing with a business. Once I began not to have little doubts, but once I hit a fundamental level, I was just, like, I don’t think this is going to work. It’s not going to work. So, you know, first thing that morning, I wrote to all the co-founders and then to our board of directors and just said, it’s over. And that definitely was not a unanimous point of view. And there was a lot of contention… And argument, and… Because we had developed an enormous amount of kind of equity.

Guy: Content — you had content, tons of content.

Stewart: Yeah, millions of frames of animation, hundreds of hours of original music.

Guy: So how did you break the news to the team? I mean, they must have been - it must have been excruciating.

Stewart: Yeah, it was a horrible experience. And I say, we’re going to have an all-hands meeting. And everyone, you know, files in and gets together. People have their coffee in the morning, and they’re chitchatting. And finally I stand up. The meeting starts, and I start to tell them that we’re going to shut down the game. Before I could even get the first half of the sentence out, I was crying.

You know, almost everyone in the room I had personally convinced that they should come work in this company, that they should accept our stock options, that they should believe in the project, that they should believe in me. It’s humiliating. There’s a real sense that I had failed all these people…

That I had an obligation to them. And I was — I had locked eyes with this one software engineer who had just three months before moved. He had an infant daughter, maybe 6 months or a year old. He was moving away from his in-laws, who were helping take care of the kid, moving to a new city with his whole family, and now I was telling him he didn’t have a job anymore.

I will now fast-forward the story several months because, here and now, I am focused on product-market fit.

Once you have the time, you should listen to the whole episode. He tells how they returned to a very small team, how they decided to transform the communications tool they’ve used internally into a new company that became Slack, how difficult it was to sell other people and companies on Slack in the early days, how their distribution strategy of enterprise software was very unconventional (and hard for VCs, like a16z to get their heads around it), and so on.

Back to Stewart telling what were clear signals of Slack’s product-market fit:

Stewart: Well, so by the time we officially launched, it was evident that this was going to be something. The leaky bucket problem that I talked about in the context of the game was completely eliminated.

We found that once people started using it — as hard as it was to get people to start, once they started, they almost never stopped. At that point, people weren’t paying us. They were just using it for free.

But we could see that they were getting utility out of it. They were logging out at the end of the day, and first thing in the morning they were right back in there.

Guy: So you launch in February of 2014 officially. It goes out into the app store or whatever and into the world?

Stewart: Yep. We also told other people who had been using it for free, like — hey, now you’re going to have to start paying. We’ll give you a nice, healthy credit to thank you for being one of our early testers.

And we found the conversion was excellent. You know, within maybe two or three weeks of launching that officially, we had sold a million bucks’ worth of Slack.

It costs $80 per person per year. So you would pay — you would only pay for the people who are actually using it. So if you have a 50-person company and there are 20 people using it, then you just pay for 20 people.

We were off to the races. Unlike almost any enterprise software ever, people would talk about it. Like, they would be in line at the coffee shop, and they would say, oh, my God, you’ve got to start using Slack. It’s amazing. It changed my life. And they would post to Twitter and say, like, I — you know, I recommend it. And that — you know, no one ever says that about the software that they have to use at work.

Guy: It’s amazing. By October of 2014 — so this is just two years after you shut down Glitch and, like, have to break the news to all these people — you raise 120 million bucks. It has a $1.2 billion valuation. I mean, that’s nuts. That’s totally insane.

Stewart: Yeah. It was completely insane. And, you know, at that same period, we’re still growing, like, somewhere between 5 and 10 percent a week.

What a ride.