A Product Market Fit Show | Startup Podcast for Founders
00:17:06 · Chris Slowe
[00:17:06] S0: Yeah, I think that with hindsight, one story we've heard over and over again for users that end up sticking around and becoming Redditors is there's this particular arc that usually goes something like, "Yeah, I noticed Reddit." It kept coming up in search results or I started seeing it in here or I'd ask a question and it would come up. I casually used it for some period of time, usually months to years. Then I found this thing, and then I discovered I was coming back every day. That thing is always very specific to the person. It's like you're a botanist or you're a hobby gardener, and you find a community that's dedicated to your particular variety of gardening and it's like, "Oh my God! I'm going to just going to stick with this forever." Another one that comes up that's also not talked about as much is we actually have really good support communities; so for people who have wanted to stop drinking, for people who have substance abuse, for people going through divorces, those can be very sticky because there's not really a place anywhere else where you can go in and be honest and talk about your actual problems and not worry about it coming back to you as a person. I mean, that's the magical model that we picked early on was we decided that we didn't want to have real names, we didn't want to have association with you. We just relied on the fact that if you're going to waste a bunch of time creating a fake user and a fake persona, if it's entertaining, great, otherwise there's no motivation to go and create a fake persona because it doesn't come back to you. It actually cuts both ways. Even though we are pseudo anonymous, the incentives are such that authenticity is what comes out of that because there's no reason to try to be an influencer.
00:18:46 · Pablo Srugo, Chris Slowe
[00:18:46] S1: That's right. I mean, unless you're trying to troll like that, that's always the only reason I can think of to, to create like a fake, fake persona. Yeah. [00:18:53] S0: Yeah, that seems a big waste of time for very little return. The other standard approaches, of course like an opera, we deal with, and again this is a tales of time on Reddit, spam. Trying to spam, trying to sell, pushing pills, that's the cornerstone of the internet. You can either be identified as spam or you can go through the front door and pay for advertising, but there's that in between. It was just like, yeah, you're just wasting everyone's time and efforts. We'll just get rid of that content. [00:19:19] S1: In those early days, like 2005, 2006, how much was it just straight organic word of mouth, and how much did Google and Google results play a role in growth?
00:19:19 · Pablo Srugo, Chris Slowe
[00:19:19] S1: In those early days, like 2005, 2006, how much was it just straight organic word of mouth, and how much did Google and Google results play a role in growth? [00:19:30] S0: Very early on it was almost entirely word of mouth. In fact, in the early days, we have a lot less insight into how things spread at that point because keep in mind, we're older than Google Analytics. We used to run our own analytics toolkit. Actually that was my first jobs was doing our traffic was we used to see a bump in traffic in those days in September and what we figured was probably happening was when people would go back, like people would go to college or go back to school, there was an opportunity for word of mouth on campus because the 18 through 20 was always – 20s range was always very large on Reddit. We had a suspicion that it was just like, yeah, you come back from the summer and now you tell all your friends about this place you found and then it just spreads that way, but it was definitely very organic. It wasn't until much later where, probably the last five or 10 years where Google's played such a large role. [00:20:24] S1: I'm trying to get a sense of magnitude. I know it's a long time ago, but do you remember that first year, that second year, we're talking about 10,000 users a month, 100,000, a million, what scale were you operating at
00:20:24 · Pablo Srugo, Chris Slowe
[00:20:24] S1: I'm trying to get a sense of magnitude. I know it's a long time ago, but do you remember that first year, that second year, we're talking about 10,000 users a month, 100,000, a million, what scale were you operating at [00:20:37] S0: I looked this up because I figured you were going to ask this question. Let me actually find it. I just was complaining about it somewhere. Okay, so we didn't hit 10 million uniques per month until 2010. We hit about a million a month in mid to late '06, but at that point we also were on a pretty good three to six month doubling curve. If anything, the rate that we were growing was about the same pace where we could actually keep the site perky. When the load's going to be twice as much in three months, it really focuses the mind to like, okay, what's the current bottleneck that we're going to hit and what's the next bottleneck we're going to hit and how do we make sure we're at least a couple bottlenecks away from the actual major outage? [00:21:25] S1: How often were there big outages in those first days, like just didn't work?
00:21:25 · Pablo Srugo, Chris Slowe
[00:21:25] S1: How often were there big outages in those first days, like just didn't work? [00:21:30] S0: I think that the outages were in the early days, very early days, not particularly often and always something that ended up being like we don't know how to use this technology yet. Also, oh man, without failure we would have outages whenever – the team was always very small. If the team decided to go out to dinner collectively, if all four of us were in the same place, somebody's phone was going to go off. It's a guaranteed thing. I remember very clearly being out to dinner and Steve's phone went off and we had to run back to the apartment which doubled as the office, hopped on, and that was the first time we discovered that – We've used PostgreSQL as our database forever, and that was the first time we actually discovered that PostgreSQL had a configuration file we should look at. The tuned version of PostgreSQL back in 2005 was optimized for a 486 with eight megabytes of Ram. It was probably the first time we had an index fall out of memory where it was like, oh wow, we can just add two zeros to this setting and all of a sudden it works. The AI was like, yeah, let's figure out what all these settings do in this particular file, and just that compounded across a bunch of things. By the way, the impression I want to leave here isn't that we didn't know what we were doing. It was like nobody knew what they were doing at that point and we're all discovering it live. [00:22:50] S1: Walk me through, so the acquisition was when, in 2006?
00:22:50 · Pablo Srugo, Chris Slowe
[00:22:50] S1: Walk me through, so the acquisition was when, in 2006? [00:22:53] S0: It was Halloween '06, October 31st. [00:22:56] S1: At that time, Facebook started 2004. I don't know exactly how big they were, but I'm sure things were flying off. I don't know what you guys were looking at if you were comparing yourself to them or other social networks, but I'm curious, what was the logic for taking that exit especially when you had that growth. Those numbers were pretty big, especially back then, right? [00:23:15] S0: Yeah, no, I mean, I think that the breakout successes that were at the same time like, yeah, Facebook was 2004, but Facebook also was still – I think at that point – I can't remember when they rebranded from the Facebook to Facebook. The original domain was even different because it was hard to get facebook.com even then. They were mostly limited to college campuses until a little later. The breakout success summer of '05 was actually YouTube. YouTube came out of nowhere and exploded, and they got bought very quickly that year by Google. I mean, for us it was like – again, this is still at a point where there really wasn't much of an acquisition path available. Companies were still just waking up to the idea of startups again. The pitch at the time from Conde was actually very good, which was Conde is this premium content generator that understands monetization and then we are an engine for getting you to content that doesn't have a monetization path. [00:24:08] S1: Pre-revenue, at the time you guys were making –