Part 3: Building Amazon.com
When you joined, the company was called Cadabra, right? “When I joined, we didn’t actually have a name. What happened was, I arrived with all of my junk in a U-Haul truck, and then Jeff told me that: ‘I incorporated it, and it is called Cadabra.’ That’s when I actually had my first ‘uh-oh’ feeling. Because I did not think that was a good name. One of the problems with the name Cadabra — which was short for Abracadabra — is that people would hear it as “cadaver” on the phone which is itself kind of bad.”
GeekWire Interview: Shel Kaphan, Amazon.com’s first employee
- Part 1: The forgotten founder?
- Part 2: Meeting Jeff Bezos
- Part 3: Building Amazon.com
- Part 4: A larger mission
Do you recall the discussion around changing the name? “Yeah. At a certain point, we realized that (Cadabra) had some problems. We broke out the whiteboard, and I remember several afternoons of just throwing stuff up. I am pretty good at kind of detecting potential negative consequences of names. You would propose a name, and usually it wasn’t very long until I … or someone else found some negative connotation.
Any examples of names that didn’t work? “I remember, at one point, Jeff was all excited that we call it Relentless with the idea that we would relentlessly satisfy customer needs or something. It took a little convincing that perhaps that was not going to have the right connotations in everybody’s minds. I don’t remember all of the different ones. But I do remember that I came in one day and Jeff said: ‘I thought of this name, and this is what it is going to be. It is going to be Amazon.com. And the reasons are that the Amazon is the world’s largest river, and we want it to connote size and selection and so forth and also because it starts with the letter A.’ ”
How did that help? “That actually turned out to be significant because one of the first things that opened the doors is we got on the Netscape ‘what’s cool’, and ‘what’s new’ pages. Because those pages were alphabetized, we were above the fold, which is important. We are on the screen when people call up the screen.”
What did you think of the name Amazon.com? “I was not thrilled with it. But I was also OK with it because names are an infinite time sink. One thing that I was sort of dubious about was having dot-com in the name. Later on, one of the other people we hired made a really good observation about that. He said: ‘In a few years, this is going to seem so 1995.’ At this point, it is enough of a brand that it doesn’t really matter.”
Was it the typical startup environment, late nights and a big push to build the site? And what did you do in those early days? “The nature of the technical challenge was much more straightforward than many other things that I had done in the past. It wasn’t something where we had to work for years and then find out if the thing we were building would actually work or if anybody would care. It had a lot of pieces, some of which were not easy. Like constructing a catalog out of a lot of non-authoritative sources, none of which was designed to be easily built into a database. At that point, there wasn’t anything like Ruby on Rails or anything where you could slap together a Web site with some dynamic behavior in 10 minutes.”
How did you build the site? “We built it all in (the programming language) C using just some really basic libraries from NCSA (National Center for Supercomputing Applications)…. There was a library that they had for building really basic things … and we used that and we built on top of Oracle. But most of the pieces were pretty straightforward to build. It was really just kind of grinding through it. And some of the things in those days, like prefiguring the RESTful architecture kind of concept and how to manage the state for user programs where the requests are completely independent and might be separated by hours, that was not something that everybody knew how to do or there was any support for in those days.”
You worked directly with Paul Davis in that first year. What was that like? “He was a terrific programmer, extremely fast. He happened to have some techniques that he developed when he was at UW for exactly the type of problem we were trying to solve.”
And what was that problem? “Your Web server is running code on behalf of the user, but the user is doing something — they might be on page three out of five on an order form sequence or something — and when they come back in, the Web site might have gone offline or something in between requests so you can’t maintain any state on the server, or at least not dynamically. It has to be in either a database on the server or stored on the user side somehow. The way we did it initially was using session IDs in URLs that were actually indexes into a database on the server side because it was before the days of ‘cookies.’ Cookies were not out there yet, and when they started coming in, people were very suspicious of them, and they’d turn them off.”
The company moved to Seattle’s Sodo neighborhood. What was that like? “As soon as we got there, they rented out the first floor to Color Tile. So, one of my first experiences, was, with earplugs in, trying to hack code while they are sawing through the cement walls to put display windows in the first floor. That office didn’t last us that long … but we then moved two blocks north to a warehouse building right where the Pecos Pit BBQ is, and there is a warehouse building that wraps around that.”
Were you storing all of the books that you were shipping at the time? “We would order things from distributors, and then assemble them into customer orders and then ship them out. We weren’t inventorying much at that point. … One of the early controversies about the company is that we were claiming that we had a million titles. Well, we did have a million titles. We didn’t have a million books, but we had a million titles in the sense that we could order them. We might find out that the publishers might take several months to get them to us. But, in most cases, it would be a matter weeks and we were always trying to make accurate promises about how long it was going to take to get something. But in the early days, no one was set up with electronic inventory systems, and what the distributors had was fairly weak. They had CD-ROMs with weekly or monthly inventory on them, but it took a while before we really figured out how to understand all of the information we could get and make accurate promises about it.”
Was there a point during the period you were at Amazon from 1994 to 1999 when you realized that you were on to something bigger? “One of the first points when I went ‘wow’ — actually there were several. We were actually on the air by July of 1995 … and that was the hard launch. The soft launch had been a couple months earlier than that. Right from the beginning we were getting some traction. First off, we got the great notice on Netscape, and I remember there were a couple times when I’d basically be by myself on the weekend working, and I’d get a phone call. I remember one time I got a call from a guy in Florida who saw our Web site, and he was all excited about what we were doing … I was just like: ‘Wow, this is really cool. We are really getting people excited about this.’
Any other memories of early successes? “When we were in the second building, I remember there was a day when Jeff’s face was on the front page of The Wall Street Journal in their little portrait engraved style, and we got really slammed right after that and managed to survive that. Both the publicity, and that we were able to manage the consequences of it, that made a big impression on me. And then … there were like six quarters in a row where our business doubled every quarter, and with no end in sight apparently. That got my attention, and I said: ‘OK, whatever this is going to be, it is bigger than I thought it was going to be.’
Did Bezos feel that way too? “I will admit that when I had conversations with Jeff about this, I don’t know if he ever believed this. But I thought what probably would have happened is that I was going to come up here and put this thing together, then it would work, and then I’d go back to Santa Cruz, and I would log-in once in a while and fix something, and I’d be working on some other project by then.”
It got a little bigger than that: “That was not the right model. To give Jeff credit, I mean, he saw those opportunities and seized them. I am not sure just anybody else would.”
