Paul Graham and his friend Robert Morris started
Viaweb in 1995 to make software for building online
stores. A few days into writing the first prototype,
they had a crazy idea: why not have the software run
on the server and let the user control it through their
browser?
Within weeks, they had a web-based online store
builder they could demo to investors. They launched
at the beginning of 1996.
Viaweb was one of the first companies to deliver
on the Web's promise of creating a level playing field.
Using Viaweb's software, small businesses could make online stores as good as
those built by big catalog companies. And many did: by 1998, Viaweb Store was
the most popular e-commerce software.
Viaweb was acquired by Yahoo in June 1998 and renamed Yahoo Store. In
2005, Graham cofounded Y Combinator, a seed-stage investment firm
Livingston: You had a different startup before Viaweb, didn't you? Can you tell
me a little about that?
Graham: Before Viaweb we had a startup called Artix. We were going to put art
galleries online. The problem was, art galleries didn't want to be online. They
still don't want to be online. We spent a long time trying to convince these
people to use something they didn't want before we had the idea that maybe we
should make something people actually did want.
Livingston: You scrapped Artix and switched to making software for websites
for online stores?
Graham: Yeah. Actually, it's pretty similar software. We realized that if we
could write software that could generate sites for galleries, we were only a shopping
cart away from generating online stores. Everyone seemed to want online
stores, so why not just do that instead?
205
Paul Graham
Cofounder, Viaweb
15
CHAPTER
At least, we thought everyone wanted online stores. There was a lot of talk
in the press about e-commerce then, because Netscape was doing a big PR
campaign for their IPO. They had to convince everyone that the Internet would
be economically important, and they picked the most literal example they could
think of. Actually most merchants didn't want to sell online, not yet. But when
they started to want to, we were there.
Livingston: Take me back to when you were first working on Viaweb. What
were some of the first things you did? Did you have any funding?
Graham: In the very, very beginning, no, we didn't have any funding. It was
just me and Rtm [Robert Morris] in his apartment. It was in the middle of
summer. Rtm was in grad school, but because it was the summer he had some
free time. We just said, "OK, we'll try and write a prototype." We wrote the first
version in a couple days.
One of the unusual things about Viaweb was that it worked over the Web.
That's where the name came from. It was a web-based application--as far as I
know, the first one. But in the very beginning, it wasn't web-based. At first it
was going to be software that you would use on your desktop computer to build
a website that you would then upload to a server. Then in the first couple days
of working on it, we had this idea, "Hey, maybe we could make this run on the
server and have the user control it by clicking on links on a web page." So we sat
down and tried to write it and, sure enough, you could write a program that
worked this way.
Livingston: This was a new idea, right? Do you remember when it came
to you?
Graham: At the time most of the hackers we knew used this program called
X Windows, where you could be using a program that was running on some
remote machine, but it would be drawing stuff on your screen. There was also
this idea of an X terminal, or xterm for short, which was a computer that did
nothing but run X Windows--all the brains were on the server. So the way we
thought of web-based applications at first was using the browser as an xterm.
Could we just treat the browser like an xterm, and have the application running
on the server?
So it wasn't that huge a conceptual leap if you came from our world, but it
was a bit of a conceptual leap. I remember very well when I had the idea. I was
staying in this spare room in Robert's apartment during the summer, because at
the time I was living in New York, and I woke up one morning with the idea. As
I was lying there half asleep this idea of making the software run on the server
popped into my head and it was so dramatic that it woke me up. I sat up in bed,
like the letter L, thinking, "We have to go try this."
Livingston: Do you remember how you felt when it worked?
Graham: I was pretty excited, because it meant we could start a company without
having to learn Windows. The prospect of having to write desktop software
was horrifying to us, because at the time, writing desktop software meant writing
Windows software. Neither of us knew how to write Windows software and
206 Founders at Work
we didn't want to learn. It seemed like this huge steaming turd that was best
just avoided. So the main thing we thought when we first had the idea of doing
web-based applications was, "Thank God, we don't have to write software on
Windows."
Livingston: So you have this major breakthrough. What were some of the next
things you did?
Graham: Pretty early, we got some funding from our friend Julian, who also
worked with us on Artix. He gave us $10,000. After about 6 weeks or so, it
seemed like it was going to be more work than we thought, so we got Trevor
Blackwell to work on it too.
Livingston: How did you know Trevor?
Graham: Trevor was in grad school with Robert. I asked Robert, "Who's the
smartest grad student in the computer science program?" and he said "Trevor."
I couldn't believe it actually, because at the time I thought Trevor was a total
goofball.
Livingston: But you were soon convinced he was talented?
Graham: Trevor is a prodigy, in the original sense of the word. When we first
recruited him, we asked him to write this little piece of image-manipulating
software, to kind of test him out. For 2 weeks we heard nothing from him, and
I had pretty much written him off. Finally he sent me an email asking me to
come to his office to see what he'd done. I went there expecting to see this new
image software, and instead he's rewritten our entire system in Smalltalk--
everything I wrote, plus everything Rtm wrote.
I basically said, "OK, you're hired. Now go and write the damn image software,
because we're not rewriting everything in Smalltalk."
Livingston: You and Robert were already good friends, right?
Graham: Oh yeah. We had been friends then for about 10 years--since way
back. In fact, I think in the beginning it was only because he was friends with
me that Robert even did this. In the beginning he was just humoring me. It was
a year before he thought Viaweb had any chance of ever making any money.
Livingston: So you convinced him to spend the summer working on this
project. What happened in the fall?
Graham: Things kind of came to a head with Rtm. We had this angry phone
conversation where he said something like, "We've been working on this thing
for a whole month, and it's still not finished." It's funny in retrospect, because
we were still working on it 3 years later. At the time, I was just thinking about
how to get him to keep working on it for another month. But that was the main
reason we got Trevor. Robert basically rebelled, so I thought, "All right, we
need more programmers."
Livingston: If Robert was so reluctant, why did you start the company with
him?
Paul Graham 207
Graham: Well, first of all he was my best friend, so I really trusted him, but he's
also one of the best programmers in the world. I'd rather have a quarter of his
brain working on some problem than 100 percent of most other people's.
Livingston: Where did you work?
Graham: In Robert's apartment. His housemate was away that summer, and I
moved into his room.
Robert used to get up early, whereas I stayed up till four and got up at noon.
So we would kind of work a 24-hour schedule. I would write some new code
during the night and send Rtm an email saying, "OK, we've got all these new
features in my part of the code." Then he would write the corresponding stuff
in his part. So we got code written very fast.
Livingston: On one computer?
Graham: Uh, well, there was a large university nearby whose computers we
sort of unofficially used.
Livingston: Nearby in Cambridge, Massachusetts?
Graham: Yes.
Livingston: What was the next big turning point after you realized you could
make this web-based?
Graham: The next turning point was when we had a working demo--when we
actually built an online store using our software and you could order from it,
and edit it through Netscape. We started Viaweb in the middle of July '95 and I
think we had this first demo in early August.
Livingston: Who were the first people that you showed the demo to?
Graham: The first people we showed it to were some potential investors. We
ultimately decided not to take money from them, because they wanted a majority
share of the company for a comparatively small amount of money. But the
existence of these potential investors did spur us to write our first version, to get
that demo working.
Livingston: Once you had this demo, did you start thinking about signing up
customers or were you focused on raising money?
Graham: What we really thought we needed to do was write more software.
We were software guys. Maybe someone who knew more about business would
be thinking about going and getting customers, but frankly the idea of customers
frightened us. We thought, "Before we go get any customers, why don't
we just write a few more thousand lines of code?"
Livingston: Why were you frightened of customers?
Graham: Being a sales guy and being a hacker are two very different kinds of
work. We were very comfortable dealing with hacking, but dealing with customers
seemed like this terrifying unknown. If it seems strange to you that we
were afraid of customers, imagine how the average sales guy would feel about
modifying the software running on his laptop. The idea would seem terrifying.
Whereas to a hacker, big deal.
208 Founders at Work
Livingston: So what did happen?
Graham: We wrote a lot of software. We thought, "That's what we're good at.
That's what we'll do." We just tried to put as much distance between any potential
competitor and us as we could.
By that fall, we probably had a better online store builder than any of our
competitors ever had, even 3 years later. In October or November I went down
to New York and did demos for some angel investors and we got $100,000
more, which seemed to us more money than we could ever possibly spend. (We
were wrong.)
Livingston: So what happened next?
Graham: We were very encouraged that the angel investors wanted to invest.
We gave demos to two investors. We only wanted to raise $50,000, but both of
the investors who saw the demos said yes. So we thought, "All right, we'll raise
$100,000 then, since they both said yes."
Then we wrote more software. It didn't look then like we had an awful lot of
competitors, so we took a risk and rewrote most of the code. Even though it was
pretty good, we thought, "If we're ever going to rewrite this thing, now's the
time to do it." Finally in December we started trying to get users.
Livingston: Who were your first customers and what did they think when you
first showed them Viaweb?
Graham: Our first customers were a pair of technical bookstores. Robert actually
went with me on the sales call to the first one. He just sat there absolutely
silent through the whole thing. I think both of these bookstores were frightened
of Amazon. Most people back then, you had to kind of twist their arm to get
them to sell online, but not people in the technical book business.
Livingston: Tell me a little about your relationships with your first customers.
Graham: We felt like we had to have five or six customers to launch. And for
these first customers, we basically would do whatever they said in order to get
them as customers. We gave them the software for free for as long as they
wanted. We built their sites ourselves. If they needed to have images in them,
we would scan the images. We were basically web consultants, because we
needed users; you can't launch a thing like this without having any users.
That's one of the problems with web-based software. If you're making desktop
software and you launch the thing, no one can tell how many other users
there are, right? But if you're making web-based software and you're hosting
the websites that these guys build, then if you don't have any users, the entire
world can see that.
Livingston: Were most people that you tried to pitch your software to online
retailers? Were there things that they misunderstood?
Graham: One of the big things we got wrong was that we thought our users
were going to be catalog companies. Now all the catalog companies are online,
but back then, they just didn't want to hear about the Web. This was late '95,
early '96. A lot of people didn't even have web access yet. So these middle
Paul Graham 209
managers at the catalog companies we called up, at that point they just wished
the Web would go away. It was just making their lives more complicated. We
would call them up and tell them how we could solve all their problems and
make an online store for them, and it was kind of like the dentist calling up
and saying, "Why don't you come in for that root canal?"
The people, it turned out, that really wanted our software were individual
merchants--guys who had some kind of specialty store selling antique chess
pieces or something like that, and up till now had relied on people coming to
their shop to buy stuff, or maybe occasionally they would mail out a xeroxed
price sheet. For these guys, the Web was huge, because it allowed them to have
what the catalog companies had. Those users loved us.
Livingston: Why did users like Viaweb?
Graham: I think the main thing was that it was easy. Practically all the software
in the world is either broken or very difficult to use. So users dread software.
They've been trained that whenever they try to install something, or even fill
out a form online, it's not going to work. I dread installing stuff, and I have a
PhD in computer science.
So if you're writing applications for end users, you have to remember that
you're writing for an audience that has been traumatized by bad experiences.
We worked hard to make Viaweb as easy as it could possibly be, and we had this
confidence-building online demo where we walked people through using the
software. That was what got us all the users.
The other thing was, we had good graphic design. Our secret weapon was
that we knew that e-commerce was really about graphic design, not transaction
processing. Unless you had a site that could convince people to buy, you didn't
have a transaction to process, and what convinced people to buy was how good
the site looked. So we made sure that our software made great-looking sites--
not just better than our competitors, but better than most of the sites that big
companies paid web consultants half a million dollars to make for them.
We didn't even process credit card transactions till about 2 years in. We
would just forward the order to the merchant, and they'd process it like a phone
order.
Livingston: Who were your competitors? Were there any that you worried
about?
Graham: We worried about different ones for different reasons. Our biggest
competitor was a company called iCat. Fortunately for us, they were not very
good at writing software. They were, however, very good at raising money and
seeming corporate. At one point they did one round of funding that was more
than our entire valuation, in fact probably twice our valuation. But fortunately
they were never a threat technically.
At first they weren't web-based; they had desktop software. Finally they
came out with a web-based version. Trevor and I were at a trade show when it
launched, and we noticed that the URLs for static pages were something like
"display-file" with a file name for an argument. So we tried replacing the
210 Founders at Work
argument with "/etc/passwd" and sure enough, the server displayed the password
file right in the browser. And there were accounts with no passwords.
I mean, this is programming 101.
There was another competitor called Shopsite that was better technically,
but still not too dangerous. Plus they were out in Utah; they weren't really connected
to the startup world. Whereas iCat was in Seattle, which was much more
startuppy. For some reason there were no serious competitors in Silicon Valley.
Livingston: Tell me about some of the other major turning points in the first
year or two of Viaweb.
Graham: There were a lot of turning points. Basically Viaweb's history was one
turning point after another, alternately up and terrifyingly down. A couple days
after we launched came the next turning point, when a giant company called us
up and wanted to buy us, right on schedule. It was just like we thought it was
going to be. We're these great hackers, we write this clever piece of software,
we launch the thing, and rrring, there goes the phone and it's some big company
wanting to buy us.
Livingston: What happened?
Graham: There was kind of a clash of cultures. First they came to check us out.
They showed up wearing these Bill Cosby sweaters, like someone in corporate
affairs has told them that when they go and visit startups, they're supposed to
not wear suits and they're like, "Uh, what do we wear?" "Wear a sweater that
looks like some macrame class knitted it collectively." So they show up in their
Bill Cosby sweaters and march up the stairs past all the landlady's kids' shoes in
the corridor, and walk in, and this company they're supposed to be buying is
just a grad student apartment with some computers in it.
But they still wanted to buy us after that, so we arranged to have a meeting
at Julian's loft in New York. One of our investors was a metals trader, so we figured
that he must be a great negotiator and we'd let him handle it. The guys
from the big company said, "We want to buy you for $3 million." And he said,
"Well, I won't sell you the company for $3 million, but for $1 million, I'll sell
you an option to buy the company in 6 months for $20 million." At that point
the guys from the big company just got up and walked out.
Livingston: How did you feel?
Graham: For the first day or so, it didn't register with me what had happened.
Then I felt really bad. I realized that if they'd bought us for $3 million, it would
have been more than a million for me personally, so I felt like I'd lost a million
dollars. I'd had a million dollars, and then lost it. I was aghast.
I called up the guy we'd been talking to at the big company and I said, "Do
you still want to buy us?" and he said, "No!" He had lost face, I guess, with his
colleagues for wasting their time on us.
Livingston: So that was a harsh dose of reality about the acquisition process.
Graham: That was my first introduction to something that turns out to be a
very important lesson for startups: it's never a deal till the money's in the bank.
Paul Graham 211
So many things can go wrong with deals, and they all do. Before we ultimately
got bought by Yahoo, we probably had nine or ten different acquirers that we
were talking to, and things always went wrong for one reason or another.
Livingston: So then what did you do? Go back to business?
Graham: Yeah. There were always two stories going on simultaneously with
Viaweb. There was the software and the customer story, which just went
smoothly and wonderfully the whole way along. We kept writing great software,
we kept getting more and more customers, the customers loved us, the growth
was this beautiful, smooth upward curve. Simultaneously, there was this story
about the business, which was one disaster after another. So most of the actual
turning points are not software or customer turning points, because everything
went great there. All the turning points are business turning points.
The next one was probably when Robert went off that summer and took a
summer job working for another company. He went to work at DEC SRC out in
California. The problem was, he didn't tell me he was going to do this until...
Well, actually he never told me. A few days before he left, we were having
dinner with some friends and one of them said, "So Robert, are you looking forward
to California?" I looked at Robert and said "California?" And it turned out
he was going to leave in a week for the whole summer.
So now I had to explain to our investors why one of the founders of the company
they had just invested in had gone and taken a summer job working for
another company. That required all my spin abilities.
Livingston: What did you tell them?
Graham: I said that this was part of his graduate student career and that it was
a common thing for people in graduate school to take jobs working in research
labs during the summer and, yes, this was another company, but it was really
more of a research lab than a company. That part was certainly true. When they
tried to turn AltaVista into a company, it was disaster.
Livingston: What was the next turning point after Robert left for his summer
job?
Graham: Our main angel investor, the metals trader, was encouraged that the
big company had wanted to buy us, so that spring he'd put more money in--still
angel-scale money. We weren't desperately running out of money, but we were
going to run out sometime in the fall. The angel investor decided that we
needed to have a business guy as CEO and that he wasn't going to give us any
more money unless we got someone. So that summer, as well as trying to deal
with Rtm being in California, we spent our time talking to various business
guys.
The problem with all of them was that they had delusions of grandeur. This
was the beginning of the Internet Bubble, remember, and I think all of these
guys saw themselves as some kind of grand CEO, while we programmers
labored in the kitchen cooking the food and washing the dishes. If the deal
were simply that the business guy would be the public face of the company, but
we would be allowed to do what we wanted and make sure everything worked
212 Founders at Work
right, that would have been OK. But we were worried about what might happen
if one of these guys wanted to actually be the chief executive officer and tell
us what our strategy should be. We'd be hosed, because they didn't know anything
about computer stuff.
Livingston: So what did you do?
Graham: We lucked out. At practically the last moment, we found Fred
Egan--or rather, he found us. Fred Egan saved us. That was a great turning
point, when we got Fred. The lowest point, well, maybe tied for the lowest
point in the company's history, was that summer when Robert was away and the
investors were pressuring us to take some business guy as our boss. When we
finally got Fred, that ended that summer of horror.
Livingston: What was so special about Fred?
Graham: He didn't need to be our boss. He was willing to be the COO and do
the business stuff and let us handle the technical stuff. He had worked for a
company that I had worked for, actually, Interleaf, and so he came with a lot of
credibility. In fact, he had been a big executive at Interleaf while I was just a
peon, so I was very impressed with him.
Livingston: Did he reassure your investors?
Graham: Oh God, that was so great. I remember Fred's first day. The metals
trader was an extremely fearsome guy. He seemed like the kind of guy who
would wake up in the morning and eat rocks for breakfast. On Fred's first day,
the metals trader called up, and Fred answered the phone and said, "Hi Alan,
are you buying or selling?" And I was so relieved. Finally I had someone to take
over that stuff.
It was such a relief to have someone who would deal with the investors,
so that we could just write software and make users happy. That's all we wanted
to do.
Livingston: Tell me a little bit about your relationship with your investors.
Graham: I think, because we didn't seem very businesslike, most of the
investors didn't really have any confidence in us as a company until we got
bought. I think it was only then that they were really convinced we were doing
a good job.
We didn't seem very businesslike for the same reason we didn't seem very
well dressed. We just didn't bother with that stuff. But we did concentrate on
the stuff that really mattered, which was making users happy.
Livingston: If the company that they've invested in was doing well, then why
was the relationship bad?
Graham: Well, I suppose they thought it could be doing better. We were getting
users at a certain rate and maybe they thought we could have been getting
users at twice that rate. I don't think we could have. We already had more users
than anybody else. There just weren't that many users out there to increase the
rate that much.
Paul Graham 213
There was one investor who I think really wanted to run the company. He
had just sold his own startup, and he was pretty young. It was hard for him to
just be a passive investor. For a while he actually came to work for us, as a VP.
You know, in retrospect I think the big problem with our investors was that
we weren't forceful enough with them. I think investors like to be bossed
around, like horses. It reassures them when you're in control. But these guys
were much older than us and had given us huge sums of their money, so it was
hard for us to boss them around.
Livingston: So now you have Fred on board and you are becoming more legitimate.
What did you do next?
Graham: Soon after we got Fred, we raised more money. I don't remember
exactly how much--maybe $800,000. A lot more money than we ever had
before. We really shifted gears at that point. Up till then, we had been operating
out of an apartment. In the very beginning, we operated out of Robert's
apartment. Then after we got the $100,000 in that first round from the angel
investors, we rented the apartment upstairs from Robert's. We had that for
about a year and then after we got Fred and we got this new round, we actually
rented an office and started hiring people. We really started to look like a
company.
Livingston: Were you worried that you didn't look enough like a company
before that?
Graham: We were big beneficiaries of that rule that on the Internet, nobody
knows you're a dog. We were just a bunch of guys in an apartment with computers.
Nowadays more people accept that startups look like that, but not back
in the mid-'90s. People still expected a company to have a real office. I think if
some of these companies whose online stores were on our server could have
actually seen the room that the server was sitting in, they would have freaked.
Thankfully they never did.
If anybody ever did want to come and visit us, we pulled all kinds of tricks to
make ourselves seem more legit. When that first giant company wanted to buy
us and sent people over to check us out, all we had in our so-called office was
one computer. Robert and Trevor mostly worked at home or at school. So we
borrowed a few more computers and stuck them on desks, so it would look like
there was more going on.
One of these Potemkin computers was Robert's, from the apartment downstairs.
And in the middle of this big visit from the company, Rtm comes upstairs
boiling mad because he's come home and discovered his computer's missing.
He's like "What have you done?" We said, "Shhh, shhh, we had to borrow it.
We're trying to look real. Don't worry, we're not using it. It's not even plugged
in." He was really mad, but he let us keep using his computer as a prop for
another hour until they left.
Livingston: So you are growing as a company, you get a lot more funding, what
happened then?
214 Founders at Work
Graham: We started to seem more real then. By that point, we were starting to
get mentioned in the press a lot. Early on, people found out about us through
word of mouth. We were the underdogs--the guys who have better technology
but nobody's ever heard of. This was the point where we stopped seeming like
total underdogs and people started to know about us.
Part of the reason was that we hired a fabulous PR firm with this money,
Schwartz Communications. We told them, "When people talk about e-commerce
and they have to mention a few examples of companies, we want to be one of
the companies they mention." The most valuable sort of press is not articles
about you, it's when people mention you in passing as a matter of course. That's
what you really want--whenever anybody talks about e-commerce, for them to
say, "companies including... and Viaweb." Schwartz got us that within a couple
months.
Incidentally, it was one of the guys at Schwartz who came up with the term
"web-based software." Up till that point we'd called it "server-based."
Livingston: Were you still getting acquisition offers at this time?
Graham: There were always people trying to buy us. There was another one
just at the point where we found Fred Egan--a Japanese company that later
made an imitation of our software and went on to become a big success in
Japan. Rakuten, they were called.
Livingston: They copied you?
Graham: Not very well. It's sort of like if you copied a dog by taking a photograph
of a dog and sticking it onto a cardboard cutout. From certain angles it
would look like a dog, but if you threw a stick and yelled "fetch" it wouldn't do
anything. The Japanese market wasn't as far along then, so in their market, this
was pretty advanced. But the entire time, there were always people trying to
buy us, of various levels of seriousness.
Livingston: Did you have to take any more funding?
Graham: We did have at least one more funding round, under the most disastrous
circumstances. One of these companies that tried to buy us--actually the
second to last one; we were almost done--was a big Internet portal. We had a
handshake deal with them and in the process of the due diligence for the deal,
it was discovered that one of our programmers had signed a piece of paper with
the company who had paid for him to go to graduate school, saying that everything
he thought of belonged to them.
Livingston: So they owned the intellectual property?
Graham: They might have. What it said on the piece of paper was that they
owned ideas relating to their business. But this was a huge company and
arguably just about anything you could do with software related to their business.
So we had to go and get a release from them and that took a long time,
during which the acquirer welched on the deal.
It turned out to be good in the end, but we had to raise our last round of
funding while this was happening. You want to raise a round of funding with an
Paul Graham 215
IP cloud over your head? It's just impossible, because potential investors have
no way of judging how serious it is. It could be no big deal, or it could be that
this other company owns half your software.
That was the second low point--tied for lowest. Ultimately we managed to
get some bureaucrat within the big company to give us a release, so we could
say to acquirers we actually owned our software. But we had to do a round of
funding before that, because we were out of money.
It was pretty miserable. Basically, the angel investors played chicken with
us. They knew we couldn't get money from anyone else, since we didn't even
know for sure if we owned our software. So they proposed to do a cramdown
round where they would refinance the company, I believe, at a pre-money valuation
of zero--meaning all the common stockholders were completely wiped
out. To keep us around, since they kind of needed us to write the software, they
were going to give us options. So we called their bluff. We said, "If you do that,
we're leaving."
Livingston: You and your cofounders said you were leaving?
Graham: Yeah, all the technical guys. So when it came down to that, they compromised
and we ended up doing a funding round at a low, but reasonable,
valuation--$12 million, I think. We got bought only a couple months after
that round closed. But we had to do the round because we were in debt at that
point.
Livingston: You must have been displeased with your investors for doing that
to you.
Graham: Well, everybody ended up rich, so it's hard to be too displeased. I'd
rather have an investor who invested in us and made our lives hell than one who
didn't invest in us at all, which is what most investors do to most startups.
I mean, we needed their money to grow the company, and some amount of
stress always comes with the money.
In retrospect, I think it was more about control than money. They weren't
trying to rob us so much as take over the company. They were offering us quite
a lot of options. The point was, we'd have to do what they said from then on, or
lose them.
Livingston: Was there ever a point when you wanted to quit?
Graham: There was one point when I almost did quit, when the investors were
telling us they were going to refinance the company. I had an appointment with
a lawyer to figure out how to quit without getting sued. I was on my way out the
front door when Fred Egan grabbed me and said, "Wait, let's see if we can fix
this." It was pouring with rain and I was not too psyched about having to go find
a cab in that, so I went back to work while he made some phone calls. I don't
know what he said, but I guess he convinced the investors I wasn't bluffing.
I wasn't, either.
We had some leverage, because the investors already had over a million dollars
in the company. I don't know if they realized how hard it would have been
to just hire a bunch of programmers and throw them in there and have them
figure out the code, but it would have been really hard.
216 Founders at Work
Livingston: So a few months later after this horrible low period, you have a
great high period because you get bought by Yahoo. How did that happen?
Graham: We especially wanted to get bought by Yahoo. If you had asked us,
"Who do you want to get bought by?" we would have said, "Yahoo." In fact, we
did say that; we kind of spread the word that we saw Yahoo as the ideal acquirer.
We'd tried to do an online demo for Yahoo about 6 months before. We
could do demos by phone where we'd talk people through editing a site and we
could see from the log files where they were clicking. I tried to do a phone
demo for Tim Koogle in the fall of '97, but he couldn't even get to our server. It
turned out some router was hosed halfway between us and them.
The way we really got onto their radar screen was through Ali Partovi. He'd
had Robert and Trevor as teaching assistants in CS classes at Harvard a few
years before. He had a startup called LinkExchange that was talking to Yahoo
at the time, and their VC was Mike Moritz, who was also Yahoo's VC. In the
end they got bought by Microsoft instead, but not before they'd told Yahoo
about us.
Livingston: How did it go with Yahoo?
Graham: We liked them. They were like us. They had hacker values, basically.
They were from graduate programs in computer science too. They were our
tribe of people, not these weird business people we kept having to deal with.
Plus they weren't jerks about the acquisition. So many companies play hardball
in acquisitions. It's so stupid. Don't they realize that the people they're
trying to squeeze are going to have to work for them afterward? Yahoo was very
upstanding about the deal. They didn't require any vesting, for example. We
could have quit the day after the deal closed. But because they'd been good
guys, we worked hard to make the acquisition work out well for them.
It did, too. Yahoo made a lot of money from this software. When you tell
people you sold a startup to Yahoo in 1998, they get this knowing look, like you
sold someone a bag full of air for a hundred million dollars, but Viaweb was a
real money-making acquisition for them.
Livingston: What was the most surprising thing about being acquired?
Graham: For me the most surprising thing was the day the deal was going to be
announced. There was a point where I had to change our front page to read
Yahoo instead of Viaweb and then it really hit me. Viaweb is gone. Viaweb
doesn't exist anymore. That was so weird. And I told myself, "Look Paul, don't
get sentimental. You built this thing to sell it. That was the whole point, and
now you've sold it, so stop whining." But boy, it was strange to think that when
I clicked on "publish" and replaced the Viaweb front page with the Yahoo front
page, Viaweb would never be seen again.
It was also kind of weird that when the deal closed, we all became Yahoo
employees. It was like one of those dreams where you have to go back to high
school. Up till that point we'd been independent, and then suddenly we were
employees, with bosses. And the weirdest thing was, we, or I at least, actually
started to think of them as bosses. Now whatever I did was either submitting or
rebelling, whereas before it had been just doing.
Paul Graham 217
I think Yahoo is smarter now about dealing with startups than they were
then. We were one of the first companies they bought, and I think the idea was,
back then, that what you should do with an acquisition is "integrate" it, in the
same way that a sugar cube becomes integrated with your tea. We basically got
dissolved within Yahoo, and all the people working on Viaweb--or Yahoo Store
as it then became--got dispersed to all the corresponding bits of Yahoo. The
engineers got put with the engineers and the people working on customer support
got put with the support people and the sales guys got put with the Yahoo
sales guys.
It seemed to Yahoo that this was the most efficient, organized way of doing
things, but actually it was terrible for us. We had been this little tight-knit group
that worked really well together and suddenly we were spread out all over
Yahoo.
Livingston: Any general thought you have on the acquisition process, since you
had several offers?
Graham: Never believe it's a deal till the money's in the bank. Even at the point
where you walk in that room to sign the final papers, there's still a 10 percent
chance the deal's going to fall through. At the point where people say, "We want
to buy you," the chances of it falling through are like 80 or 90 percent. So you
can't let yourself believe. If someone wants to make you an offer, fine, but don't
change your plans based on that. Just keep going.
Livingston: Looking back, what surprised you most in your experience with a
startup?
Graham: One thing that was surprising was that it actually worked. There we
were, in the summer of 1995, thinking, "We don't know anything about business,
but we're good programmers. Maybe if we write a really good program,
we'll make something all these users will want and we'll get lots of users and
then some big company will buy us." And 3 years and enormous numbers of ups
and downs later, that's exactly what happened. We had this theory about how
business might work, and we sort of forced it to conform to our theory.
I know Robert was surprised that we made any money, because I have a real
index of how he was feeling about Viaweb early on. A couple months in, he and
I were having dinner, and I made a bet with him that if he ever made a million
dollars out of Viaweb, he would get his ear pierced. So the day after the Yahoo
deal closed, Trevor and I grabbed him by one arm each and took him down to
the Garage in Harvard Square, where all the teenagers get their nose rings, and
we got his ear pierced. He spent a long time trying to pick out the smallest one.
Livingston: Some aspects of business turned out to be less of a mystery than
you had thought. What did you find you were better at than you thought?
Graham: I found I could actually sell moderately well. I could convince people
of stuff. I learned a trick for doing this: to tell the truth. A lot of people think
that the way to convince people of things is to be eloquent--to have some bag
of tricks for sliding conclusions into their brains. But there's also a sort of hack
that you can use if you are not a very good salesman, which is simply tell people
218 Founders at Work
the truth. Our strategy for selling our software to people was: make the best
software and then tell them, truthfully, "this is the best software." And they
could tell we were telling the truth.
Another advantage of telling the truth is that you don't have to remember
what you've said. You don't have to keep any state in your head. It's a purely
functional business strategy. (Hackers will get what I mean.)
Livingston: Were there things that nontechnical people misunderstood about
what Viaweb was doing?
Graham: Constantly. No one ever seemed to get that the software ran on the
server. Nowadays there are so many web-based applications that you take this
for granted, but this was a year before Hotmail. We would explain to people
how the thing worked and give them a demo and they would say, "Great. Where
do I go to download it?"
After we got bought by Yahoo, a reporter who had been covering us for the
past 2 years wrote an article about the Yahoo acquisition and at the end said "It
only takes 10 minutes to download." After covering us for years, the guy still
thought this was client software.
Livingston: Is there anything that you would have done differently?
Graham: I wouldn't worry so much about seeming like a real company. Now I
would just say, keep it a bunch of guys operating out of an apartment for as long
as you want, because there's nothing to be ashamed of in that, especially if
you're writing great software.
Another thing I would do is open an online store ourselves. We did use our
software for building our website. We were the only one of all our competitors
who actually used our software for building our own corporate website. But we
didn't have anything people could buy online. If we had been selling stuff,
we would have understood what life was like from the merchant's point of view.
Livingston: What was one of the funniest moments?
Graham: Probably the time we tried starting a gas generator inside our office.
There was this huge blackout in Cambridge that lasted for about five hours. We
always had our servers in our offices with us. We didn't trust this collocation
stuff. Nowadays, collocating is the standard thing to do and even big companies
do it, but we felt like we had to have those servers in the room with us. So when
the power went out, our servers were really dead.
We had some battery-powered UPSs, but they would only last for half an
hour. They were really designed for power spikes, not for the power going out
for 5 hours. So I dispatched Trevor to Home Depot to buy a gasoline-powered
generator as fast as possible, while I sat there watching the UPSs' power go
down, turning off servers one by one--thinking about which customers were on
each server and which ones would be the maddest, and turning off whichever
server would have the least mad customers on it. Eventually I had to turn off all
the servers, because it took Trevor a while to get to Home Depot and back.
Finally he showed up with this gas generator, and we weren't really sure
where to put it because we were in this small office building in Harvard Square.
Paul Graham 219
We were on the top floor; we didn't really have a place to put a gas generator.
The first thing we tried was putting it in the office next to the server room. We
started the thing up and it sounded like the end of the world. It was the loudest
thing I have ever heard in my life. You might think the problem with starting a
gas generator inside your office would be the exhaust, but it never got to that
point. It was so terrifyingly loud. We thought, "Even to avoid our customers
calling us up angrily because their stores are offline, we cannot endure this."
After about 5 seconds, we just looked at one another and shook our heads and
turned it off.
Then we tried putting it out on the street in front of our building. The problem
was, we were up on the third floor. We got every extension cord we could
find in the place and stuck them together end to end, and they were just long
enough to get out the window and down to the street. But only just--it was so
close that the extension cord was actually tight. It was running through our
office at chest height and you could kind of twang it and it would go
"boinnnnnggg." Then we started the gas generator up in the street and that was
just about bearable, so we ran the servers on that for a couple hours until the
power came back.
Livingston: Can you remember any other hair-raising moments?
Graham: At one point, in the Spring of '96, when we only had about 20 users,
we all went off to this trade show down in New York--the first trade show we
ever went to. We came back and it turned out the server had crashed soon after
we'd left and had been down for 11 hours. And nobody noticed! We kept waiting
for the angry phone calls, and they never came. It was so early in the history
of the Web that nobody was ordering from these stores anyway, and they
weren't even checking themselves to see if their sites were up. Half of these
people who had online stores with us probably didn't even have Internet access.
Livingston: Do you have any regrets from the experience?
Graham: One thing I regret is how pathetic we were during much of this whole
process. We all had practically zero assets when we started, and this was during
the Internet Bubble, remember--very early in the Internet Bubble, but still,
there were people starting companies and getting them bought for like $5 million.
Millions of dollars, when the most money I'd ever had in my bank account
was about $10,000. There was a point where we started to seem like a real company--that
is, real enough that someone might actually buy us--and this made
us just pathetically eager to sell the company. We must have seemed like such
losers.
So I can understand now when founders want to sell out for a couple million.
Investors say, "No, you should wait," but it's easy for them to say. A million
dollars seems just overwhelmingly attractive when you have nothing. You don't
care if it's a good deal or not.
I also kind of regret being a zombie for several years straight. I really had no
life during Viaweb. If people are talking about some famous movie and I've
never seen it and have no idea what it's about, it's usually a movie that came out
220 Founders at Work
between 1995 and 1998, because at that point, I was on Mars. I was not part of
the ordinary world of humans. I was sitting glued to a computer all day long, or
asleep.
Livingston: What did you worry about the most?
Graham: Running out of money. That was the big worry. Running out of
money and having to go and get more funding. Getting funding is very painful.
It's so much harder than actually making a successful company.
Livingston: What advice can you give about raising money?
Graham: The advice I would give is to avoid it. I would say spend as little as
you can, because every dollar of the investors' money you get will be taken out
of your ass--literally in the sense that it will take stock away from you, but also
the process of raising money is so horrible compared to the other aspects of
business. You can't work your way out of it like you can with other problems.
You're at other people's mercy.
The way not to have to raise money is not to spend money. Do everything
as cheaply as you possibly can. What you want in a startup is this feeling of
cheap and hip. Not miserly cheap, but cool, bohemian cheap. That's what we
strove for.
Livingston: So investors were your biggest worry?
Graham: Probably, but I worried about all the different things that could kill us
and all the different ways they could kill us. People start startups to get rich, but
what keeps them going day to day is the fear of failure. You've said, "OK, I'm
starting this startup and I'm going to get all the users and be successful," and
once you've told everybody that's what you're doing, if you fail you'll look like
a fool.
So when we did sell the thing finally to Yahoo, in the eyes of the world,
because we got bought, we were a success. Arguably we were already a success,
since we had more online stores than anybody else. But getting bought kind of
locked that in. At that point you would think someone would be thinking,
"Wow, this is great. I'm rich. I can go buy everything I want." But all I was
thinking was, "Thank God we didn't fail."
Livingston: You write a lot of essays with advice for startup founders. What is
the most important piece of advice?
Graham: What Y Combinator prints on our T-shirts: make something people
want. If you make something users want, they will be happy, and you can translate
that happiness into money. That is the basis of a startup. A startup is a company
that builds some kind of technology that people want. The mistake that a
lot of founders make is to build something they think users want, but that users
don't actually want.
Livingston: Do you think having done a startup yourself makes you a better
judge of startup founders now that you are an investor?
Paul Graham 221
Graham: Oh yeah. In fact, I don't know how people who haven't done it can
pick founders. How can they tell? I often see people who seem kind of clueless,
and I can remember, "Yeah, we seemed clueless in exactly the same way." So
those are the guys we invest in.