Nirav Tolia Cofounder, Epinions

Nirav Tolia: To illustrate by analogy, if you're a TV channel — TV is another media that relies on advertising — or a TV show, and you only have five or 10 viewers, you won't make money. Because what advertisers pay for is the access to many, many consumers. So think the fundamental observation I've made with B2C sites is you really need scale in numbers. You need lots of users coming to the site to actually capture money. We have 2.5 million users coming to the site every single month. I'm not sure that's enough to actually get profitable. You need to get to 5 million or 10 million. Yahoo's at 150 million users. So all of the websites out there that assume that they're able to get to 10 million users and 20 million users, those ones are going to have a hard time.

What we've found is that it's easy to get to the first million users; it's a little more difficult to get to the next million. So when you're 2 million you're feeling pretty good and you're thinking, yeah, I can get there. Getting from 2 million to 3 million, incredibly hard. From 3 million to 5 million, even harder. So as you gain more audience, it's more and more difficult, but you need to do so in order to monetize the business to consumer side.

Jeffrey Rayport: Nirav, you're describing a dynamic which on the one hand you've seen, it's empirical data. On the other hand, doesn't it fly in the face of all the early promise of increasing returns models on the Web? Where actually getting to the first million was tough and expensive, but then the next million started to look like a higher-margin business, and in fact because of viral effects and all the rest of it, network dynamics, you'd build that curve very fast.

Tolia: I think increasing returns is still a principle that applies to the Internet. The problem is that people don't know where that inflection point is going to occur. So you have to keep on working, keep on working. Maybe it's a million users, maybe it's 2 million, maybe it's 5. Getting there is very difficult. The example I'll give is eBay, which is considered an increasing-returns type model. In four years, eBay went from zero users to 4 million users. That's fairly respectable, but what we call organic growth. It took time. It has actually been quicker for us to get to 2.5 million users than it did for eBay. However, in Year 4, eBay went from 4 million users to 15 million, because they hit an inflection point. When they hit an inflection point, they went into increasing returns, and they experienced what's called non-linear growth.

So I think increasing returns and being out there and having the viral growth and all the benefits of scale, those are all true things, but don't get caught into the negative or positive hype that it will take a million users to get there or 2 million or 3 million. We just don't know when these systems really explode. Microsoft didn't know when it sold its 1 millionth copy of Windows that it would suddenly have a lock on the market. You're just sort of doing as much as you can, building your business. Then, when increasing-returns models hit, you just aren't even prepared for it. All of a sudden you hit an inflection point, and things explode. So I still believe in that concept. We just don't know where those inflection points are.