I was talking to a VC about a company with a million plus users and I realized how many web applications I've signed up for and never went back to, or only used once and never went back to again. I'm a registered user on those sites, but so what?
What's really important to driving value is that someone uses your service more today than they did yesterday, and startups need to make sure their data capture is robust enough to be able to slice and dice this.
The services that have created the most value for me are ones that I use more now than the first day I started with them: Typepad, Flickr, del.icio.us, Twitter, LinkedIn, Last.fm, Facebook, Highrise. Sometimes, my usage of them grows because others join or I see what others are doing with it, like Twitter. Other times, the product is flexible enough that I can find new uses for it and grow into it, like del.icio.us and tagging things for: myself as a memory tool. All of these applications are ones that I started out with slowly but I use them more and more everyday.
If I was a VC, I'd want to know account level data. Andrew is looking into time spent on the site for the USV portfolio, and of course that's good if it grows, but I'm not sure how comparable that is across different types of sites. Plus, sometimes I'll leave a site open all day and never do anything with it because I forget about it. For me, logins is a good proxy, but not perfect either.
Run the numbers... How many of your users logged in more in the 2nd week of usage than they did in the first? Week 3? Month 3?
Registered users is an absolutely meaningless statistic. A million registered users and $5.35 will get you a Jamba Juice in NYC.