Friday, September 25, 2009

The First Ad Server

Over the years, I've been asked when the first ad server was built many times. Sometimes the question comes from people in the industry, curious about history. Sometimes it comes from lawyers looking to respond to a patent holder. Back in the Nineties, this was a fun story to tell, but honestly, I've told it so many times that I've decided to write it down. If people ask, I'll just send them a perm URL.

So here goes: Back in the Fall of 1994, Jason Strober and I were second year students at Stanford's Graduate School of Business. We both disliked our summer jobs and loved the Web. We met regularly on the weekends to brainstorm businesses we could start online. We considered and rejected many of the most lucrative opportunities (i.e. selling books, auctions, etc.) because we couldn't convince ourselves that people would change behavior fast enough to start buying things online. What we were convinced of, however, was that people would surf the Web in large numbers and for a variety of purposes, both predictable and unpredictable. The early signs were there to support this belief, if you were paying attention. We also concluded that where the eyeballs go, advertising will follow. So our focus turned to web advertising.

The other insight that propelled us was when we realized that an image source URL could point to a remote server. We realized that we could host ads on a server and deliver them across sites. This seems so obvious today to anyone who knows HTML, but at the time the idea of an ad server was novel, let alone central ad serving.

We began iterating on business ideas to leverage this technology. We came close to building a marketplace, much like today's ad exchanges, but concluded that it was a bridge too far for advertisers in 1995. Instead, we focused on building a hosted infrastructure for serving ads for advertisers.

In the Spring of 1995, during our final quarter at Stanford, we completed our business plan, began seeking VC funding, and even pitched Netscape to build an ad server for them (we didn't get the business, but we did influence Netscape to adopt the CPM model for selling ad space online). Our business plan was completed on June 18, 1995 (if memory serves), and our first ad server went live that same month. Here's a picture of the internet's first ad server (or what's still left of it) hanging on the wall in my office:



Mango was commissioned in June 1995, marking the first time an ad server was placed into production, and de-commissioned five years later. If you click on the image, you can even read the vintage of the processor (a Pentium 66) dating from way back when 66 MHz processor was state of the art!

In July, 1995, we officially launched Focalink Media Services with this press release:



While the launch got covered by a trade magazine, which had an early web site, the web coverage has been lost in the mists of time (if anyone can find coverage of this event available online, please comment on this post with a link -- you can read the date and text if you click on the image). The launch did enough to help us get our first customers: Saturn, Prudential and Microsoft. Not too shabby.

Eventually, we hired a larger team, brought on professional marketing help, and relaunched our service in early 1996. There's just so little of the coverage of the early days left online, but there are a few articles still available (1), (2). By this point, we were off to the races. By the end of 1997, we had merged with ClickOver and changed the company name to AdKnowledge. In late 1999, we sold the company to Engage, a CMGI company.

Monday, September 21, 2009

Who's better: Derek Jeter or Scribd?

At the beginning of this major league baseball season, I got curious about Derek Jeter's chances to surpass Pete Rose's all time hits record. I was curious mostly because I'm a huge Yankee fan, but also because it would be great for baseball to have the class act of the league finally relegate Pete Rose to second standing.

I posted a file called "Derek Jeter Lifetime Hits Projection" on Scribd in February of this year. Check it out. It is mildly interesting to note that Jeter is tracking as I projected. Only nine more years and he's there -- no problem! :-)

The more interesting thing to note is how effective Scribd is at organic search ranking. The document I posted was not designed with SEO in mind, and actually doesn't have very many words associated with it. Still, it has been, and continues to be, the top ranked result in Yahoo and Google when you search on "Derek Jeter Lifetime Hits Projection", "Jeter Lifetime Hits Projection", "Lifetime Hits Projection", "Jeter Hits Projection", and other variants.

Recommendation: if you've got something that you want people to find, give Scribd a try.