Serial entrepreneur Frank Addante knew everything at his latest venture, the Rubicon Project, must scale from the very start.
00:07 Frank Addante: I sat them down and I said, "Look, here's $2 million. I want you to go spend that as fast as you can."
In 2007, Frank Addante co-founded the Rubicon Project, a technology company that automates the buying and selling of online ads.
The company launched on April 28th, 2008 with a small team.
00:22 Addante: When we did our big public launch, we had this huge influx of interest in the product. We had about 500 customer leads coming to our website on the very first day alone, and this came from large companies like Fox and CBS and NBC. So, we were pretty overwhelmed with demand when we first launched the product. So, for the first year or so, we were just focused on, "How do we go and just keep up with the demand, put more and more servers online, hire more and more engineers to write code, and build products for customers?"
By February 2011, Frank hired more than 100 engineers.
00:59 Addante: So, one day I got a call from one of our sales reps, and they told me that one of our best clients was threatening to leave. And I dug into it, and I said, "Why?" It seems like we have such a great relationship with this client and they were one of our early clients. And what they told us was that our smaller competitors were developing features and reports and had various technologies faster than us. And it just really surprised me that anyone can be developing features faster than us, especially because when we started, our biggest benefit to our customers was our ability to move rapidly and be agile.
01:38 Addante: So, I went out to our data center in Los Angeles and met one of our engineers there. He was taking me on a tour of the data center and he was so proud. He was so proud of how little equipment we had in the data center at the time. And I looked at him and I said, "Mike, you were processing billions of transactions and seeing my prior company, when I built the data center, we were doing far less transactions but had far more hardware." And he looked at me and said, "Yeah. Look at what we've done. Look at how little money we have spent." And I said to him, I said, "Well, do you think we have enough?" He said, "Well, no. We could always use more." And I said, "Well, we can always use more or we need more?" And he said, "Well, no, there's a backlog of software engineering projects that have been developed that we just can't operationalize at scale, because we just haven't put that hardware online." And it was quite a powerful moment, because what I found was that we can write software and code faster than we can operationalize it.
Frank gave the head of the data center $2 million for new hardware--and told him to spend it fast.
Within 45 days, the company had enough data storage in place to handle rapidly-developing software.
02:54 Addante: While we were chasing this demand and this growth, what we failed to recognize was that, we transitioned from just being a software engineering company, to also being an infrastructure engineering company. So, what used to be simple is no longer simple, it can be a very costly thing. What we realized is that we can't do anything little anymore. So, at the volumes of transactions, billions per day, what we found was that, it might cost us just a few thousand dollars to write the code, but then we have to go invest millions of dollars in infrastructure to support that code and operationalize it at scale for our customers, because anything you do needs to scale from day one, and that was a huge lesson that we learned.
As of November 2012, Frank points out the Rubicon Project delivers digital ads to 97% of the U.S. Internet population--more than Google or any ad technology.
That's 6 billion ads and 500 billion bids on those ads processed every day.