Uncategorized

What is the Rubicon Project? Part II: Solving a massive problem.

October 8th, 2007   |   by Frank Addante

Things have been a little crazy (good crazy) since last Monday’s funding announcement. I wasn’t prepared for all the activity that followed. We had thousands of visitors hit our website in anticipation of our beta and received tons of calls from potential partners, press and investors. I still can’t believe the overwhelming response! Lesson learned: always stay on your toes because you never know what’s going to happen. The team and I have been working overtime to make sure everything is all set for our private-beta invitation site launch that goes live today (see announcement).

As promised, I’m going to give you more information and insight into the Rubicon Project. To avoid this blog turning into a shameless plug for my company, I want to give you some insight into the thoughts behind the decisions we’re making as we move forward.

Validating our hunch about the massive problem

I talked to 100 websites before we started this business. They all complained that their ad networks don’t generate enough money for their ad inventory. So, I went to talk to the ad networks. The ad networks told me that website publishers don’t give them the “right” inventory, and if they did, they could make them more money. So, I went back to the websites and said “why don’t you give ad networks the right inventory?” They said that they have no insight into what the “right inventory” is, nor did they have the technology to do it.

From this, two trends emerged to illustrate the industry problem: first, there is a lack of technology available for websites to monetize their ad space; second, is a rapid growth in the number of advertising networks.

The Internet advertising industry is a huge, yet highly inefficient market. Most business today is transacted via Excel spreadsheets. As I mentioned in my previous post, the current industry is operating like the stock market without a NASDAQ system. Today, 80% of online ad inventory goes unsold (by a website directly) with that 80% being sent to ad networks (e.g. Google AdSense, Yahoo! Publisher Network, 24/7 Media, HispanoClick, etc.). The relationship between websites and ad networks is inefficient, somewhat blind and complicated. Given that $27 billion was spent advertising online last year, growing to $65 billion by 2011, it will be difficult for the industry to scale without an automated system.

How do you go about solving that massive problem?

The short answer is: first, recognize that massive problems usually require simple (to use) solutions; second, you solve it one step at a time. It’s easy to lose focus and try to boil the ocean.

The first step for us was to create an automated system for websites and ad networks to work more efficiently and effectively together. The trick is to create a solution that takes all of the complexity and simplifies it to the point where a small blog can use it, yet is feature-rich enough for a very large site.

Trust me, it’s easier said than done. There is some pretty sophisticated technology required to make it work. It is important to me to build a business that is driven by technology and fully automated so that it can scale.

Our goals with the product were: a) simplicity and ease of use b) “whiz-bang” (our favorite words) and c) full automation. Our engineering team rocks, I think they nailed it (you can check out a demo here by clicking on “Watch a 3 minute demo”).

Deciding on a strategy

We’re playing in a space where there are lots of players on both sides. On one side, there are hundreds of thousands of websites that make money from advertising. On the other side, we have hundreds of ad networks that work with thousands of advertisers. So, we’re putting ourselves in the middle and helping them work better and more efficiently. Ad networks are our channels and we want to be very good channel managers. In order to be successful, we need to provide our channel partners with information and technology to make them more efficient and effective. We don’t work directly with advertisers because that would put us in conflict with our channel partners.

We want to be the “Visa” of online advertising. Essentially, we’re creating the backend transactional system to automate buying, selling and billing for the online advertising industry. Websites are like our retailers. Ad networks are like banks who manage the relationships with the end customers (or in our world, the advertisers). Our job is to help them work efficiently together.

To Beta or not to Beta?

Originally, we planned 5 stages before launch: prototype, working prototype, alpha, private beta, public beta and then launch. Because we have a team of rock stars we’ve already blown through the first 3 stages way ahead of schedule. We’re feeling aggressive and ready to hit the market based on the overwhelmingly positive feedback from websites and ad networks and the promising results with our initial alpha customers (more on that later.)

But, it’s easy to get caught up in the excitement and get ahead of ourselves, so we’re staying focused and disciplined. For example, take our private beta invitation site. There are hundreds of thousands of websites that could use our service to make more money from ads on their site but we’re opening up our service to only a limited set of 500 sites. Why only 500? We want to make sure to walk before we run and focus on those first 500 and turn them into great success stories and happy users that we leverage to launch the service to the public.

Helping the websites

The value proposition for websites is pretty simple: make more money, do less work.

We have created a free to join, self-serve solution for websites to optimize their ad space with over 300 ad networks. We do this by learning the strengths of each of the ad networks and then our smart matching technology deciphers a website’s ad inventory and matches each ad impression with the network that is best able to monetize it. How do we do that?

First, we break down a website’s inventory into about 9,000 micro-segments using demographic, geographic and contextual information. For example, we might create a micro-segment for “female, sports enthusiasts in New York.” Then, we take these impressions and match them up to ad networks that are qualified for this kind of ad inventory. We test and optimize the ads with the networks and then allocate ad impressions accordingly. The inventory from any micro-segment can be spread across anywhere from 1 to 100’s of ad networks. The service does all the work and all the website needs to do is set some preferences on the kinds of ads they want to run, check out some reports and collect checks. We consolidate all the reporting into one place and aggregate billing so we can provide one check for the websites.

Helping the ad networks

Today, ad networks have a big problem with churn. Working with ad networks for a lot of websites is often an “all or nothing” decision because they don’t have the technology to automate and manage multiple networks and slice and dice their ad inventory to send the right traffic to the right networks. Therefore, websites are constantly flipping ad networks, creating a somewhat unpredictable business for the networks. We help reduce that churn.

The ad networks have been incredibly receptive to our solution. We’re able to provide them with more intelligent inventory. Since we pre-qualify and label ad impressions for them, we’re able to give them richer information about each ad impression. We’re essentially serving up ad inventory to them on a silver platter. It gives them more of the inventory that they’ll be successful with, providing them with greater reach and high quality inventory.

It’s all about the beta steps

We’ve been very fortunate to receive an enormous amount of interest in the solution. Now, we need to prioritize which websites make the most sense for our private beta. We want a good mix of sites ranging from small blogs to large sites, different types of content, various user bases and in diverse geographies. We want to learn where the service is most effective and where it is least effective so we can either improve the service or focus in areas where we know we’ll succeed. We’ve got a lot work to do and a lot to learn.

As promised, I’ll give you insight into the day to day progress, decisions that need to be made and the thoughts behind those decisions… Hopefully I’ll be able to share some “in the moment” learning and lessons that you might find interesting…

This should be fun, I’ll post again when I reach an interesting decision point or learn another lesson!