Exclusive: The Rubicon Project And The Tech Behind News Corp’s Ad Exchange


EDITOR’S NOTE: This article, originally published on August 27, 2013, placed at No. 13 in our 20 most popular articles of the year.

ADOTAS – Last week, News Corp announced it would launch its own private exchange and discontinue any remaining arrangements with third-party ad networks.

“The News Corp Global Exchange, which will roll out over the coming weeks, comprises more than 50 leading web sites and mobile products, reaching millions of highly engaged subscribers and users worldwide via WSJ.com, TheTimes.co.uk, NYPost.com, TheAustralian.com.au, News.com.au, MarketWatch.com, TheSun.co.uk and more,” read a press release announcing the move.

On the tech side, News Corp will be working with The Rubicon Project as its global sell-side platform.

The Rubicon Project launched its pioneering automated advertising platform six years ago, and now works with over 100,000 advertisers in the U.S. and in several key international markets.

The company claims that its success is due to feats in engineering with the most comprehensive platform (automating buying and selling), big data (3 trillion data signals monthly/2 million every second), and network effects (more sellers attract more buyers and vice versa).

According to the company, an auction equals RTB ($3b ad spend through DSPs) + Static Bids ($5b – $7b ad spend through ad networks and agency trading desks) and orders are defined as buyers and sellers doing direct deals, but the process and operations are automated through its platform.

In this exclusive Q&A, Greg Raifman (pictured), who became president of The Rubicon Project in January, discusses his company’s strategic partnership with News Corp and the state of the $10 billion ad automation market.

Q: How is the News Corp. announcement validation for the programmatic ad space?

A: The automation space needs no validation. It is already generating billions of dollars in ad spend, and is the fastest growing category of spend in display advertising. The overall trend is automation. Programmatic is one form of automation. Automation includes auctions (both RTB from DSPs and static bidding from ad networks and agency trading desks) and direct orders between the buyer and seller directly. It’s growing much faster than even we anticipated when we pioneered this category of automation at Rubicon Project six years ago. News Corp’s clearly defined strategy provides global advertisers with programmatic access to their premium, global audience at scale.

Q: Tell us about the tech side of launching such a massive undertaking as News Corp’s ad exchange.

A: Establishing a global private marketplace with News Corp and its many affiliates worldwide is a massive undertaking that involves all aspects of our company working together in a cross functional symphony. As many already know from our prior releases, we have spent years building a mammoth infrastructure of over 27,000 CPUs into one of the largest, and most technologically advanced real time computing and data management clouds in the world. As a result, supporting large global enterprises such as News Corp have become part of the fabric of Rubicon Project. Currently, according to comScore’s measurement, Rubicon Project’s Advertising Automation Cloud reaches about 97 percent of all U.S. Internet users and processes transactions between 500 of the world’s top publishers, and more than 140,000 advertisers. The hyper-scale performance of Rubicon Project’s Advertising Automation Cloud required advanced engineering brilliance in hardware, network and software. To fulfill this innovative need of engineering,Rubicon Project, along with Advanced Micro Devices, Inc. created the RubiCube, a single SeaMicro SM15000 based hardware platform custom engineered for high-speed advertising automation. Resultantly, Rubicon Project’s Advertising Automation Cloud now powers 40% of the Top 500 publishers at such a massive scale it currently manages many colossal private marketplaces, including the one for News Corp. and its affiliates.

You can read more on our hardware engineering efforts in (CEO and Founder) Frank Addante’s Founderblog here.

Q: Who are you top competitors, and what differentiates yourself from the rest?

A: Our main competition is manual effort via telephones, fax machines and paper. Our mission statement from day 1 has been to automate the buying and selling of advertising. Our founders and management team pioneered the first ad servers 15 years ago which was the first form of automation (i.e. it automated the process of putting ads on websites vs. hard-coding them via HTML.)

Today, there are two major automation platforms: The Rubicon Project and Google. These are the only two complete solutions that automate both buying and selling with exchange capabilities to power both. Google’s stack was built mainly through acquisition, while Rubicon Project’s technology platform is mostly organically engineered from the ground up. The main difference is Rubicon Project’s platform is an open platform that all buyers and sellers participate in. Google has a conflict with many publishers and buyers, so some choose not to buy or sell through Google. It’s similar to American Express and Visa/Mastercard. While both are very successful, American Express is the closed network (that banks don’t get to participate in) whereas Visa and Mastercard is an open solution for all banks and retailers.

Rubicon Project pioneered advertising automation 6 years ago and our success is formed on a foundation of engineering excellence. Rubicon Project’s Advertising Automation Cloud is one of the most comprehensive cloud computing platforms in the world with algorithms and infrastructure leveraging 3 trillion data signals per month, processed by more than 25,000 CPUs delivering value to buyers and sellers of display and mobile advertising. Based on our powerful hardware, data and software engineering, the largest advertisers, ad agencies, and publishers worldwide have adopted Rubicon Project’s platform and as a result Rubicon Project ranks #1 in the US and in several key international markets.

Q: In your own words, where is the industry heading, what will help keep programmatic alive?

A: Advertising is the last big frontier of markets to be automated. If you look at any other major market where billions exchange hands between thousands or more buyers and sellers, each of those markets have become automated. Much like numerous other industries like finance, travel, and ecommerce, automation is a driving force in the digital advertising industry and the fastest growing channel in display, mobile, and video advertising. Automation will continue to flourish on a path to providing total control to buyers and sellers for premium orders and auctions based on the preferred way advertisers, agencies, and publishers wish to do business together.

The industry often confuses automation with RTB. RTB is a protocol. Automation equals auctions and direct orders. Auctions (via RTB and static bidding) has grown to $10Bn globally today. Direct orders is on a rapid trend as well. We have more than 100 customers (like News Corp) who useRubicon Project for direct orders.

We started by automating the auction first 6 years ago (automating the sell of unsold inventory from publishers through ad networks). The first deals to be automated on our platform were around $0.10 CPM. Today, we are seeing $20+ CPMs direct orders; we even saw a $68 CPM deal the other day. Similar to other markets, the first forms of automation are always the low-value inventory first. Eventually, it all becomes automated. The first stocks traded on NASDAQ were small cap stocks, Sabre Systems (which automated travel) was built by American Airlines to sell its unsold airplane seats) and eBay automated the garage sale. Today, all stocks are traded electronically, first-class seats and five-star hotels are bought on the Web and there’s a car sold on eBay every minute. Our vision for the industry when we started Rubicon Project 6 years ago still remains the same: to automate the buying and selling of all advertising. Our products and go-to-market strategies have always matched our mission and now we’re seeing the same trend happening in digital advertising today.

Q: How do you incorporate predictive data to improve campaign performance for clients?

A: Rubicon Project’s Advertising Automation Cloud is one of the largest big data processing platforms on the Internet. Currently processing 3 trillion data signals (2 million per second), that’s about 75X the number of searches done in a month’s time and 35X the number of “Like” button requests on Facebook. Our platform also processes transactions across 97% of the total U.S. Internet population (that’s more than any other company in the world, #1 as measured by ComScore.) We deploy a technology called Semantic Grid Computing (“SGC”), architected and created by Dr. Mona Yousry Ph.D. and her team, to analyze massive quantities of data in order to identify patterns and trends, and then use these to make inferences and overcome the current shortcomings of market segmentation approaches. Mona’s background is in big data at AT&T’s Bell Labs and the Sempra Energy Exchange. As a result, with the use of SGC it is now possible for Rubicon Project to market on an individual user basis. By analyzing the digital browsing trends of a user our algorithms gain insight into that user, but more powerfully, by analyzing the digital browsing trends of hundreds of millions of users, the algorithms gain insights into patterns for sets of users, or user clusters. When some of the members of a cluster exhibit an attribute of interest it may be reasonable to assume that others in the cluster may also exhibit that attribute, (without the technology having yet observed it), or may have the potential for exhibiting it in the future. In the case where the attribute of interest is that some members of the cluster have bought a specific product, then others in the cluster may be good candidates for receiving advertisements for the product.

The SGC team has developed a concept for associating an individual with his or her browsing behavior, which we call a “silhouette.” We don’t want to know who a user is, i.e. his/her name, address, phone number, gender, marital status, parental status, etc. These are the types of attributes used in typical demographic segmentation. The technology is more interested in how they browse, what they focus their attention on, what products they have purchased, and how they relate to others who have demonstrated some similarities in browsing behavior. We refer to this behavior as the individual’s silhouette. The algorithms are familiar with you, they don’t know who you are, (nor do we want to), but they know something about you. Something that will help us direct advertisements to users that we hope will be relevant and perhaps useful to them. If our customers are able to present their ads only to those for whom the ad is relevant, then the same value to the buyer can be achieved with fewer impressions so the corresponding CPM price to achieve the marketing objective can go up, and publishers can achieve higher prices for their inventory. Everyone wins.

The News Corp Private Premium Programmatic Global Exchange is a good case study for this issue. News Corp was receiving interest from buyers who wanted to secure specific targeted audience segments in specific markets. For individual publications, those audiences were always difficult to capture. But on a global level, as you accumulate these kinds of audiences in bulk, and then apply rich third-party data to those audiences, the value of that inventory increases dramatically.

Q: Any clients you can publicly talk about with metrics?

A: Globally, publishers like Time Inc., eBay, Viacom, News Corp, and LaPlace Media have embraced the Advertising Automation Cloud to work directly with the world’s largest advertisers, agency holding companies, and agencies and have seen spend increase 40X year over year via Rubicon Project’s order automation platform.


Please enter your comment!
Please enter your name here