Advertising News

As the ninth installment of our quarterly market report declares, the state of the digital advertising market is good–and growing! Through Q1 2010, the Rubicon 20 Index, measured through proprietary algorithms that track performance across a roster of 20 of the Web’s most heavily-trafficked properties, displayed 25% growth year over year. And merger and acquisition activity in the online advertising sector grew 28% compared to Q1 2009, as technology and innovation continues to drive growth in the industry. In this report, we evaluate the emerging technologies and developments in the space, including DoubleClick’s relaunch of DART for Publishers (DFP), the impact of RTB; and the increasing threats posed by ever more sophisticated versions of malware:


[Excerpted from the 2010 Online Advertising Market Report: Q1 Emerging Trends & Outlooks]

Brands and revenue are at risk

Tracking and stopping malware has been a nearly impossible process with many publishers still unable to detect and stop malicious ads before the a flurry of consumer complaints start coming in, by which time the brand’s reputation has already been damaged. A few more examples of malicious attacks on popular websites and ad companies include:

  • Yahoo Top Ad Malware Distributor, Says It’s Not Their Problem, according to TechCrunch
  • CA Predicts More Malvertising, Mac Attacks in 2010, as reported by eSecurity Planet
  • Avast Indentifies Malvertising Strain, according to BizReport

Customers expect Web sites to protect them. Of course, this would not be such a problem if endpoint security suites protected end users but that is not the case. Today the products on consumer and business desktops, like those offered by market leaders Symantec (Norton) and McAfee, are struggling in this modern-era of dynamic and polymorphic threats and simply do not offer sufficient protection against these new, unknown, and fraudulent attacks on users.

“For publishers, advertising is about making money, but malicious ads change the equation. Publishers need better solutions to protect their customers from malvertising and the potential for malicious content on their Web sites,” noted Rob Lipschutz, co-founder and CEO of SiteScout, a company whose technology protects publishers against malicious ads and other dangerous Web content.

“The advertising ecosystem faces a stiff challenge and the problem is widespread and found in both direct advertising as well as more distributed ad networks. New ad formats also make the problem increasingly complex.”
Given that much of the ad serving technology still on the market today was developed in the late 1990’s, those products are not equipped to deal with today’s huge volume of different buyers and sellers from around the world, which creates a constant stream of new vulnerabilities in the system.

Agencies and advertisers have their own way to tag ads and publishers don’t have control, as there are thousands of different ad tags running on a publishers’ site at any given time. These disparate systems have had no universal quality control because nothing is tied together, driving the need for automation and technology innovation to eradicate the vulnerabilities of this process.

The need is clear for a solution aimed at publishers and advertising companies, the producers of content, rather than end-users, that provides visibility and advanced protection against the new kinds of attacks to prevent direct loss of revenue or risk to brand (leads to loss of revenue). In January, the Rubicon Project launched Rubicon Security, its first foray into protection against malware attacks on publisher customers’ sites. The issue of malware will only increase as a key risk to publishers’ advertising businesses – and to the consumers who drive those businesses with their patronage – in the months ahead.

To read the full Q1 2010 Online Advertising Market Report, as well as past published reports for free, visit: http://www.rubiconproject.com/market-intelligence

Share

{ 0 comments }

Fly on the Wall is a series that spotlights a Rubicon team member with the hopes of providing industry insight and an in-house look at the company’s latest developments.

The mad scientists behind the Rubicon Project’s optimization and data technology aren’t so much mad as they are overachievers. In addition to working to keep the internet free and maintaining a pace of constant innovation, our very own Neal Richter has been passionately pursuing his research on Artificial Intelligence, balancing school with his sideline hobby: creating ground-breaking systems for data analysis. The ultimate multi-tasker already has 3 patents under his belt with 5 more on the way. We asked Dr. Neal Richter to take a few minutes out of his busy schedule to update us on some of his recent accomplishments.

neal-richter


So what’s this about two-timing us?
I’ve just defended a PhD at Montana State University in Computer Science. I picked MSU both since my family is 5th generation Montanans as well as wanting to flyfish more than fight traffic. The dissertation is on evolutionary algorithms, a process of simulating evolution and competition to ‘grow’ computer programs to solve problems. I used some mathematical techniques to prove and demonstrate a few new things about this process. I also worked full-time through school, building artificial intelligence and data mining software for various companies.

In your 13 years of working in the online technology industry, what are your most interesting victories?
My favorite might be building an emotional analysis of text system at RightNow Technologies. We built version 1.0 in 2001, and it’s been churning away, processing untold billions of emails, customer service requests and now tweets. We did it with a team of 4 people over 6 months while working on other tasks. There have been entire companies created and sold doing the same thing since then!

What are you working on now at the Rubicon Project that has you all fired up?
I’m working on a few projects, all of which revolve around using data to drive publisher yield within the REVV technology platform. To start, the platform throws off mountains of data. We mine that data to feed our core AdEngine with pricing and relevance information and set up its internal “ad tournament”. We were hard at work before the REVV for demand launch last week, finalizing the core algorithms behind the new real-time bidding capabilities within our platform. In the past, I’ve worked on the Data Intelligence stack, which you can think of as a combination between a web-analytics product and a semantic indexer. It finds and measures the intensity and frequency of topical browsing patterns on publisher websites.

What is your typical approach when you take on these kinds of projects?
There are really two answers to this type of question: the tools used and the methods applied. The methods question is more interesting. We’re using various machine learning algorithms, combinatorial and numerical optimization techniques, natural language processing and pattern mining, just to name a few. Also in play are auction models and pricing-finding algorithms from economics.

With all the time you’ve dedicated to helping the team develop all of these new methods, does it ever take a toll on your academic pursuits?
What we’re doing at the Rubicon Project in the core stack is really called computational advertising. This topic exists at the nexus of optimization, machine learning, data mining, micro economics and information retrieval. There are more than a few interesting problems embedded there. Getting a PhD (while working full-time) was challenging but immersing myself in these sub-topics has been very beneficial. I don’t think I’d be prepared to tackle these industry problems the right way without all this research through the university. For instance, one part of the Rubicon Project system forces ads to compete with each other based on relevance and price. My research on evolutionary algorithms (making solutions compete for resources and survival) really helped to illuminate the best practices on how to reconcile that in a non-naive way.

What technology must still be developed to increase digital advertising performance for publishers?
The core problem for publishers in display advertising is lack of information. They have an impression to sell and want to market it in a value-added way to maximize revenue. Here’s a simple analogy: imagine you are a barley farmer in the grain belt of Montana, and you have a crop to sell. You have direct contracts for some of that crop but for the rest, you need to find a buyer. Now imagine that you have no idea what other barley farmers are getting per bushel nor who is buying it and for what purpose. You take your excess crop to a grain-exchange and they cut you a check. Is it fair price? Who are the down-stream buyers? How can the crop be segmented and sold efficiently (brewers, for example, want low protein barley versus the food industry, which seeks high protein). If your barley is organically grown, how do you find the best buyer who is willing to pay for that added value?

This highlights the broad strokes of the problems encountered by publishers today. To overcome these obstacles, they need:

  1. Information on the demand for their inventory,
  2. Comparative pricing data;
  3. Technology to segment their inventory;
  4. Mechanisms to protect the value of the segments

The inventory a publisher does not sell directly needs to be protected from being offered as low-value remnant on advertising exchanges. To do this, they need an automated and scalable system to act in their best interests. At the moment, the vast majority of ad-exchanges are biased to benefit the buyers as they expose remnant inventory to be sold cheaply. Currently, the information bias is mostly on the side of the buyer and sellers have little to no leverage on their data. Econ 101 tells you that when both parties have unequal information, then the transaction is not always fair.

the Rubicon Project’s technology is addressing these four core needs and we’re confident that we’re doing it the right way, with the right data, algorithms and tools.

What research and insights have you made thus far that give you hope that this technology is achievable?
One insight that jumps to mind is related to real-time bidding and the popular second-price auction. It’s common knowledge in the industry that this mechanism is great for fairly allocating opportunities to buyers. However, it’s less well-known (outside of economics literature) that this method is not necessarily optimal for sellers. One flaw is its susceptibility to collusion among buyers, and explicit collusion is not necessary. The implicit collusion of the buyers having information that the seller does not exposes this flaw.

The simple addition of pricing intelligence available to all parties helps fix this flaw. the Rubicon Project’s engagement with all real-time bidding platforms and DSPs will enable publishers to gain valuable insight to market pricing as well as force those exchanges to compete with all the ads in the REVV Marketplace.

You seem really excited about what the Rubicon Project is doing for publishers–is it safe to assume that even with your doctorate now securely in tow, you won’t be deserting us for the world of academia?
I’ve been here just about 1 year and it’s fantastic. I love a big challenge and the open opportunity in front of us. We have a great team of engineers that are picking up steam. I’m certain there’s not a better job to have at the moment!

It’s been a great year since you’ve joined us, Neal, and we can’t wait for all the innovation to come in the next!

Share

{ 0 comments }