Ad Industry Moves Analytics Toward Forecasting Demand

by Laurie Sullivan, Nov 28, 2008, 7:00 AM

• Comment
• Recommend
• Tweet
• 

Publishers that are looking for a method to visualize their inventory and understand the constraints of selling to advertisers will soon have an application that theoretically forecasts availability and price.

The application will officially become available in January through the start-up Yieldex, a San Mateo, Calif.-based Web-services company founded in 2007 that analyzes and predicts available online advertising inventory to help optimize campaigns. Three unnamed companies have been testing the platform for six weeks.

Targeting ads to specific demographics, bundling two or more media, and collecting registration information to package and sell ad space have created complex issues. These supply-chain problems are unique to the advertising industry, but the underlying problem is not.

Many companies have tapped analytics that attempt to forecast demand for parts and service in industries from electronic component manufacturers such as Intel and Texas Instruments, to enterprise resource planning software providers Oracle, Microsoft and SAP.

While forecasting has never been a perfect science in any industry, the application should help publishers sell and keep better track of ads, especially as available "premium" inventory tightens.

The Yieldex application aims to set the advertising industry on a similar path by providing an application supported by Web Services that can analyze and automatically sort through raw data based on content, quantity, market price and demand.

"Media companies facing supply chain management problems don't understand supply because inventory is volatile and complex," said Larry Allen, Yieldex president, who estimates that large publishers could have tens of thousands of ad spaces for sale. "Many times someone is sitting in the back room reading numbers off a spreadsheet to the salesperson about how much inventory is available and where. The salesperson oversells ads because analytical tools are not in place to keep track of inventory."

Allen admits that in a test with one of three companies trialing the Yieldex platform, the unnamed publisher's ad server predicted 40% more available ad inventory then Yieldex; however, the Yieldex application forecast 5% more than actually available.

Yieldex's application relies on cloud computing from Amazon, which provides the processing power. Individual publishers collect massive amounts of log data daily because they have a large number of transactions on their Web site. Each transaction via ad impression contains a variety of information such as what the consumer read and targets associated with the ad. The log file information is fed into the Yieldex application and processed in parallel on many machines through Amazon EC2, cloud computing Web Services. From the data, Yieldex can characterize each piece of information and examine the permutations to determine the available inventory and how it interacts. The combination of the processing power through Amazon and the Yieldex application makes this possible.

Allen, a former senior VP for network development at AOL's Platform A, said Yieldex solves a specific problem in the advertising industry that has existed for more than 10 years. "No one has gone after solving this problem in the advertising industry because it's a very difficult math problem--very complicated," he said.

The underlying engine is complete. The front end, or user interface, still needs work, Allen said.

Recently, Yieldex took first place in the Amazon Web Service Start-Up Challenge. Not only did it win the $100,000 cash prize from the second annual contest, but winning meant the judges deemed the Yieldex solution an original way of meeting a marketplace need using AWS. The Yieldex app was selected from many start-ups, all of which built solutions on the Amazon Web Services platform.

• Comment
• Recommend
• Tweet
• 

Be the first to comment on "Ad Industry Moves Analytics Toward Forecasting Demand"

Leave a Comment