With ad pricing from ad networks changing constantly, publishers not only have to keep a pulse on revenue from impressions, but also on which ad network is delivering the best value. In fact, publishers can benefit greatly from diversification in this dynamic environment. PubMatic, an online ad optimization service and TechCrunch 40 company, is launching a new product, Ad Price Prediction, that matches the optimal ad network with every ad impression with real-time functionality.

Ad Price Prediction’s algorithm has been developed using data collected by PubMatic’s ad optimization service over the past three years. The company processes more than 100,000 data transactions per second through ads served for 6,000 publishing customers, and has applied this data to its technology to enable real-time processing.

PubMatic offers a meta ad server that sits between online publishers and online ad networks like Google AdSense, Yahoo Publisher Network and Value Click. Their service helps publishers manage and maximize their advertising inventory by communicating with multiple ad networks to help them find the optimal ad layout and the highest paying ad network. They also provide users with a central dashboard to track all their ad networks and ad configurations.

Ad Price Prediction’s algorithm compares the price difference between behavioral targeting and contextual targeting on a single ad impression and then serves the publisher the higher paying ad. PubMatic says that its premier customers, publishers with more than $5 million in annual online revenue, were able to lift ad revenue by as much as 70% using Ad Price Prediction.

Competitors to PubMatic include YieldBuild and the Rubicon Project. AdWhirl also does dynamic ad optimization, but only for iPhone developers.