????By Kim Thai

????Rajeev Goel, CEO of upstart PubMatic, thinks his company can help print publishers recapture advertising revenue from their glory days.

????Print media is dead! Yawn.

????You’d think media analysts and bloggers would find another catchphrase. This executioner’s call is as tired as Jon and Kate’s tabloid tussles.

????So when Rajeev Goel, co-founder and CEO of PubMatic, told me that not only would print publications survive, but he knew how they could, he definitely got my attention.

????PubMatic offers what it calls "real-time ad price prediction technology." In other words it lets publishers of premium content (read: traditional magazine and newspaper companies) decide in real time which ad networks on which to sell unused advertising inventory.

????Others, including Google's (GOOG) DoubleClick and the Rubicon Project, make similar promises. But Goel touts that PubMatic is the only advertising optimization company completely devoted to publishers, pushing his close competitors into a separate category. Not to mention he’s doing this all in real-time.

????Still not convinced? I wasn’t either. But Goel says his two-year-old company has already been able to increase publishers’ advertising revenue from 30% to 70%. (In one case, PubMatic was able to catapult revenue by 300%.) He only charges a 15% commission fee.

????We sat down with Goel to see how he thinks PubMatic can help the ailing print industry.

????Fortune: What trends and mistakes do you think publishers are making in choosing ad networks?

????Goel: As people spend more and more time online, there’s a huge growth in the supply of inventory, which is changing the pricing dynamics. Second, publishers typically have a wide audience in terms of their global reach of their readership — and that’s typically an under-monetized opportunity for the publisher, where they might not be focused on selling that ad inventory. And third, a place where we place an increasing amount of importance is in ad quality. We’ve all seen the not-so-desirable ads online and I think typically the large premium brand publishers are looking to not only maximize revenue but also protect the user experience.

????We’ll source the right networks and then pass information to those networks in real-time about the audience that’s on the website — such as household make-up and income of the users and other demographics — so the right ads can be targeted.

????Fortune: Many have tried to take the route you are doing right now and have either failed or changed their business model. What makes you think you’ll succeed?

????Goel: Where most others have failed is that they felt they could deploy a couple of engineers on the problem for six months without a clear understanding of what is required to be successful. Our ongoing technology investment and investment to-date are a significant competitive barrier that makes it very difficult for most newcomers to be successful. What’s unique to PubMatic is our patent-pending Ad Price Prediction technology, which looks at every impression in real-time along with dozens of factors and predicts how each ad network or exchange will price that impression. We do this for every single impression on a unique basis in less than four milliseconds. This is what is unique relative to any other solution provider, including DoubleClick or the Rubicon Project.

????Fortune: What’s the advantage of being real-time?

????Goel: We are consistently able to outperform non-real-time solutions, such as publisher’s in-house ad operations team, through our real-time optimization. One of the latest trends in online advertising is known as “real-time bidding,” in which a buyer of media will evaluate an ad impression in real time and determine what they are willing to pay for that ad impression. The vast majority of publishers can earn approximately 50% more revenue from real-time bidding advertising campaigns.

????Fortune: Tell us about the technology behind this.

????Goel: The technology is a predictive ad server. It looks at the demand across these hundreds of ad networks and in real-time, it predicts how each network is going to price the next impression that comes in. So when the user goes to the page, the publisher will call our system and our system will look at all the networks that they’re working with, predict which network is going to best monetize that impression — based on the geography, time of day, what the page is about — and we select that network to serve the ad. Doing that, we process over 100,000 data transactions per second. It’s just a huge amount of data to know about the site, to know about the audience, and also know about the networks and in terms of what is their demand.

????Fortune: Has the advertising slump given you an opportunity or, adversely, has it brought down business?

????Goel: Clearly, online advertising has gone through quite a spike and a downward slope in the last couple of years. I think there are a couple of things that are driving that. One is the dramatic growth in the supply of inventory and that’s led to a precipitous drop in pricing. What we found for the industry as a whole is that in’07 and ’08 pricing decreased anywhere from 30% to 40% in annual pricing. So pretty steep declines in pricing. Every month since January of this year, there’s been an uptick in pricing. The pricing hasn’t returned to the level it was to ‘07, but it’s clearly bottomed out. Everyone is waiting for the recovery to take root.

????Fortune: Do you think advertising can support publishers online the same way it used to in print?

????Goel: I do believe with the right model of content creation and the distribution of it, there can be online-only, advertising-based models that work. One of the things we advise publishers to do is to figure out ways to deliver differentiated ad products to your premium sponsors — to sponsors that really want to engage with your audience. If you’re trying to sell the same ad product or the same ad unit via the direct sales force or the ad network, then it’s no surprise that there’s going to be conflict there. But if you can differentiate the ad products sold, then I think you can have a very successful ad strategy.