Ramsey McGrory, who runs the Right Media Exchange, recently said, “In its current state, RTB is lopsided and problematic for publishers” and cited “Potential data leakage, sales channel conflict and general upending of the publisher sales model.”
Ramsey is on to something. Real-time bidding (RTB) is an exciting new technology, but after two years, it’s become clear that it has strengths and weaknesses. As one of the pioneers of the real-time bidding revolution, I’d like to dig into how far this emerging technology has come and where it needs to go.
2009 was the year of adoption. In January, we saw only a few million ads sold daily on an RTB basis. By the end of the year, we were already seeing a billion ads a day. The initial implementers were fairly primitive. Although some platforms supported premium features like brand block lists to protect publisher channel conflicts, most buyers and sellers were still trying to figure out how to turn on RTB – forget trying to actually understand the impact and how it would truly interact with their other sales channels.
2010 was a watershed year. I’ve never seen a new technology adopted faster than RTB. Over the course of a year, we went from seeing about a billion ads a day to over 8 billion by the holiday season. Companies started to realize that RTB wasn’t just an experimental trend, but a viable channel through which publishers could make serious dollars, using newly available impression-level data. The distinction between “remnant” and “premium” inventory became less important in the face of an emerging category: user-based targeting.
At the commencement of 2011, real-time bidding is a commonly accepted monetization channel, but publishers now realize that RTB poses a whole new set of yield, brand, quality, and channel challenges. To allow advertisers to buy the users for which they have valuable data, a publisher has to expose its entire inventory through RTB, which introduces a set of challenges. In theory, an advertiser can now just “listen in” and collect pricing data, build behavioral profiles, and try to buy that precious premium inventory for significantly lower prices.
So, how do publishers address the main challenges of real-time bidding?
- The sales channel conflict. For example, an agency trying to purchase premium inventory at a remnant price by buying it on a “spot market” rather than through the premium sales team. There are many “knee jerk” reactions to this problem. The most popular one is to set a high floor price to force buyers to pay at least a certain amount of money for a publisher’s inventory. To effectively enable this new RTB channel and allow buyers to have maximum flexibility but also protect the premium channel, very granular price controls are necessary. In certain cases, there are user sets that are known to be highly valuable that should never be sold at a low price. On the other hand, there’s also a pool of high-frequency inventory that is impossible to sell at a premium and is perfect for an RTB marketplace. To strike this tricky balance, sophisticated yield management controls are necessary.
- Brand safety concerns. Microsoft and Yahoo, for example, each have over a billion-dollar premium display business, so they can’t afford to play guessing games with their advertisers. (Disclosure: Microsoft is an investor in my company and we serve as a technology partner to it.) It’s critical for these brand guardians to analyze the potential impact of a real-time buy across various sections of their properties, specific ad buyers, and users/viewers, and each and every creative needs to be audited.
Overall, I believe that real-time bidding is here to stay and offers an incredible new solution for buyers and sellers alike. And in the same breath, I know that almost everyone in the business will tell you that an efficient real-time marketplace needs to factor in brand and channel safety in a manner that ensures there will be no cannibalization of their existing business. If RTB is to truly soar in 2011, addressing these publisher challenges will be critical.