Despite the fact that real-time bidding is a complex and sometimes confusing space, media buyers and sellers alike continue to flock to RTBs, a trend that experts don't believe will end any time soon.
Real-time bidding, of course, isn't the end-all and be-all of digital advertising, and there are numerous areas for concern.
But is the entire model for how ads are bought and sold via RTBs broken?
According to Konrad Feldman, the CEO of analytics firm Quantcast, the answer is essentially yes.
In an AdAge op-ed, he argues that the CPM model, which just about everyone in the industry acknowledges is as imperfect as it is popular, "is now preventing the online advertising industry from realizing the full potential of what may be one of its biggest game changers -- real-time bidding (RTB)".
As Feldman sees it, buying ads on a CPM basis is particularly detrimental to the value proposition of RTBs because real-time bidding offers the opportunity to price ad impressions on "an impression-by-impression basis".
That's powerful because advertisers know that not all impressions are created equal. "Some consumers are more likely than others to purchase certain products" he points out, and he observes that seemingly small details, such as the time of day, can influence an impression's effectiveness.
If CPM is the problem, is CPA the solution?
Unfortunately, the CPM model can result in an advertiser purchasing ads in a less-than-optimal manner. According to Feldman, "In some cases, higher-priced impressions have been 100 times more effective at influencing consumer behavior" and while "this doesn't mean that a buyer should simply pay a higher CPM," it raises the point that many advertisers focus more on the cost than the benefit received in the CPM environment.
So what's the solution?
Feldman suggests that the CPA model would be a dramatic improvement that eliminates some of the barriers he believes the CPM model has put in front of RTBs. "CPA simplifies the transaction and enables the advertiser to take full advantage of the dynamics of a real-time marketplace".
The need for speed
Moving to a CPA model might look good on paper, at least from certain angles. But is it realistic?
Feldman notes that there are challenges. CPA requires a closer connection between advertisers and publishers, and that's not necessarily going to be easy to forge. Attribution complicates this incredibly, because a publisher which delivered the last click will likely want full credit for a completed action while other publishers which delivered earlier impressions will likely claim that those impressions contributed to the conversion.
Perhaps even more problematic is the notion that all of the parties involved will need to estimate the value of every impression and make a decision "in milliseconds." Feldman says that this "requires large volumes of data, the algorithms and processing power to make sense of that data, and a platform with the scale to apply those models to billions of buying decisions every day."
That sounds like an attractive proposition for VC-backed ad tech startups, but for advertisers hoping to reach customers and publishers hoping to move their inventory, it sounds like a potentially costly pipe dream that will likely never be realized.
Which raises an important question for RTBs: should they really be treated as platforms for real-time bidding, or should they be viewed as a means to buy and sell digital impressions more fluidly and efficiently?
Perhaps there's no right or wrong answer to this question, but for the advertisers that treat real-time bidding as the ad world's equivalent to high-frequency stock trading, it's important to realize that when moving fast, losses can be magnified as easily as gains.