"Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Arthur Hainline, director of analytics at Bidtellect.
Real-time bidding's emergence promised a lot of efficiencies and advancements in digital advertising, one of which was to establish fair prices for advertising inventory. Various demand-side platforms (DSPs) and their bidders bid on equal footing in the open market and assign CPM prices that accurately reflect the value of ad placements.
In theory, a bidder should easily be able to establish the quality of a placement, bid at the rate that reflects that quality and win a fair percentage of that inventory. Trusting in this process, marketers are relying more and more on audience data to improve the value of their ad dollars.
Unfortunately, the appraisal and pricing of ad inventory in the open market is not a simple matter. There exists no universal metric used to establish the quality of ad inventory; how each ad placement’s quality is measured varies wildly between various DSPs, clients and agencies, all of which have different goals. No amount of first-party data, third-party data or other clever targeting tactics can be effective without a proper evaluation and understanding of the inventory.
Consider a placement on the home page of a website.
It may be viewable more than 70% of the time and have a high click-through rate. Bidders highly value this placement on behalf of clients with upper-funnel goals, especially in a media landscape where viewable impressions are at a premium.
This same placement, however, does not tend to generate engaged users willing to make a purchase at a high rate post-click. The highly viewable impressions are a bit intrusive, and some of the clicks are unintentional. The rate set by the open market on behalf of upper-funnel clients exceeds the rate that a lower-funnel valuation justifies.
Now consider a second placement on the same website near the bottom of an article page. It is viewable less than 30% of the time and has a low click-through rate. Because this placement is not desirable on the basis of clicks and viewability, its ability to convert users post-click can determine its market rate. Bidders buying on behalf of clients with lower-funnel goals establish the price for this placement, while bidders buying on behalf of upper-funnel clients are left unable to win inventory at a rate that backs into their clients’ goals.
The problem illustrated in this example mushrooms when contemplating not only the monumental scope of programmatically available inventory, but also the numerous, distinct and sometimes subtle goals of brands who are purchasing it.
However, the vast amount of inventory available and extensive number of auctions occurring each second present an important opportunity. Billions of data points can allow for more accurate inventory evaluation and predictive modeling. Optimization technology can express a brand’s discrete goal in the form of intelligent dynamic bid prices and inventory selection, even within any post-targeting subset of available auctions.
While the potential value that user data and targeting offers advertisers should not be overlooked, those are not sufficient standalone strategies for smart media buying. In a landscape where brands and agencies have increasingly sophisticated and unique goals, the importance of decision-making in programmatic media buying cannot be understated. Understanding the nuances of programmatically available inventory is critical, not only to bid at the correct rates for specific brand goals, but also to bid on the appropriate inventory where those rates can find the best value.