Counterpoint: CPM Shouldn’t Go Anywhere

Clicks, views, engagement and other “performance” goals of digital advertising are not end goals. They have no value unless they impact the real business goals positively. Unless significant causal or correlative connections can be made between these measures and real business goals, these measures are simply overvalued currencies in which advertisers are speculating for the lack of effective measures. Unless these connections can be discerned, it is better to stick with CPM.

Contrary to surface-level assessments, CPMs have a better chance of correlating with real business performance, at least at this nascent stage in programmatic media buying. Advertisers have bought on CPM for many decades and have developed an intuition for what types of media are most effective. This intuition, based on analysis, inference, and experience, is certainly not perfect. But it is based on many years of experience, and within it is contained wisdom.

There is a reason that some Web properties consistently command higher than average CPMs; they can. Advertisers will pay for them because in their expert opinions, advertising on these properties is worthwhile. It’s the sort of effect that occurs in an English auction, in which bidders orally submit bids, competing against each other, until the highest bidder wins. These types of auctions typically result in the highest selling prices among the traditional auction formats because the open competition relaxes the apprehensions that the bidders may have. When they see other bidders competing for an item, it reduces the perceived risk in their minds that they may be overvaluing the item. If many bidders value the item highly, then there must be something to the item even if no one bidder can totally quantify its value.

Similarly, the ability to command consistently high CPMs for media/audience bundles can be achieved only if a significant number of advertisers are willing to pay those CPMs over a long enough period of time. This confidence does not occur in a vacuum.

Clearly, measuring performance in this manner is not perfect. However, clear correlations can be drawn very easily over a long enough time frame. One can simply purchase a bundle of media at a given price, measure its impact on key business metrics over a significant time frame, and then optimize the bundle iteratively. Programmatic buying makes this easier. This experiment is much harder, if not impossible to conduct for some “performance” metrics given the attribution issues that persist. For others, there is no point, given that if, say, engagement leads to improvement in some business metric, then the CPMs for the highly engaging media will increase over time anyway. At that point, it’s just a matter of whether one likes the pretty pictures on the CPM note, or the engagement note.

The conversion should be happening on the programmatic buying end of the market, so that publishers can simply focus on their CPMs, and the market does not experience currency-conversion costs in the form of friction. In any case, the ad tech/ programmatic buying market should not waste its time on paper transactions. It should rather figure out whether any of the currencies is actually backed by something real, as in real business impact for advertisers.