• by Diaz Nesamoney , Op-Ed Contributor, September 13, 2012

Measurement remains one of the most pressing issues in the online advertising industry -- yet no one can agree on the best approach. Meanwhile, as the debate carries on, too many online ads are being served up to an empty theater: A recent comScore study found that of 12 big brands advertising online, 31 percent of the 1.7 billion impressions sampled were delivered but never viewed. 

The lack of eyeballs is a key reason for the industry’s ongoing dissatisfaction with cost per click (CPC), along with the inability to measure today’s engagement-oriented campaigns. CPC’s limitations have led to the proposal of new measurement models such as cost per engagement (CPE), which bases cost on how a user engages with an ad, and cost per view (CPV), which was proposed by TubeMogul and endorsed by Google and bases cost on whether a user chooses to initiate an ad. But just like CPC, CPE and CPV have drawbacks. 

First, CPE is difficult to guarantee, it’s impossible to predict the level of viewer engagement and the definition of “engagement” itself is difficult. CPV, meanwhile, gives advertisers a stronger tool for measuring the effect of an ad, and a pricing model that corresponds. But CPV is unable to measure whether a viewer watched a full in-banner video without clicking, which means that publishers who place auto-play in-banner video above the fold in premium positions are losing out if they don’t get paid just because a user didn’t click to play. This is akin to saying advertisers will not pay for a TV ad impression if you walked away from your TV to the kitchen to grab a beer, how would they know?

These are not the only options, however. A promising approach is cost per viewable impression (CPVI), which is similar to TV’s measurement of gross rating point (GRP), and this is attracting advertisers who can familiarize themselves with the ad buy. CPVI tracks code on Web sites that identifies whether an ad was served above or below the fold, and can calculate a viewed impression.

MSNBC.com was one of the first publishers to start pushing for this new measurement model. But industry acceptance has been slow. One reason is that adoption of CPVI could raise agencies’ accountability for not reaching targeted audiences. The industry also would have to be more transparent about whether an ad was actually viewed by a user. Today, ads are counted according to “served impressions,” which can result in substantial over-counting of impressions. This has led to a decline in cost per impression, as well as the creation of more ad inventory at lower cost.

Convincing publishers to cut available inventory is obviously a tough sell. But let’s consider another perspective: If fewer ads are published, it could be argued that the impressions they make on viewers may be more valuable. Also, this will enable publishers to sell ads at a premium. And even if the industry is reluctant to move to CPVI, the writing is on the wall: The IAB wants to make viewable impressions a standard and count real exposures online. (See #1 of the IAB’s five Guiding Principles of Digital Measurement.). Also, as mentioned in an earlier article, CTR is clearly dead, so the CPC model of pricing is on its way out at least for branding campaigns.

Other arguments for rallying around CPVI sooner than later are as follows: The industry needs to think about which measurement will be most adaptable to work across any screen. In my view, CPVI is a move in the right direction, although we should still consider looking at valuable measurement beyond this that tracks engagement and purchase intent. And while it’s not as predictable as the old standard, CPC, I believe measuring viewable impressions will result in better targeting and less clutter for users. The effectiveness of an ad and giving consumers what they want is what matters most, after all.

There is also an increasing sense that perhaps we need to separate media pricing models from media measurement metrics. Media buyers tend to gravitate toward simplicity in pricing and buying media, which is why GRP and CPM remain pricing models by which the bulk of media is purchased. Attempting to tie engagement metrics to pricing models results in a confusing plethora of models which makes media buying very challenging. That’s a subject for another article.