Injecting Viewability in Conversion Attributions

As my industry colleagues know all too well, success of direct response (DR) campaigns are inevitably measured using cost per acquisition (CPA) metric and media partners – or ad tech vendors providing data/algorithms to buy media – who have the lowest CPAs get rewarded with more budgets and the ones at the other end of the spectrum get booted off the media plan. It’s that cut and dry!

Clearly, everyone on the media plan is extremely motivated to be the media partner who brings home the acquisitions or conversions. To be attributed a conversion, a) a user must click or at least see an ad from the media partner, b) and then the user must complete the conversion action within a designated window, typically 7 to 30 days. But these are necessary, not sufficient conditions.

As there are several media partners on any plan and a typical user is browsing increasingly higher number of web sites, it’s likely that multiple media partners have managed to serve ads for exactly same campaign to her before her conversion action. Then who gets the conversion credit? Conversion attribution is all about deciding answer to that very question!

Conversion attribution is a topic into itself with different methodologies with varying levels of sophistication that I cannot adequately cover here. But for an overwhelming number of campaigns attribution is given to the media partner who served the last click (higher priority) or view prior to the user conversion. Since clicks are relatively rare (shocking!) the partner who generated last view invariably wins the conversion derby. So what does a media partner do? It will keep showing ads to likely converter – mercifully, most stop when a user converts – with a hope that the user will indeed go on to convert and it will get the credit.

Given the ‘last view’ is so pivotal to conversion attribution, it’s worth asking a simple question: What exactly counts as an ad view?

Unfortunately, it’s not what any rational person would think it should be. An ad view is an ad impression delivered to a browser or a device – whether it was on the portion of the page that was in user’s view or not is, alas, immaterial for the purposes of attribution. This counting/attribution methodology encourages media partners to keep serving ad impressions to likely converters even on ad slots that seldom, if ever, get seen just to maximize their chance of winning the ‘last touch’ conversion attribution. In this dogged pursuit of conversions enormous media budget is wasted and ‘better price for better quality inventory’ equation simply goes out of the window.

Can online ad industry do better than that? Absolutely! How? By making Viewability a required condition for conversion attribution. There are some companies (checkout C3 Metrics) who have highlighted this issue and even productized a solution but I don’t see a broad acknowledgement of the issue and adoption of Viewability based conversion attribution.

I can imagine these types of solutions facing market resistance in the past because Viewability was not a standard metric. But earlier this year, Media Rating Council (MRC), with support of the Interactive Advertising Bureau (IAB), approved the use of Viewability as a metric for display advertising. So far, the ad industry seems to have viewed this development with distinctly ‘brand’ advertising angle. Viewability is, of course, a no-brainer for brand advertising. Why would you ever pay for an impression that no one saw and therefore had no effect? But I would argue that it can be, in fact should be, equally applicable to performance marketing. Think about it. Why an ad that was never seen by a user should ever be considered for attribution? It just doesn’t make sense! Impressions that are not visible should simply be discarded from attribution considerations.

What more needs to happen for Viewability to be an integral part of conversion attribution modeling? In my observation, it’s the 3 party ad server that acts as the referee deciding who to grant a conversion. So I believe that until DFA, the de-facto ad sever used by advertisers and agencies, injects Viewability in attribution modeling it simply won’t happen.

So it all comes down to Google (DFA is part of Google’s product suite from DoubleClick acquisition) taking the lead and set the right example. All it will take to bring sanity in attribution modeling is for DFA to switch from ‘last view’ to ‘last actually viewed’ model. Once DFA layers in Viewability in conversion attribution, other ad servers will follow suit, IAB and MRC will, I imagine, throw their weights behind such a move, and Viewability technology vendors will be more than happy to lend their technology to make it happen. And the online ad industry and ecosystem will truly be better off with this change. Anyone listening at Googleplex?