The first issue, and one that has been well-publicized over the last year or so, is the prevalence of non-human traffic hitting website servers. Depending on who you believe, this traffic, whether from deliberately fraudulent bot networks or innocent (though equally irrelevant) spiders and other web crawlers, makes up anywhere from 15% to 50% of server requests, the implement by which impressions are "served."
The second issue is that even if the site visit is from a real person, there has historically been little or no attempt made to determine if the user was ever able to see the ad itself. For instance, if a page loads but the particular inventory that was purchased is below the fold, and the user then clicks a link in the header to leave the page, an ad impression would be recorded despite the fact that it never appeared on the screen. The New York Times ran an article about this not long ago, which helped bring attention to a wider audience, but largely covered existing topics of discussion.
The IAB's attempt to define "viewable" impressions and instruct the marketplace to deal only in those impressions that fit the description (“must be in the viewable portion of an internet browser for a minimum of one continuous second to qualify as a viewable display impression”) is well-meaning, but unfortunately fraught with problems. Basically, if those conditions either aren't met or can't be guaranteed, they recommend that publishers don't sell that inventory and that media buyers don't pay for it. And it is here that we marketers run into a bit of a conundrum.
As digital marketers, one of our big selling points for years, in terms of justifying budgets, was the fact that digital is just so much more measurable than traditional advertising channels, which in turn meant that we were comfortable being more accountable. Over time, it became clear that display ROI wasn't quite so cut-and-dry as search and other direct response marketing, but we were quick to assure those who control the purse strings that it was ok, because we could still measure the impact. Using a variety of modeling techniques and brand lift studies, we claimed that we could show the benefit of display advertising on other channels, whether it was increasing searches on our brand terms or improving conversion rates on page for those exposed to a display ad.
So now that we find ourselves balking at the fact that we may, in fact, have been paying for bogus traffic all along, we are in something of a bind. Some marketers are suggesting that we should simply not purchase inventory from those publishers and networks that can't promise "viewable" impressions now, but I don't see that as feasible. For one thing, a lot of people selling inventory, especially targeting niche audiences, simply don't have the technology built into their ad serving to track and guarantee such things. More importantly, however, the idea of cutting off a channel for this reason has a real logical flaw to it.
Here's the deal, the product (display impressions) hasn't changed significantly, we are simply more aware of shortcomings than we were before. So, if we are going to suddenly tell those to whom we marketers are accountable (boss or client) that we should no longer be buying certain inventory that we used to be, we are forced to acknowledge one of two things:
1.) The product has always not been worth what we have been paying for it, and thus all of our past attempts to demonstrate its value through various methods have been either inept (bad), or deliberately deceitful (very bad).
or
2.) The product always has always been worth purchasing for all of the reasons that we said it was, and still is, but we are recommending shutting off that valuable channel for what essentially amounts to moral reasons.
Since the whole point of most measurements for display had to do with the results, not the delivery itself, what we should be able to say is something like the following:
"Even with what we now know to be only 65% viewable delivery of purchased impressions, we have still been able to achieve [X, Y, and Z beneficial results] on our display campaigns. While we will continue to work with our publishers and providers and encourage them to improve their delivery methods, recent industry findings have done nothing to change the core value returned by past campaigns, nor do they suggest any reduction in returns on future campaigns, even if circumstances remain the same. As such, we do not recommend any change in current investment in this channel."
Some marketers seem to be tempted to cut off their noses to spite their own faces, but this is a wrong-headed approach. There is a solution here, and it is a market solution. In the same way that demographic information and services like comScore have allowed inventory sellers to specify their audience and separate into tiers that are reflected in pricing, so too will these provably "viewable" impressions command a premium. This new information may give media buyers a little more leverage over pricing with some publishers/networks, but it doesn't give us the ability, or need, to re-write the history of our past results. Think about it this way: if display has been as effective as it seems despite so many served impressions never being seen, there is basically nowhere to go but up.
I for one, however, will certainly be looking for a lower CPM next month from those providers who can't make the '50 %on screen for one continuous second' guarantee.
