If you work in the online advertising industry, you’ve probably heard the debate in one form or another. In one camp: those who believe that display ads are an important marketing tool, even when no one clicks on them. In the other camp, those who believe that unclicked ads are essentially worthless.
In Chango’s white paper, “View Through Attribution Exposed: What last touch isn’t telling you,” this debate is tackled head on, giving you both sides of the argument.
What is View-Through Attribution?
When an ad impression appears on your browser, it often drops a cookie. Thanks to this cookie, an advertiser is able to identify which visitors to its site first viewed a display ad, even if the user has never clicked on an ad. This is known as a view- through visit, or post-impression visit. The ad that led to the visit is the view- through impression.
For most marketers, the value of a view-through impression is clear enough. After all, unclicked display ads work the way off-line ads have always worked. When a brand puts an ad in a magazine, no one can click on it. But that doesn’t mean such ads don’t work. Off-line advertising might not be a perfect science, but advertisers know they work because they see powerful correlations between their campaigns and new conversions.
And view-throughs are actually more easily measured than most off-line campaigns. When Target puts a billboard up on the highway, the store doesn’t know which customers saw it. With view-through, you at least know which customers saw your impression.
So, what’s to debate? The view-through skeptics point out that view-through metrics can be misleading. If someone sees an Amazon banner and visits Amazon a few hours later, there’s a very good chance the ad led to the site visit. But what if the visit took place three weeks later? Should an ad impression get credit for a visit it might have had no connection to?
The critical issue here is the view-through window, which is the length of time after the ad appears that a visit to the site is counted. When the view-through metric was first introduced, the default window was generally set to 30 days, meaning that if someone saw an ad and then visited the site within the next month, it would register as a view-through. But 30 days is far too long for most campaigns. And, to this day, many marketers aren’t aware that view-through windows can be set for as little as 24 hours.
And then there’s the fraud problem. In the early days of view-through, it wasn’t uncommon for scammers to buy up millions of super cheap impressions so they could take credit whenever a user ended up on a site that was running a display campaign. Fortunately, this practice, known as “cookie stuffing,” is now quite rare. But the damage it did to view-through’s reputation lingers.
In light of view-through’s checkered past, some skepticism is certainly understandable. But there’s an important point that many of the skeptics are overlooking. Sure, it’s true that view-throughs are not a perfect science, but view-throughs can account for over 90% of site visitors and over 90% percent of pageviews once a view-through user arrives. And that means the view-through is actually a much more reliable metric than the click-through. After all, just because clicks can be counted doesn’t mean that they should be counted.
According to a study by comScore and Starcom MediaVest, 84% percent of Internet users never click on an ad, and the clicks from the remaining 16% don’t correlate to actual on- site conversions. View-throughs, meanwhile, correlate with increased search traffic and revenue.
The key to using view-through is setting the right window and using the right measurements.
In Chango’s white paper, we explain how to determine the incremental value of exposure to unclicked ads. As the white paper makes clear, we think this is one debate that can finally be put to rest. Fortunately, the world of online advertising has plenty of others to keep things interesting.
You don't need to remember this website again. Just subscribe to updates on email.
Google Reader is long gone, anyways!