Ah view-through! Rarely does a topic in our industry solicit as much debate as this seemingly innocent metric. Display planners rely on it, search marketers don’t trust it (but would secretly love to get credit for it from PPC), and the clients are confused and uncertain.
When 3 CMOs all asked for help on this topic within the space of a single week, I decided we needed a new resource that looked at the arguments ‘for’ and ‘against’, and so have published a new white paper on the subject: View-Through Attribution Exposed: What Last Touch Isn’t Telling You.
In order to reach a conclusion, we looked at data from comScore, my own published studies when I ran media at iCrossing, a new TagMan / IAB study, as well as opinion pieces from organizations like Salesforce Radian6 and Chango‘s retailer clients.
In this summary article, I want to explain view-through from the perspective of a search marketer and give you the answers you need to grab more display dollars, and be the informed voice for your bosses or clients.
Look Away Now…
Not to spoil the ending, but I believe in view-through under the right circumstances. The right answer is to not give 100% view-through credit, but nor is it to give 0%. Somewhere in the middle is a happy solution we can all work with, as we will discover.
The Basics: What Is View-Through?
View-Through, also know as post-impression, or just VT, refers to when an individual is exposed to an ad campaign, takes an action such as visiting or site a making a purchase, but did not click on the ad itself. As hard and complex as our industry tries to make it, that’s the sum of it.
We see VT in many aspects of marketing; perhaps the most accepted is in TV, radio or outdoor advertising.
Companies spend billions talking to their audience using these channels, more than in digital, and even though they can’t directly see an engagement path from billboard to store purchase, they continue to spend. Offline advertising relied entirely on their equivalent of VT.
Search vs. Display – The Click vs. The Impression
One of the reasons search marketers don’t like or understand view-through is because it does not exist as a metric or a capability in the PPC world. With PPC, we see a searcher and can choose to respond to that person with an ad, and that individual either clicks on it or they don’t.
The click is a good measurement in PPC for two reasons – first, it is really our only metric, and secondly because the message is only shown when an individual asks for it (with their search), and so as part of that interaction the click is the final word.
Display is very different of course – unless you are looking at a tactic like ‘search retargeting’ then the ad is not being shown because it has been requested, it is being shown under best-guess circumstances that it fits that individuals wants.
Display is not a response to a question; it is there to influence the question to be asked. The click might fit the search world but it does not fit the display world.
The Click Means Nothing In Display
It could be argued that the click not being a valid metric for display is merely opinion, but what is not just opinion are the well-documented studies that demonstrate the click is a random, inconsequential event in the display media world.
It turns out that those people who click on display ads are often the youngest and oldest members of the online community, less digitally experienced and of lower income (if you are a high-end credit company targeting the 25-45 year olds, the click is the exact opposite of what you should be measuring!).
Most clicks are now the responsibility of less than 16% of Internet users.
In one study, the demographic profile of click convertors was shown to match the national average of the US population rather than be correlated to the typical customer profile on which the conversion took place. Conversely, the VT convertors matched the site’s profile perfectly.
Consumers simply don’t click, and so VT is our default primary metric.
OK, But What Percentage Of VT Makes Sense?
There are many ways to do this, but the most common is to use a cookie to correlate the ad impression with the subsequent conversion – this can be done with an ad server like DoubleClick, or using a media vendor’s own pixel. Next, the marketer must understand what percentage of the VT revenue is an accurate number for his or her own campaigns.
Clearly, some individuals would have come to their site whether the ad was shown or not, and so the purpose of the study is to work out how many are truly incremental.
This is typically achieved with a PSA (Public Service Announcement) test, of which more information can be found at:
- 3 Simple Alternatives to Attribution Modeling for Search Marketers (Search Engine Land)
- Viewthrough & Incrementality Testing (Domenico Tassone)
Display’s Other Impacts
Beyond the measured impact within its own channel, display has also been demonstrated to have a real and calculable impact on PPC, SEO and direct load traffic.
During my agency days, I was part of a team that evaluated this using real client data that was entirely under our control, giving me the confidence that we were cutting through the BS of such studies and were proving something concrete!
What we found among other things was that a well-structured display program could increase the CTR in PPC by as much as 15% for a national travel brand, and lowered the CPC cost by over 11%. Further, the study showed a 13.7% increase in natural search clicks and a small uplift in overall site traffic.
An Atlas study speaks to similar results showing a 4x lift in the overall program when adding display to search, and comScore can demonstrate more than a 50% increase in branded searches in PPC when using smart display buys.
So What Did We Learn About View-Through?
- That VT is not a small topic, but is far less complex or mysterious that it is made out to be
- Display is very different to PPC and requires its own metrics to understand its success
- The action of a click in display is actually misleading and will lead to bad campaign decisions
- VT is often not trusted because it is not just about the math, and in the past the wrong VT windows have been used and studies have been hard to run
- Smart use of PSA and / or overlap reports will help generate the right answer for your own circumstances
I encourage discussion and debate on this topic in the comments below or do reach out to me if you wish to discuss your circumstances. In addition, check out the white paper and also this recent SEMPO webinar (Retargeting Exposed) where we discussed some of these attribution model concepts live.
Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.