30 Jan Building Accurate Measurement in the Age of Fraud, Non-Viewable Ads and Cookie Proliferation
As Display has moved from a contextual set-it-and-forget-it advertising vehicle to an audience based, optimized machine, precise and accurate measurement has become the priority. This is not an attempt to bring yawns to the faces of many a marketer (often typical on this very subject) but rather, an opportunity to emphasize the importance of measurement to inform improved targeting which results in superior campaign performance.
Accurate Measurement – Key To The Purchase Cycle
As with most initiatives, we expect things to improve over time. But macro trends in our industry’s technologies and in consumer behaviors have made the road to accurate measurement complex. Measurement enables marketers to tell stories across the entire purchase cycle, serve the best performing personalized ads to consumers across devices as well as determine which new audiences to target based on return on ad spend (ROAS). Some of the barriers to accurate measurement reside on the publisher side and some on the advertiser side. This raises the complexity in getting to a solution and most likely necessitates a brick by brick approach to building one.
Filtering out the Non-Human/Bot Traffic
The first step to accurate measurement is for advertisers to implement a fraud solution to fully understand whether real users are viewing the ads. Fraud has the ability to muddy metrics significantly, making it difficult to determine the best performing ads, and wasting ad spend on media buys that should not have been purchased in the first place. In a recent ANA & White Ops study, 11% of display ads, 17% of programmatic ads and 19% of retargeted ads were served to bots in September of 2014. This major prevalence in bot traffic creates significant, painful inefficiencies before media can even be successfully optimized.
Viewability: Giving Credit where Credit is Due
Viewability is also an important factor in determining whether a user viewed an ad, and there are currently multiple technology vendors providing solutions. The most egregious blind spot in regards to viewability is re-targeters who use last-touch attribution without a viewability solution. This has equated to attributing 100% of the credit to the retargeting ad (prior to conversion) despite advertisers being unclear on whether or not the purchaser actually viewed the ad. In December 2014, Google’s Active View determined that 56.1% of ads served were not deemed viewable by the user, so there is a high likelihood of incorrectly credited attribution on a regular basis. Viewability is another measure that is essential to take care to move towards accurate measurement and optimization.
Mobile: Errant Clicks are Commonplace
With more consumers using mobile devices for commerce research as well as transactional purchases, many marketers are using the same metrics to optimize their mobile campaigns as those conducted on desktops. But the mobile web contrasts with the desktop in that there are a huge proportion of errant clicks; as a result, changes are required in terms of what metrics should be used for optimization. A report by xAd, Nielsen and Placed in September 2014 showed that clicks showed no correlation (or a negative correlation) to secondary action rates, which were stronger signals of purchase intent. This makes a strong argument for using different metrics and different optimization across channels and devices.
Cookie Proliferation Muddies the Waters
On the cookie front, third-party cookie tracking has become unwieldy due to the increased cookie proliferation due to browsers and mobile devices not accepting third-party cookies and security software deleting them on average every seven days. Users also operate multiple devices including business and personal computers, tablets and of course, mobile phones. In a recent Cisco study, it was estimated that by 2017 each user would possess five different Internet devices. Fortunately for advertisers, vendors such as Facebook are providing persistent identity management solutions that will tie together interactions from users across devices. This provides a much more accurate baseline to view traditional metrics such as reach, frequency, site overlap etc. and provides step function improvements in optimization. If a third party cookie is deleted on average every seven days, over a month period a user who views an ad four times per week, will be reported in two different ways. In a typical third-party cookie environment, this user would be measured as four different users with four different impressions in a given month. In a persistent identity management solution environment, this activity will be reported as one user having viewed 16 different impressions. The separate views on this simple metric would have profound implications on media buying and campaign optimization.
Processing Signal not the Noise
As more and more advertisers are making critical campaign decisions based on measurement data and are moving to a continuous optimization process for their display campaigns, it is important to build an accurate measurement framework to inform targeting and enhance campaign performance. This likely will come from a brick by brick approach versus a single black box solution to create a transparent measurement framework that can effectively process the signal-not the noise.