One of the most seminal shifts in digital advertising over the past 24 months is the migration of media from content to audience. Much to the chagrin of major, branded content sites, the rise of audience targeting promotes two fundamental marketing tenets.
First, it’s about the customer, not the content. Publishers don’t own their users. In fact, users are individuals with unique attributes, attitudes, and behaviors — beyond mere content consumption — that predict their receptivity to brand messaging or purchase activity.
Data providers like Blue Kai and Bizo use these descriptors to create targetable segments, each with their own implied ROI and eCPM. This is good for both marketers and publishers.
The second tenet, perpetuated primarily by DSPs and the exchanges, suggests that marketers don’t need to pay premium CPMs to find those user segments on branded portals. Instead, advertisers can purchase less expensive run-of-site campaigns across a multitude of sites — regardless of the content they’re reading — only showing the advertising when someone from the targeted audience arrives.
On its face this makes perfect sense, right? Television has worked this way for decades. IBM is a long-time advertiser on the professional golf tour, in hopes of finding senior corporate executives on the couch. But, does content have no impact on user ROI? Is targeting just about audience or just about content?
With this debate in mind, AdJuggler ran a controlled test over the summer to understand the incremental yield to advertisers — if any — based on the content users are reading when the ad is shown. We also wanted to independently substantiate the yield improvement by targeting ads, based on demographic variables appended to site users.
In short, does content drive yield? And, are audience attributes predictive of higher yield. To keep the test simple, yield was defined as CTR, lead rate (completing a contact form post-click), and sales rate (entering a commerce zone).
For a sample of 872 million impressions, every ad was categorized into one of 51 product or service categories, and every web site was categorized into one of 15 content segments. Approximately 37% of users — an equally significant sample — were also matched to an Equifax consumer lifestyle database (now owned by Alliance Data/Epsilon), comprising more than 150 demographic and purchase attributes.
The AdJuggler analysis suggests three broad conclusions for marketers and publishers. First, yields vary significantly by advertiser product or service segment — CTRs, on average, as high as 0.3% for certain entertainment and education offers and as low as 0.005% for certain CPG and technology offerings. And, separately, the CTRs achieved on different publisher content channels also vary significantly — as high as 1.20% for certain niche content categories to below 0.05% for multi-category portals. This range of results – and the unique performance of advertising vs. content – is fairly intuitive.
Second, and more striking, we found that the yields vary even more significantly when targeting certain advertisers on certain publisher channels. In the case of, say, the health care marketer segment, the yields varied even wider, based on the content publisher channel. The top-performing publishers — in this case, medical sites — achieved CTRs of two to three times higher than their segment average yield.
And, third, we found that the Equifax user attributes are also strongly predictive of advertising response. Many of the attributes correlated to yield in similar ways, enabling us to group users into consumer segments based on a collection of financial, demographic, purchase, and attitudinal factors (e.g., “affluent buyers”, “hobbyists”).
Across advertiser segments, financial attributes are most predictive of yield, but the real insight is specific to each advertiser segment. In the case of automobiles, for instance, certain lifestyle attributes were most predictive of advertising response – and these impacts produced 50% to 300% yield improvements when targeting the those segments.
Yes, we may all understand that different advertisers and different publishers produce different yields. But, today we see minimal evidence that publishers have the tools to understand these yield differences, and, as such, are not monetizing these differences in the form of higher CPMs.
The analysis does take a step to validate the industry’s focus on audience targeting. But, equally striking, is the correlation we found between advertiser and publisher content, suggesting that CTR and brand impact is a combination of audience, the offer, and the content a site visitor is consuming.
Content matters — sometimes. Audience matters more.
Read the full whitepaper at AdJuggler.com.