RIP, CTR: Advertising Has Changed, So Should Our Metrics
EDITOR’S NOTE: This article, originally published on January 14, 2013, placed at No. 9 in our 20 most popular articles of the year.
ADOTAS – Within the industry, so many of us are still holding onto stale weights and measures that may be great for beefing up agency campaign reports, but don’t actually mean much to the business anymore. Impressions are still an industry standard, with CPMs used as a way to quantify the ad buy by slapping a price on display advertising.
But do impressions really matter these days? Just because an ad is served when a page loads, who’s to say if it is really making an impact for your customers or even reaching your potential customers? Impressions used to be a primary indicator of campaign success; that was quickly replaced by CTR, the infamous click-through rate. But now some companies are claiming that this metric is no longer as relevant as it use to be.
A study from Pretarget (now Sequential Media) and ComScore revealed that even when a user clicks on an ad, the correlation between that click and a conversion is virtually nonexistent. Engagement can be measured in many ways and today marketers can track virtually every data point imaginable besides clicks and traffic driven through a creative like: time spent, mouse-overs, hovers, views, shares, tweets, and even maximizing and minimizing ad units.
As part of the study, Sequential Media analyzed more than 260 million ad impressions across the campaigns of 18 advertisers over a span of 9 months, and tracked conversions ranging from filling out online form sto downloading software. In the study, Sequential Media found that the Pearson correlation (a common correlation methodology) between clicks and a conversion was 0.01, the lowest correlation rate among metrics tracked in the study (a 0 result would mean there is absolutely no correlation, while 1.0 would signify the strongest possible correlation.)
“My key takeaway,” said Sequential Media founder and CEO Keith Pieper, “is that optimizing to viewable impressions or hover time is a better proxy for a brand advertiser than a click-through rate. Metrics like ‘hover’ and ‘view’ that have typically been thought of as more brand-focused metrics actually can end up being super appropriate for direct marketers.”
Others tend to agree with Pieper’s assessment. According to Eric Franchi, co-founder of online advertising company Undertone, CTRs are quickly becoming a dead metric. “Unless an advertiser’s goal is a post-click action, click rates should never be a primary indicator of campaign success. In proper context – versus other creative concepts, data sets and ad units – they might add a layer of dimension to the campaign story. Besides, barely anyone clicks on standard display ads, currently less than one click per thousand ads served across the industry.
When asked what could potentially be replacing CTR, Franchi said, “There is no single metric, just as the click should not be the end-all for all things interactive. For campaigns with brand goals, there are far better indicators to use, such as engagement with the ad, or classic brand indicators such as awareness or purchase intent. Further, this focus will drive innovation towards larger, more immersive ads can deliver this type of connection.”
Then what do we use next besides pixel to track conversions? What is the best way to track consumers engagement with ads?
It seems to me that if the advertiser was to achieve a conversion with no click, and the conversion were ‘ranging from filling out online forms to downloading software’ then the attribution of that conversion should really go to another source because the ad being measured was ineffective at driving the conversion.
I also love that the second highest correlation to conversion in their study is if the ad is viewable at .35. That is to say that un-viewable ads have a SIGNIFICANTLY higher correlation to conversion than ads with clicks since the gross impressions have a .17 correlation to conversion. I wonder how many clicks were in their sample size?