Stop throwing good money at bad impressions and maximize the value of your investment in digital media.
In 2009, comScore and Starcom produced a joint study of clickers that concluded 8 percent of the Internet audience is responsible for 85 percent of the clicks.
This report elicits an examination of your frequent clickers and whether they are the audience your brand is looking to reach. This is fundamentally important to consider before selecting click-through rate (CTR) as a campaign metric.
My own research suggests something similar to comScore/Starcom. The analysis is constrained to an audience of just clicks and not Internet users. However, the results show a similar skew to the comScore/Starcom results.
In a sample of 200,000 random clickers taken in March 2012, I found that 40 percent of the clicks were produced by 20 percent of the clickers.
This data suggests there is substantial bias to a macro segment that can be described as frequent display ad clickers.
If there were no bias, the graph above would show a straight line, which would mean that the clicks are evenly distributed and therefore would not skew to any particular descriptive characteristics, but as you can see this is not the case.
What are the audience characteristics of the frequent vs. infrequent clickers?
I ran an analysis to describe the differences between the frequent vs. infrequent clickers. Here’s what I found.
Hour of day (EST):
Clickers have a macro profile. My interpretation of the charts above leads me to believe that frequent clickers in this sample have a tendency to be interested in:
Clicks aren’t enough. If you think your digital campaign has this problem, then you should consider consolidating your targeted media buying into a platform that can help focus inventory toward the right clickers. It’s as important to not optimize to a bias as it is to target the right audience.
Platforms that can help you will have some type of page level targeting and audience extension model as part of their offering.
If the audience and page planning constraints are established prior to the campaign, then the digital campaign does not waste money by chasing audience skews that the advertiser does not wish to target. Your campaigns will stop throwing good money at bad impressions and therefore maximize the value of your investment in digital media.
AS AN INDUSTRY WE HAVE POSITIONED OURSELVES AS THE MOST MEASURABLE. BUT HAVE WE PICKED THE WRONG METRICS?
The rise of the quant on Madison Avenue has led to many analogies that bridge the financial services industry with that of digital marketing:
High frequency trading = real-time bidding (RTB)
Spot markets = ad exchanges
Futures markets = insertion orders with option exercise fees
Commodities market = blind inventory
(Please insert your favorites in the comments section.)
I have used many of them and find them both helpful and fun. However, there’s one big analogy that’s missing from the quiver – what is the marketing equivalent of currency? I don’t believe there’s an appropriate answer.
What I mean by currency is a cold, hard measure of objective value. Without it, all transactions are relative.
The primary problem is not that a currency does not exist but that while a currency has been promised, false positive or relative metrics have been offered in its place. This is because, as an industry, we have positioned online to be the most measureable of all media. I believe we have picked the wrong metrics.
Consider CPA. It seems to be a clear marketing objective, but on closer inspection, I would contest that it’s often more misleading than not.
Example 1: Cookie Bombing
At one point in my career I had the privilege to work with a digital marketer that tried to spend $20 million in one month in digital display. During that time period, the digital channel could take at least partial credit for almost every online conversion. The client also realized eCPMs that were near $0.10, so the effective reach and frequency of the campaign was absolutely massive.
This is the same tactic to a lesser degree that is currently deployed by many affiliate networks and current sellers of CPA. It is commonly known as cookie bombing and it works if you’re using DFA to measure effectiveness, but it does nothing to strengthen brand marketing effectiveness. The tell-tale sign of a cookie bombing marketing plan is when nearly 100 percent of all conversions are claimed by the digital marketing plan and there is no correlation between whether a campaign/partner serves zero or 50 million ads in a day, and actual real activities generated that day or at lag = t-1, lag = t-2, lag= t-3 (time lag will capture any latent effects of the advertising).
Example 2: Below the Fold vs. Above the Fold
My company’s research suggests that pages below the fold are two times more likely to receive attribution credit than pages above the fold.
If advertisers are measuring solely on last view-through attribution, the RTB algorithm of their demand-side platform will gravitate to either one of two options:
I believe this to be true because the last ad on a page is the last ad to load so it will receive the last impression attribution credit.
Example 3: Upper-Funnel vs. Lower-Funnel Tactics
Today, most optimization methods gravitate toward the use of retargeting. This supports the buyer’s request of optimizing toward attributed credit or view-through optimization.
Retargeting at its best is a friendly reminder to a likely purchaser to come back to the advertiser’s site and purchase. This is the goal.
In reality, retargeting often takes credit for shoppers that are likely to purchase a product anyway. Even when done right, retargeting takes more credit than it is due.
In a case study done by my company for a U.S. airline, we measured the extent that our own retargeting line items “stole” attribution credit from upper-funnel-focused tactics. We did this by comparing the activity allocation of the upper-funnel tactic and the retargeting tactic through a first-impression vs. a last-impression attribution model.
The results of the first-impression attribution model showed that our own retargeting impressions diverted approximately 40 percent credit from the upper-funnel tactics. If the advertiser’s agency canceled the upper-funnel line, the advertiser’s CPA would go down; however, impression volume would suffer over the long term. Unfortunately, despite our sincerest objections, this is exactly what the agency for the advertiser decided to do.
A single publisher, ad network, ad server, DMP, or DSP that hides their marketing tactics and algorithmic approach as part of their “secret sauce” cannot fix the game. It’s the assembly of all the parts that delivers the full value.
It can only be done through a targeting platform that ties together multiple targeting tactics, algorithms, and any inventory (RTB or publisher direct) under a single arbitrator of value.
Learning and tactic transparency are critical in order to train the media planner and advertiser on the relative value of each tactic for their brand. It trains their marketing intuition. But this approach also disrupts the current state of media planning and therefore full adoption via platform consolidation is a necessary but difficult leap of faith.
It’s further complicated by the fact that even though there are clear benefits to the advertiser, platform consolidation puts the agency model at risk. This is similar to what we’ve witnessed over the past two year as the ad exchanges/SSP/DSPs have put the ad network model at risk. However, the networks that provide value through meaningful and tangible differentiators have done far better while those that have not have done far worse. The same should be true for digital agencies as we see more targeting platform consolidation over the coming years.
My problem with CPA is not that it’s used as a measurement but rather that it’s used as currency to pass judgment on the effectiveness of one “inventory” supplier over another when that measurement is clearly flawed. Bad actors will always game univariate measurement. It’s in their best interest to do so.
One of the most heralded benefits of digital – its ability to measure – has slowly become a major deterrent to good advertising. Instead of informing strategy, it has begun to distract it; instead of pointing our efforts in the right direction, it has veered them off course.
But there is much to be learned from the game, as well. Until planning teams become more educated about these phenomena, the industry should view CPA metrics with a cautious eye. This doesn’t mean dismissing them altogether, but simply recalibrating them with the understanding that CPA is a relative, not objective measure.
As long as digital marketers continue to offer the current state of CPA as currency, then marketers will continue to underinvest in digital. Digital media was promised as the most measurable media, but we got the measurement wrong. Our mistake keeps scale out of the industry. It’s a cruel paradox, but one I believe to be true.