Adometry Founders Blog

If you studied psychology, you probably learned about Maslow’s hierarchy of needs. Proposed by Abraham Maslow in 1943, it describes human needs from base physiological needs, up through the need for safety to self-actualization. The model’s hierarchical nature implies that if a person’s base physiological needs are not being met, it’s very difficult for him to concentrate on higher level needs. In other words, if you’re dying of thirst, you’re probably not worrying about maximizing your creativity.

Figure 1. Maslow's Hierarchy

What does Maslow’s Hierarchy have to do with digital advertising? As it turns out, Maslow’s insights provide a good model for thinking about digital impressions and their impact.

When an advertiser (or an ad agency on their behalf) buys an impression, the result of that impression can roughly be categorized into one of three categories as shown below.

Figure 2. Advertiser's Hierarchy of Digital Impressions

Brand damaging impressions – No one at an agency wants to receive a call from a client complaining that his brand showed up in an advertisement on an inappropriate site or next to unacceptable content. For some brands, any association with pornography, sex, violence, etc. is unacceptable. Like Maslow’s hierarchy, it’s hard to think about optimizing ad effectiveness when you’re apologizing to a client upset by a placement on an inappropriate site. A recent study by Econsultancy and the Rubicon Project surveyed 130 online advertisers and found that 65% were worried about their ads appearing next to inappropriate content¹.

Our own tests have found that potential brand damaging impressions are a small percentage of the total, ranging from zero in direct buys to as high as 1.5% on some ad networks.

Wasted impressions – Wasted impressions aren’t damaging to the brand, but they may damage the relationship of the agency with the advertiser who pays the bills. Examples of wasted impressions include missed geo targets, ads with very short exposure times, never viewed ads (below the fold or hidden), over-exposed ads, or ads on sites with an audience unlikely to engage (e.g. ads for retirement planning on a site for children). Fraudulent impressions can also be included in this category.

Our data suggests that wasted impressions constitute 20-50% of impressions in large campaigns. Wasted impressions are not limited to ad networks or blind buys – some buys on portals or even direct buys include wasted impressions, particularly “run of site” buys.

Effective Impressions – Effective impressions constitute the remaining 50-80% of the impressions, but not all effective impressions are created equally. Some impressions deliver greater conversions and engagement. Much of the data gathered to identify wasted impressions can also be used to identify the most effective impressions: site category, location on the page, time the ad is visible, frequency. All of these variables can be correlated with conversion or engagement data to determine the relative effectiveness of an impression. This is the highest level of the hierarchy and the one where the value of advertising is realized.

Action – The first step is to eliminate the brand damaging impressions. These tend to be small percentages of network buys and are very rarely seen on premium publisher sites. Distinguishing between wasted and effective impressions is next and the greatest potential area for improvement. By identifying ads that are never seen or poorly targeted, comparing networks by actual delivery performance against cost, and using conversion or engagement data to determine the ideal time exposure, frequency, day part, and page location, marketers can improve both the effectiveness and the efficiency of their online display campaigns.

1. Ad Network and Contextual Buys – Marketers certainly have the ability to purchase ad inventory through networks, and will frequently request specific kinds of content based on their target audience. Through the data we’ve collected we can tell that several ad networks buy from other ad networks, which again may buy from another ad network, and each network will define categories their own way. What the Veracity data has shown is that as much as 20% of impressions purchases through ad networks are complete misses from a contextual targeting standpoint, and even higher with sites that are marginally related to the desired context.
2. Timed Exposures – When buying by CPM, all impressions are treated equally, but there is a lot of variability.  One of the metrics that Veracity provides is around exposure time, looking at how much time the advertisement spends visible to the user (taking into account things like multiple tab browsing, other applications, etc). What we’ve seen in early campaigns is that the percent of impressions that are active for 20 seconds or more varies between 40% and 85%.  If as a marketer your goal is to expose a user to you message, those numbers indicate a large difference and can change the effective CPMs for a publisher or ad network considerably.
3. Reach and Frequency – We know that so much more can be done online, but we feel looking at reach and frequency is still a fundamental way to look at advertising and comparing across channels.  Additionally, looking a reach and frequency variations by other metrics is really interesting.  At the placement level, we’ve seen average frequencies ranging from just over 1 to 5 or 6.  If a campaign is attempting to optimize against reach, these variations in frequency can modify the effective costs substantially.  Even within a placement on an ad network, different sites in that network may vary even more than that, with some sites getting the lion’s share of impressions and generating frequencies as high at 10 or more, potentially saturating an audience with the same message.

Check out John’s latest on nPost: Some Well Worn Advice for Any Startup

Forrester recently came out with a new round of financial predictions through 2014 for Interactive Marketing. The report was followed by the usual round of articles saluting or decrying Forrester’s numbers, methodology, etc. I, for one, take the flood of predictions throughout the year as I think almost everyone else does, with a grain of salt. I tend to see the real value of most of the surveys and research in terms of overall trends. The interesting things are usually around spikes in one type of advertising (video or mobile) or the long-term increase on one form of advertising at the expense of another (newspaper, magazine). I expect that is what most of us look for.

We have taken the stance with many of the people we have spoken to over the last several months that there may be another trend, to which we hope to contribute, that is somewhat overlooked. I’m talking about driving additional investment in interactive advertising via better analytics. Our point of view is that advertisers and agencies too frequently simply can’t quantify what it is that they are getting from their online display advertising. And if that is true, the mystery of how interactive and traditional media interact must be an even bigger mystery to those creating the overall advertising plan.

Given the choice of real numbers around your online spend versus the ‘tried and true’ estimates most marketers live with from print and television, wouldn’t you be more inclined to buy interactive to the point of maximum effectiveness at the expense of other mediums? Through delivering greater understanding of the trends and drivers in the interactive advertising space to those that will make the most out of them, we believe overall interactive spending may increase faster than the pundits expect.

I saw again the other day that US government officials are looking at behavioral targeting to determine if it is a breach of privacy. They follow a number of other political entities including Canada and the E.U. in their investigations. The BT companies argue that the practice is entirely anonymous, and will only help the consumer get exposed ads which will interest them more. Having personally been involved in some behavioral targeting at a previous job, I tend to think of it as a relatively harmless practice. However, I do get that it is widely misunderstood by all parties. The public sees it as being more intrusive than it really is. Some advertisers who employ it, think of it as the silver bullet that will fix all their online advertising problems. Regardless of which camp you sit in, there is enough of a public perception is that BT is a bad thing. Otherwise we wouldn’t see politicians pursuing the issue.

I think there is value in behavioral targeting, but we should all be prepared for the possibility that their may one day be something like a ‘do-not-target list’ or some other limitation, which, depending on how widely it is adopted, immediately devalues the whole proposition. Online advertisers should be prepared to look beyond targeting to more detailed campaign results. There is a plethora of useful, directional data available to get the job done without crossing the invisible line of public perception. We all know intimately how important relationships are in business, and this extends to the perception of the general public as well (aka, brand perception). All of us in online advertising should be concerned. The bottom line is that we can do the job without being perceived as the bad guys.