ARF Engagement Recipe: Surprise, Utility, and Emotion

In a series of webcasts and meetings during December, the ARF stimulated and educated the industry on the subject of Engagement. A great deal of valuable discussion took place and many companies participated. We did too. We were grateful when we saw that one of the ideas we had input, the importance of surprise in creating Engagement, stuck. Joe had undoubtedly already been thinking about it, judging from his instant agreement.

Joe Plummer, ARF’s Chief Research Officer and the leader of the MI4 Joint ANA/AAAA/ARF Task Force, put forth the following three possible drivers of Engagement in one of those webcasts:

1. Surprise/Discovery: the ad is in some way unexpected and therefore catches the consumer’s attention in a way that it would not if it were predictably in line with previous advertising seen. Primarily a creative control dimension, but with important media control aspects as we will see below.
2. Utility/Relevancy: the ad speaks to the individual’s interests. This dimension is the reason why targeting is important. If an ad is well targeted, this dimension will be highly functional in driving Engagement among the True Targets in the audience reached. Primarily a media control dimension in most cases but with some creative control aspects too. The ARF Adworks studies showed that targeting is the most robust predictor of high ROI.
3. Emotional Bonding: the ad creates positive affect such that the consumer is willing to be drawn closer to it. “Touching the consumer’s heart” is the high end of this dimension, and “liking” is more broadly descriptive of all instances of it. ARF’s Copy Testing Validation study had found that in addition to Persuasion, the best predictor of ROI among communications effects measures, Liking is a secondary measure of importance which can increase the ROI predictivity of Persuasion. Primarily a creative control dimension, but, again, with important media control aspects as brought out below.

Joe’s enunciation of these three dimensions sparked some new realizations in me. Some of the research we at Next Century Media had been doing had left some unresolved questions in my mind, and acceptance of Joe’s model would clear these questions up. Of course, the status of the three-driver model is that it is hypothetical at this point; much more research will have to be done before its truth is empirically established. Yet the explanatory power of a model is often a clue as to its veracity.

Here’s how the Plummer Model helped me reach closure on some intriguing questions.

Interestingly, these questions primarily relate to media, not to creative variables, although these two domains are abstractions, and the consumer never sees media divorced from creative or vice versa.

Next Century Media conducted a series of 30 studies over the past five years, for companies such as P&G, Nestle, American Express, Toyota, Volvo, et al, which show that “Brought to you by” Internet sponsorships, without advertising, average 7.5 times the Persuasion effect of average TV commercials.

Having been trained first at Grey which looked down on such sponsorship as mere signage, and always argued for product benefit demonstration at the time, I wondered “How could this be?” How could a simple “Brought to you by”, with no explanation of product benefits let alone demonstration/ dramatization of same, on the Internet, have more Persuasion effect than the average TV commercial?

All but one of these studies are proprietary. The one that we can show here is the one we did with the IAB’s help for Volvo and its agency Euro RSCG Circle. Volvo sponsored Yahoo’s coverage of the Annual New York Auto Show at the Javits Center. Volvo’s little box saying “Brought to you by Volvo For Life” appeared on every Web page next to photos and detailed editorial coverage of the features of all the new Mercedes, BMWs, Chryslers, etc. – all of Volvo’s competitors. The results: 7.7% of those exposed to the sponsorship message, afterward indicated that Volvo is one of the cars they will consider the next time they buy a car. The corresponding figure among the identical control group which did not see the sponsorship message, but saw everything else just the same as the test group, was 1.6%.

In the car funnel, longtime averages indicate that the average new car buyer considers 6 makes, test drives 3, and buys 1. Therefore, getting into the consideration set means you have nominally a one in 6 chance of making the sale. Not in the consideration group your chances are effectively zero. Therefore it was possible to model the ROI.

The model showed an ROI of 90 to 1. This is obviously very good in a world in which the DMA published average ROI of direct marketing is 11.49 to 1, and in which the Ephron/Pollak study showed that the average ROI of TV advertising is .54 to 1 for packaged goods and .87 to 1 for non packaged goods – in other words, these media on average lose money for the advertiser. (Counting only short term sales effects; Adworks studies suggest that one must multiply these by about 2.6 on average to model longterm sales effects, which would make the ROI on average TV slightly positive but nowhere near what Volvo achieved with this sponsorship.)

These results are hard to believe but directionally consistent with each of the 29 other studies. What is going on?

The Plummer Model makes this all understandable. This is why my gut feels that there is probably a heap of truth in the Plummer Model.

1. Surprise: the users seeing Volvo sponsor a show that trumpeted the glories of its competitors while never stooping to hawking its own cars, would have been quite surprising. It suggests an advertiser that sincerely wants to give good editorial content to people and does not ask for anything in return. This is gracious. The kind of behavior one sees in heroes. Not what the consumer usually sees advertisers do. Surprising and impressive.
2. Utility: coverage of such events in such depth had not been seen a lot on the Internet. Being able to “be there” brought a lot of interest and satisfaction to people interested in new cars.
3. Emotional Bonding: even before the Plummer Model we hypothesized a “Gratitude Effect” in response to this “gift” of good content. The lift in willingness to consider Volvo in all likelihood stemmed from an increase in liking of that manufacturer, not because of its product features, but because of the gentility of its management.

Another set of research findings we developed recently also became less puzzling when looked at through the lens of the Plummer Model.

Since the early 20th Century if not earlier, advertisers have wanted to put their ads into editorial contexts where the advertiser’s product category is central to the editorial itself. Cars in car books, golf clubs in golf magazines, etc. This is called Contextual Targeting or Content Targeting. Two benefits are seen in doing this:

1. Reach targets in concentration.
2. Get more attention to the ad, because people are in that environment due to an interest in the product category, so why wouldn’t they pay more attention to the ads for that product?

The second of these two assumptions fell apart in the last 18 months as Behavioral Targeting on the Web made it possible to reach people who had just been to a product related site, sending them an ad for that product in a lower-CPM non-product-related environment. For example, sending a car ad on a news site to someone who just went to Autobytel.

Behavioral Targeting controls for the targeting dimension by reaching the same people, so that the attention/ engagement metrics comparing Behavioral to Contextual Targeting relate only to the second hypothesis, i.e. that people will pay more attention to product X ads in a product X editorial environment.

In a significant proportion of cases they didn’t. Analysis of case studies from TACODA, Revenue Sciences, Advertising.com, 24/7 and others found that in about a third of cases, Behavioral beat Contextual in clickthrough rate and/or purchase intent lift and/or ROI and/or other communications measures. How could this be?

Dave Morgan, CEO of TACODA had showed me this puzzling situation and asked my help in figuring it out. This led to TACODA and NCM in December announcing a year-long research program to understand how Behavioral and Contextual Targeting really work on the Web, and how they can best be used together. The work reported next was all commissioned by TACODA. Thanks to TACODA for letting NCM design and analyze the research however we feel it will be of most value to the industry.

We hypothesized that two factors could be at work: excessive clamor/clutter of product X ads in Contextual, and/or the surprise effect of seeing an ad for a product in a place where you wouldn’t expect to see it.

An Eye Tracking study designed and analyzed by Next Century Media was carried out by The PreTesting Company. That study shows that Behavioral Targeting averages 17% more Ad Looks than Contextual Targeting, and this average increases to 54% more Ad Looks after the first exposure. In other words, as frequency builds, the same ad increasingly gets filtered out in the Contextual environment, but apparently as a result of surprise, the level of Ad Looking stays up in Behavioral. We are now planning a brainwave study to confirm this, because the P300 brainwave has been validated as a signature of surprise. The P300 brainwave occurs as an autonomic response when something in the environment does not match the expectation model in the brain for that environment. Because P300 is an autonomic response, it appears to not wear out in the first four exposures (the maximum frequency tested in the Eye Tracking study). Perhaps it doesn’t wear out at all.

These results become more understandable when looked at through the Plummer Model:

1. Surprise: this Engagement driver appears to be well leveraged by Behavioral Targeting. Good news for the users of BT – they never expected to get a bump in Engagement, BT justifies itself based on lower CPM against targets, now there appear to be two benefits.
2. Utility: Behavioral Targeting reaches the same targets as Contextual Targeting, hence high utility/relevancy – people interested in the product category.
3. Emotional Bonding: this is up to the creative. Behavioral Targeting in itself probably does not offer any special edge on this driver. But two out of three isn’t bad!

Thanks, Joe, for your contribution to Engagement theory. You’ve helped me make more sense of counter-intuitive data.

Please tune in next time when we will consider other theories of Engagement.

All the best,

Bill