ARF Media Model Illuminates Current WFA and MRC Initiatives

The ARF Media Model is a tool for helping practitioners think about measuring the effects of advertising. Work on it goes back to 1948, and it was most recently updated in 2003 by Erwin Ephron and myself, along with Jim Spaeth, Bill Moran, Denman Maroney and Phil Brandon. Although its name suggests otherwise, it is actually concerned with the holistic effect of media, creative and context. The first three of its eight levels are media-only effects (including context), and the higher order media effects (levels 4-8) are combined with creative effects. At level eight, sales, the effects are also strongly influenced by non-advertising variables such as sales force, product/service, packaging, distribution, shelf facings, adjacent products on shelf, price, promotion, PR, word of mouth, weather, economy, competition, pandemics, elections and possibly sunspots (jocular).

The current WFA/ANA/ISBA et al initiative is focused on measuring reach/frequency across TV and digital so as to enable reduction of excess frequency on the heaviest viewers, thereby improving the consumer experience with advertising. It represents the first time that advertisers have stepped in and committed to paying at least a significant part of the cost of achieving this simple-to-state, not-so-easy goal.

MRC's Outcomes Standards Working Committee (of which I'm a member) is attempting to write a set of standards which could be used to MRC accredit suppliers of Time Series Analysis, Marketing Mix Modeling, Multitouch Attribution, Singlesource, Matched Market Trials, Lift Studies, Random Control Trials and other measures/models of business outcomes driven by (or at least correlated with) advertising.

In the past two months there has been a lot said about the need for WFA, MRC and the many other parties involved in these two somewhat overlapping and to some extent coordinated activities to re-read the ARF Media Model. Much of this sound advice has been coming from the energetic Tony Jarvis (also a member of the MRC Outcomes group). Tony has often pointed out that the industry ought not stray from our past high water mark accomplishments that can continue to guide our clarity of thinking if we do not ignore them.

My intent in this post is to bring forth some of the most relevant text in the 2003 ARF Model. The ARF Model supports the growing consensus within both the WFA and MRC "colleges" for using level three of the ARF Model, Advertising Exposure, as the basis for determining whether advertising has actually reached a specific person or household, i.e., the ARF Model plays a role in defining what an impression is, and therefore whether it can have an outcome involving that person or household.

First the definitions of the eight levels:

Note that one must be careful not to confuse the two cases in which eyes are mentioned: in level three, the eyes must simply be open. In level four, the eyes must be on the screen (or the printed page or sign).

Also, as one digs into the ARF Model, certain clarifications become apparent. For example, although the intent is to count ad exposure when eyes are open, pragmatically the 2003 Model states, in the words of Erwin Ephron [adding my years-later further two cents in square brackets], "In broadcast, it is impossible to obtain an objective measure of the people whose open eyes/ears are confronted by the vehicle [or ad]. The diaries and meters used to measure persons viewing or listening [/hearing] are not objective, for they require respondent input." (page 14 of the ARF 2003 Model)

In other words, we infer both vehicle and ad exposure if the respondent says that they were viewing at that time. In the Nielsen national peoplemeter system for example, if the respondent has pushed his/her button to indicate he/she is viewing at the program level but does not change button settings for a brief room-leaving during the ad, ad exposures will be inflated.

In fact, this is what concerns Ed Papazian and Tony Jarvis. Ed estimates that the inflation is 2X-3X, which is supported by innumerable studies done over the past half century.

Part of my motivation in hounding the industry through the '70s-'90s to move to passive peoplemeters (which led to the PPM on whose original development I consulted) is that we would objectively detect room-leaving. (This utopian state was envisioned in this ancient Myers Report post.)

Alas, at the moment the best we can hope for from the WFA and MRC efforts (and those of so many other fine people) is to not fall back from the sanity of the ARF Model into earlier confusion. We can also enhance it, for example by adding clarifying adjectives such as "device" and "person" with regard to certain measures. For example, there can be a device Ad Impression or a Person Ad Impression (the latter would use an objective passive peoplemeter, or today's push-button peoplemeter subjective viewing claim). The same could be applied to viewing duration, which could be measured at the device or person level, and be labelled accordingly.

New technologies being introduced today create an additional potential level to be added to future updates of the ARF Model: predictive measures. Adelaide is an example of a metric that predicts attentiveness; RMT Resonance is an example of a metric that predicts all of the ARF levels from four to eight as validated by Nielsen NCS, Simmons, 605 and others.

Once we are all speaking the same language, we can get to the advertisers' desired Northstar and MRC Outcome Standards sooner.

Don't miss the next post in which I compile deep thinker reactions to this article from CIMM's Jane Clarke, MASB's Tony Pace, Ed Papazian, Jim Spaeth, Tony Jarvis and Ethan Rapp, and add more dimensions to this discussion relevant to the future of measurement.

Click the social buttons to share this story with colleagues and friends.
The opinions expressed here are the author's views and do not necessarily represent the views of MediaVillage.com/MyersBizNet.