This study uses a moment-by-moment copytesting technique to examine the differences between new product and established brand TV commercials from an information theory perspective. Based on a theoretical dichotomy suggested by earlier researchers, two types of visual information were identified in pictures taken from a sample of 41 commercials. P-type information was explicit, product-related content and E-type information was the esthetic, execution-related content. Using the TLK Picture Sort recognition technique, it was found that viewers process more of the E-type information present in established brand ads, while more P-type information was processed by viewers of new product ads.


From a theoretical standpoint, advertising for new products clearly differs in a number of fundamental ways from advertising for established brands. Coming in the critical first stage of the product life cycle, a new product commercial has the job of generating awareness of the new product starting from a zero base. It must, therefore, communicate a large amount of new information. It must communicate the brand name; it must communicate the category in which the product competes; it must communicate the attributes of the product and how those attributes are different from other products in the category and what the benefits of those differences are. By contrast, advertising for established brands has the benefit of prior advertising or marketing history. Usually its job is to remind consumers of the brand and to reinforce existing attitudes and loyalties toward the brand. Typically, established brand advertising has much less factual information to convey than advertising for new products.

It is reasonable to assume, therefore, that consumers would in general respond differently to new product than to established brand advertising and this assumption has been confirmed empirically by a number of researchers in recent years. Interestingly, the differences that have been found confirm not only that the information content of the two kinds of advertising is different but suggest that how consumers process the information content of new product ads is different from how they process the information in established brand ads. This difference in information processing is not fully understood.

This paper uses a moment-by-moment copytesting technique, the TLK Picture Sort, to provide a new perspective on this subject. In particular, we will attempt to bring a new precision to the measurement of the information content of an ad by defining two types of visual information that might be present in a tv commercial. Then we will report empirical results that demonstrate significant differences in how viewers process the information content of new product versus established brand tv commercials.

Literature Review

Olson, Schlinger and Young (1982) analyzed a large sample of TV commercials for new and established packaged goods using the Viewer Response Profile, a multidimensional rating system. They found significant differences in viewer response to the two types of advertising on a number of dimensions, with new product ads rated higher on the dimension of relevant news but lower on the dimensions of familiarity and stimulation. Notably, their interpretation of the lower stimulation scores for new product commercials was that it was due to the information overload of product news that would typically be canted by introductory advertising. They concluded that in terms of viewer response “new product advertising forms a distinctive and important genre or category of advertising.”

Stewart and Furse (1986), in their analysis of the recall and persuasiveness scores of a large sample of tv commercials, suggested that the “communication and persuasion process may not be the same for new product commercials as for established product commercials.” For new product commercials they found recall and persuasion scores to be highly correlated, whereas for established brand commercials the two measures were uncorrelated.

Jones (1986) pointed out in his review of the hierarchy of effects models first described by Ramond (1976) and based on the low involvement theory propounded by Krugman (1965) that the “Learn-Do-Feel” hierarchy is “relevant to the vast majority of packaged goods in their introductory phase” while the “Do-Feel-Do” hierarchy, or reinforcement model, is relevant in the majority of cases of established brand advertising. Implicit in these two models of advertising are two types of information content that might be found in commercials. “Learning” in this context pertains to the factual content of the ad, such as the product name, product attributes and benefits, etc. which produces a cognitive response from the viewer. “Feeling” pertains to the other executional content of an ad that produces affective or emotional response.

Young (1972) drew a similar distinction between two types of advertising copy. The first type she described as “explicit” copy, which “communicates concrete, product-related benefits.” The second type is “implicit” copy, which “communicates less tangible or more psychological benefits.” The implication that new product ads tend to be more explicit while established product ads tend to be more implicit is clear. Jones, in the same review, commented that “It is obvious that in most circumstances, with increases in a brand’s store of added values, implicit copy will become relatively more important as it takes on some of the prominence that was held by explicit copy during a brand’s introduction.”

Shannon’s invention of information theory (1947) has spawned many information processing models of communication but perhaps the most interesting in the light of the preceding discussion is the work done by Moles (1966) on information theory and esthetic perception. Moles proposes the existence of two types of information in messages in general. The first type of information he calls “semantic.” Semantic information refers to logical translatable, utilitarian information about the state of the external world and pertains to decisions about present or future actions. It would seem to correspond to the type of copy Young labeled explicit and which produces a learning response in the hierarchy of effects models. Esthetic information, on the other hand, relates to internal states. Instead of to a universal or logical repertoire, esthetic information refers to a repertoire of knowledge common to a particular transmitter and to a particular receiver and as such it is like “personal” information. In general, esthetic information is untranslatable from one channel of transmission to another. For example, the esthetic information content of a picture cannot readily be translated into words. Esthetic information would seem to correspond to the type of copy Young labeled implicit and which produces a feeling response in the hierarchy of effects models.

Young and Robinson (1989) used a moment-by-moment picture recognition technique to show that the type of information contained in the peaks of their video attention curve is related to the recall score generated by an ad. Specifically, peak experiences of explicit product information appears to drive recall. They have also shown (1990) that the number of peak experiences produced by an ad is related to its persuasiveness, at least as persuasiveness is measured by the RSC copytesting system. The relationship between persuasiveness and the type of information attended to has not yet been explored, but anecdotal results obtained to date suggest that it need not be explicitly product-related.

Building on this research, this study uses the moment-by-moment picture recognition technique to explore in further detail the differences between new product and established brand commercials. Specifically, we will examine differences in the information content of new and established brand commercials and how that information is processed by viewers.


The sample consisted of 41 finished 30-second commercials. This included 23 commercials for established national brands of consumer packaged goods and 18 commercials for both successful and unsuccessful new products. All of the commercials had been tested prior to airing within the last five years.

The tests were conducted in one-on-one consumer interviews during which respondents individually viewed the test commercial one time and than answered a series of open-ended and closed-ended questions describing their reactions to the advertising. Sample sizes typically consisted of from 50 to 100 respondents recruited by mall intercept on the basis of category usage.

Part way through the interview respondents were taken through the TLK Picture Sort (See Young and Robinson, 1987). The procedure uses a deck of still photographs taken of the commercial directly from a television screen. This deck represents a visual “sample” of the commercial’s images and typically consists of from I5 to 20 photographs for a 30-second ad. Respondents were given a randomized deck of photographs to look through and asked to sort them into two piles—the pictures they recognized from the commercial and the ones they did not recognize. The resulting recognition scores were then plotted as a time series representing the viewer “attention curve” for the ad.

The information content of the individual frames or pictures used in the test was subsequently coded using a two-way classification scheme. Pictures were classified as either P-type or E-type. P-type pictures were pictures containing explicit product-related content, such as the name, the package, visualizations of product attributes or benefits, or pictures of the product in use. E-type pictures were basically all other visuals in the execution. The images in E-type pictures could be said to represent much of the esthetic content of the video portion of the commercial.


Examples of the patterns produced by this coding are shown in Exhibit 1. Three commercials are shown. Each series of P’s and E’s represents the sequence of pictures taken from one commercial, in the order shown, and coded according to our categories. As can be seen, the proportion of P-type to E-type pictures present in a commercial and the order in which each type of picture occurs in the flow of commercial images varies considerably from ad to ad.

The data for the new product commercials were then aggregated and compared to the data for established brand commercials.

Discussion of Findings


A summary of the basic recognition measures for the two samples of commercials is shown in Exhibit 2.

In a sense, the total information content of the video portion of a commercial is a function of the visual complexity of the ad. A measure of that complexity is the number of pictures needed to describe a commercial with a picture sort deck. Looking at that measure as a starting point, we see that the total information content of the visual component of new product ads is comparable to that for established brand commercials. The average number of pictures used to describe new product commercials, 16.3 pictures, is only directionally higher than the number needed to describe established brand commercials, 14.6 pictures.

The amount of visual information actually processed by the viewer, as measured by the percentage of pictures recognized, is higher for established brand commercials, 66% versus 60% which is significant at p less than .05. Importantly, this difference in the amount of information processed is due to the higher rate of “peak visual experiences” viewers had of established brand commercials. (Significant at p less than .001.) We define a “peak experience” as a picture recognized by 75% or more of viewers. Here we find that while one-fourth of the pictures in new product ads were peak experiences, over one-third of the visuals in established brand commercials were experienced at peak levels.

A lower rate of information processing was expected for new product ads given the finding of Olson, Schlinger and Young that new product ads are perceived to be less familiar and more newsworthy. Since the information content of new product ads is in general more original, it should be more difficult to process. However, the higher rate of peak experiences for established brand commercials was not expected, but this is possibly related to the finding that established brand advertising is generally perceived to be more stimulating or entertaining.


The difference in the amount of pictorial information processed, while significant, is smaller than what we might expect given the greater load of information that new product commercials are generally expected to carry. To reconcile our intuition with empirical results we must first be clear about the type of information to which we are referring. Exhibit 3 shows the results of our analysis for our two categories of information, P-type and E-type visuals.

New product commercials were found to contain substantially more P-type information than established brand commercials. Less than half, or 47%, of the visuals in established brand commercials were of the P-type while 69% of the visuals in new product ads were, a level nearly one-and-a-half times higher. This is consistent with the commonly held perception that new product ads tend to be loaded down with “information”.

Now if we look at the type of information that is actually processed by the viewer we see a much larger difference than before, with 41% of the P-type visuals being recognized in the new product ads and only 29% of the P-type visuals recognized in the established brand ads. This is potentially misleading, however, because these differences simply reflect the proportions of P-type to E-type visuals in the two categories of advertising.

The results for peak experiences, however, are more interesting. Here we see the same number of P-type visuals in the viewer’s peak experience of both categories of advertising, 17% for new products and 16% for established brands, despite the beginning imbalance of P-type information in favor of new products ads. Moreover, E-type information occurs in peak experiences at a rate three times higher for established brand commercials than for new product commercials.

This finding helps to explain a number of previous research results.

For example, the finding that the same amount of P-type information is processed into the peak experiences of both new and established product commercials explains why the recall norms for most major copytesting systems for new and established product commercials are so similar. Young and Robinson have shown that it is the P-type content of peak experiences that drives recall.

Also, to the extent that the P-type content is motivating to consumers, this would explain the correlation between persuasiveness and recall scores found by Stewart and Furse for new product ads. P-type information, that is, explicit product-related information, is the dominant type of information processed at peak levels by viewers of new product ads.

It is likely, however, that E-type content can be just as motivating to consumers as P-type information. Given our previous research which found no relationship between E-type information in attention curve peaks and recall, we also have an explanation for the lack of correlation between recall and persuasion for established brand commercials. E-type information is the dominant type of information processed at peak levels by viewers of established brand ads.

From a theoretical standpoint, P-type information is the type of information we would expect to produce the cognitive or learning response predicted by the first hierarchy of effects model, learn-do-feel. This is exactly what happens with viewer processing of new product ads. E-type information, that is, esthetic or emotion-generating information, is the type of information most frequently processed at peak levels by viewers of established brand commercials. In terms of the hierarchy of effects, this is the do-feel-do model. And this, importantly, is consistent with the assumption made by many advertising practitioners that established brand advertising often works by an emotional rather than a rational mechanism.

Finally, we should point out that we do not yet understand how peak experiences are created in an ad—that is still one of the mysteries of the creative process. Given the relationship between peak experiences and recall and persuasion, however, this is clearly an important subject. An intriguing line of inquiry for further work on this subject is suggested by the recent studies of Csikszentmihalyi (1990) who has developed an analogous concept in the larger arena of the psychology of optimal human experience. Writing about his concept of “flow”, or peak human experience, he says,

“Because attention determines what will or will not appear in consciousness, and because it is also required to make any other mental events–such as remembering, thinking, feeling, and making decisions—happen there, it is useful to think of it as psychic energy. Attention is like energy in that without it no work can be done, and in doing work it is dissipated. We create ourselves by how we invest this energy…. When a person is able to organize his or her consciousness so as to experience flow as often as possible, the quality of life is inevitably going to improve…. In flow we are in control of our psychic energy and everything we do adds order to consciousness.”

Understanding the “micro-flow” of images in a tv commercial, whether it be for a new product or an established brand, would appear to be an important step towards understanding advertising effectiveness.


Csikszentmihalyi, Mihaly (1990), Flow: The Psychology of Optimal Experience, New York: Harper and Row

Jones, John Phillip (1986), What’s in a Name? Advertising and the Concept of Brands, Lexington: DC Heath and Co. pp 132-49

Krugman, Herbert E. (1965), “The Impact of Television Advertising: Learning Without Involvement,” Public Opinion Quarterly 29: 350-56

Moles, Abraham (1966, trans. by Joel E. Cohen), Information Theory and Esthetic Perception, Urbana: University of Illinois Press

Olson, Dave, Mary Jane Schlinger and Charles E. Young (1982) “How Consumers React to New-Product Ads,” Journal of Advertising Research 22 (June/July) 24-30

Shannon, C.E. and Weaver, W. (1949) The Mathematical Theory of Communication. Urbana: University of Illinois Press

Stewart, David W. and David H. Furse, (1986) Effective Television Advertising: A Study of 1000 Commercials, Lexington: DC Heath and Co. pp 23-24

Young, Shirley (1972), “Copy Testing Without Magic Numbers,” Journal of Advertising Research 27 (June/July), 15-22

Young, Charles E. and Michael Robinson (1989), “Video Rhythms and Recall,” Journal of Advertising Research 29 (June/July), 22-25

Young, Charles E. and Michael Robinson “Video Connectedness and Persuasion,” pending publication in the Journal of Advertising Research

Advances in Consumer Research Volume 18, © 1991 pp. 545-549

The Ameritest Flow of Attention and Flow of Emotion are registered in the U.S. Patent and Trademark Office.