Addressing New Media with Conventional Media Planning

Hugh M. Cannon

Wayne State University


Accepted industry wisdom is that many new, and particularly Internet, media cannot be addressed through conventional media planning procedures. This paper takes a contrary position. It not only argues that new media can be addressed through conventional planning procedures, but it contends that they should be. Increasingly, all media compete for the same budget. Furthermore, they play critical roles in the same integrated marketing communications programs. To suggest that they should be treated differently argues against truly integrated media planning. While this paper does not suggest a comprehensive integrated planning solution, it does outline the directions such a solution should take. Most important, it shows how all media selections can be addressed through a common evaluative process.

Addressing New Media with Conventional Media Planning

One of the side effects of the media revolution has been a shying away from traditional research in conventional quantitative media planning. There are a host of reasons. One is that media are prolipherating faster than the sources of data needed to measure them. But research is addressing this problem. In Internet media, measurement is relatively easy. The question is more what should be measured (Novak and Hoffman 1996; Pavlou and Stewart 2000). Leckenby and his colleagues have made considerable progress in developing measures of reach and frequency for evaluating Internet media plans (Leckenby and Hong 1998; Kim and Leckenby 2000).

The greater problem is media integration. How does one develop common standards for evaluating such varying media types as the Internet and television or magazines? The issue of intermedia comparisons ranked a close second to the effect of advertising frequency in Schultz’ (1979) study of media practitioners’ research priorities. While we have no data to support this, we can infer that the problem has become more pressing with the passage of time. In reality, media plans have never focused on a single medium. But the problem is exaggerated with the increase in the number of media alternatives, and, in particular, the advent of Internet media. The popularization of integrated marketing communications (IMC) has further raised the stakes, placing more pressure on planners to find methods of integrating diverse types of media (Schultz and Kitchen 1997; Kitchen and Schultz 1999)

One approach is to retreat in the face of the increasing complexity of the problem and rely on qualitative media selection criteria. While we understand why people would take this position, the fact is that modern campaigns often involve millions of dollars in media expenditures. With this kind of money at stake, the benefits of a more systematic, theoretically rigorous, and ultimately, quantitative approach are enormous. The purpose of this article will be to propose a basic system for developing a common, quantitative planning system that can be applied across multiple media classes.

Conventional Approches to Media Planning

While we speak of “conventional media planning,” we recognize that there is no single, universally accepted model. But virtually all models address concepts such as gross impressions, reach and frequency, effective reach and requency, rating points, gross rating points, share, duplication, audience composition, households/persons using television/radio, and cost per thousand (Novak and Hoffman 1996), if not by name, at least in concept. These grow out of a relatively recent tradition, from the early 1960s, when media planning was still struggling with the basic concepts of quantitative media analysis. In 1961, Agostini published his famous reach estimation formula, followed by a host of competing formulations (Bower 1963; Caffyn and Sagovsky 1963; Kuhn 1963; Marc 1963; Hofmans 1966; and Claycamp and McClelland 1968). These formulas enabled media planners to estimate the reach, and by extension, the average frequency of a schedule. Reach and average frequency were superior to raw media weight in that they incorporated the notion of duplicate exposure. Thus, they addressed two key elements of media strategy — how many people are exposed, and how much advertising these people receive.

Reach and average frequency is a crude planning tool, even though it is superior to the alternative of raw media weight it replaced. When computers were introduced into media planning, it became practical to begin working with frequency distributions — estimations of not only how many people were reached, but also how many were reached with various levels of exposure. In the early 1970s, Gensch (1973), and later Chandon (1986) and Rust (1986), compiled extensive reviews of the media schedule simulation literature. Many of the later models would estimate frequency distributions (Metheringham 1964; Green and Stock 1967; Beardon, Haden, Klompmaker and Teel 1981; Rust and Leone 1984; Leckenby and Boyd 1984; Leckenby and Kishi 1984; Leckenby and Rice 1985; Danaher 1988; 1989, 1991; 1992). However, it was not until the advent of personal computers, and the corresponding drop in the cost of computations, that these gained broad acceptance in day-to-day media planning. As late as the early 1980s, the popular advertising literature was filled with discussions of why frequency distributions should be used instead of simple reach and average frequency analysis.

When cheap computational power arrived, media planners were faced with a catch-up game. They recognized that using reach and average frequency analysis was naïve, and that it should be replaced by an analysis of frequency distributions. The question was, “How?” Some visionary organizations, such as Foote, Cone and Belding Communications, developed more sophisticated planning systems (Wray 1985). But the industry settled on the simpler notion of effective reach and frequency, or simply effective frequency planning (EFP), as articulated in Naples’ (1979) classic book on the subject. Estimates vary of how widely spread the use of EFP became, but it was clearly the dominant media planning paradigm in the 1980s and 90s (Kreshel, Lancaster and Toomey 1985; Leckenby and Kim 1994).

While EFP is still widely practiced in the advertising industry, the concept has been shown to have major conceptual problems – namely, that it still involves crude rules of thumb that did not fully utilize available media data and that it presumed an “s-shaped” advertising response curve that rarely exists in real advertising situations (Cannon and Goldring 1986; Cannon and Riordan 1994). We will build the model discussed in this paper around the concept of frequency value planning, or FVP (Cannon 1996; Cannon, Leckenby and Abernethy 1996, 2001). FVP is an emerging model that addresses the major shortcomings of EFP.

A Media Planning Model

Hoffman and Novak (2000) suggest that, while quantatitive media planning is essential for Internet media, conventional approaches do not apply:

“As with advertising programs in the real world, online managers want to know whether advertising is achieving the company’s business objectives. Traditionally, however, advertising has not been able to provide meaningful measurements of results. At best, conventional broadcast and print advertising has only allowed measurement of the mass amount of advertising delivered – as opposed to actually consumed – usually expressed in terms of exposures or impressions and at best broken down by demographic or psychographic segments.

Thus, for managers seeking measurable results from online advertising, traditional media models offer little guidance, and may even hinder the process of capturing the value in consumer response in virtual environments. Further, the Web possesses unique characteristics compared to conventional media. These differences warrant new strategies. Borrowing advertising practices and metaphors from traditional media like exposure-based advertising pricing models and billboard-like “banner ads” is likely to be appropriate only for limited advertising objectives such as awareness in the context of branding goals.”

This paper takes a contrary view. Exhibit 1 portrays a basic media vehicle planning model that is based on conventional approaches, including those criticized by Hoffman and Novak. It begins with the well-accepted premise that media planning ultimately stems from marketing plans (box A), as suggested by the classic DAGMAR paradigm (Colley 1961). In support of this is an integrated marketing communications (IMC) program (box B), also prominent in the literature. While IMC is relatively new as a formal concept and has no clear, universally accepted definition (Schultz and Kitchen 2000), a growing body of literature and practitioner experience suggests that it grows out of a reaction by advertising agencies to such factors as the development and usage of databases, client desires for interaction/synergy across media, and the need for coordination on both a global and a regional basis (Kitchen and Schultz 1999). Following the DAGMAR paradigm, marketing strategy provides sales-oriented objectives for the IMC analysis. The IMC analysis itself consists of an effort to determine the kind of promotional support needed and divides it up into logical tasks that can be assigned to specific, synergystic media programs.

Exhibit 1:
The Media Planning Process


Media Planning Process

It is at this point that the process begins to address actual media decisions. As Exhibit 1 suggests, the first step is to allocate promotional tasks to various media classes (box C). Then, within each media class, the process seeks to identify the specific media vehicles that are most cost-effective in performing each task (box D). Finally, the process draws on the most efficient media vehicles to develop media schedules, again in support of each task. The process seeks to adjust these schedules to maximize their frequency value — their impact, given the possibility of duplicate exposures (box E).

We will now turn to the task of discussing the three media stages of the planning process that support what we have termed “media strategy.” These are portrayed by boxes C, D, and E in Exhibit 1.

Allocating Tasks by Media Class

As Exhibit 1 suggests, the first media stage of the planning process seeks to allocate the various IMC tasks to specific classes of media. This is consistent with traditional planning, but it also lends itself very well to new media. Indeed, new media (particularly Internet media) tend to play more highly differentiated roles than conventional media. In this sense, the traditional model shown in Exhibit 1 actually works better for campaigns that include both new and old media than for traditional campaigns. The only requirement is a model to guide the integration.

Exhibit 2 provides such a model. It assumes that IMC tasks will consist of advertising-based marketing support strategies that must ultimately be expressed through some kind of creative strategy and execution. Exhibit 2 provides a framework for reconciling executional factors to media environment.

The Exhibit 2 framework defines media environment in terms of consumers’ typical psychological environment during advertising exposure. The environment is defined by consumers’ need for information and the capacity available to process it. That is, how much is the typical audience member likely to be actively looking for, or at least receptive to, advertising information? And, how much actual processing of advertising information is the typical audience member able to do, given both his/her level of involvement and given the number of competing mental tasks competing for her/her attention.

Exhibit 2:
Allocating Tasks by Media Class

Allocating Tasks


Note that the framework is closely aligned to the creative strategy models growing out the Foote, Cone & Belding (FCB Grid) traditon (Vaughn 1980, 1986; Rossiter, Percy and Donovan 1991; Rossiter and Percy 1997). As with our framework, the creative strategy models are built around two dimensions, although they are labelled differently – (1) thinking/feeling” in the case of the FCB Grid, and Wells’ now classic “informational/transformational” distinction in the case of the Rossiter/Percy Grid, and (2) “high involvement/low involvement.”

The Need for Information Dimension

The best way to understand the “need for information” dimension of media environment is perhaps to review the logic behind the “thinking/feeling” or “informational/transformational” dimension of the creative strategy models. Ultimately, these describe the orientation of the promotional messages that must be delivered in an appropriate media environment. Understanding them, then, should provide a key to media classification.

The scale items used to operationalize the “thinking/feeling” dimension of the FCB Grid (Ratchford 1987) are suggestive of the distinction between what Shimp (1997) calls the consumer information processing model (CIP) and the hedonic, experiential model (HEM), respectively. The CIP model parallels the functional attitude theoretical notion of “utilitarian” attitudes (Katz 1960) – the idea that people evaluate alternative behaviors by weighing the consequences, in our case the benefits derived from product attributes (Fishbein and Ajzen 1975). In the FCB Grid scales, this dimension is operationalized by items such as “decision is mainly logical or objective” and “decision is based on functional facts” (Ratchford 1987). The HEM tends to involve holistic evaluations that are “felt” and “experienced” rather than reasoned out (Hirschman and Holbrook 1982; Holbrook and Hirschman 1982). These seem to correspond with the FCB “feeling” scale items such as “decision expresses one’s personality” and “decision is based on a lot of feeling” (Ratchford 1987).

Rossiter and Percy (Rossiter, Percy and Donovan 1991; Rossiter and Percy 1997) argue that the FCB “thinking” dimension relates roughly to “informational” advertising, but that it suffers in its simplicity. In the Rossiter/Percy approach, “informational” advertising draws on five different “negative” motivations. Rossiter and Percy argue that the “feeling” dimension corresponds relatively closely to “transformational” advertising, which, in turn, implies positive consumer motivations.

In our view, the general distinction between the two modes of information processing is “logical” versus “associative.” The “need for information” dimension in Exhibit 2 is based on the premise that both “thinking” and “informational” messages are processed logically, consciously using information to make inferences regarding how to achieve consumer objectives, as posited by the CIP model. “Feeling” and “transformational” messages are processed associatively – experienced — with only a casual regard for logic. This is important because consumers do not consciously look for associative information. They do not seek information or cues that will transform the way they feel about a product. Contrary to Rossiter and Percy’s contention, we maintain that this applies to some negative motivations, such as feelings of inadequacy that are triggered through an ego-defensive appeal, as suggested by Cannon and Boglarsky (1991). Consumers do not address threats to their egos by logically choosing to abandon their feelings of inadequacy any more that they consciously evaluate attributes and choose to fall in love. Both are experiential events, the key to which lies in unconscious associations. This explains why ego-defensive messages do not tend to work in information-seeking media environments, such as magazines, and certainly, the Internet.

To summarize, then, IMC messages that are highly associative in their persuasive approach — that address HEM motivations — are best delivered in media environments such as television, where people tend to process cues by association. In these media, consumers tend to “experience” rather than consciously “evaluate” input. Conversely, IMC messages that rely on conscious CIP evaluation are best delivered in media environments such as magazines or the Internet, where consumers tend to process decision-related information through logical inference.

The framework goes beyond the obvious distinction among major media classes. Within classes, we may also distinguish among media that call for logical versus associative processing of advertising cues. One interpretation of Aaker and Brown’s (1972) classic study of vehicle source effects is that quality claims are more effective in prestige magazines because consumers associate the claim with the media environment, independent of the logical merits of the claim. Similarly, the framework suggests that the “peripheral processing” effect discussed in the context of the elaboration likelihood model (Petty, Cacioppo and Schumann 1983) might be triggered, in part, by media context rather than simply low involvement. In environments where consumers traditionally process advertising cues holistically, by association, consumers would eschew central processing simply because they are not psychologically predisposed to think logically. This may hold true for Internet media as well as other environments. On a gaming site, for instance, users would have no need for product information, and advertising cues would likely be processed associatively, partaking of the ambience and feelings consumers were experiencing while visiting the site.

The Available Processing Capacity Dimension

The “available processing capacity” dimension of our media-class selection framework (Exhibit 2) corresponds to the “involvement” dimension in both the FCB and Rossiter/Percy Grids. Recognizing a large and complex literature on the subject of involvement, Ratchford (1987) notes, “At risk of gross oversimplication, a reasonable summary of the [involvement definitions] would seem to be that involvement implies attention to something because it is somehow relevant or important.” Both the FCB Grid and Rossiter/Percy Grids address a kind of product involvement. The media-class selection model addresses the complement of this, involvement capacity. That is, how much involvement are audience members ready and able to mobilize for an ad in the media context. This capacity depends, in part, on the format and uses/gratifications of the medium itself. The Internet has enormous capacity for involvement, because users can extend the time they spend indefinitely, searching a site, investigating links, and so forth, until they have achieved their usage goals.

Information processing capacity also depends on the competing tasks audience members are likely to face in a particular media environment – what else is vying for their attention (Web 1979). The effects of advertising clutter (Web and Ray 1979) and “zapping” from one station to another during commercials to see portions of other programs (Abernethy 1991; Zufryden, Pedrick and Sankaralingam 1993) are illustrative of only two of the problems faced in a tevision environment. These may be expanded to include everything from side conversations to visiting the bathroom, the net effect of which is to reduce the processing of advertising messages within a media context (Abernethy 1990). The model portrayed in Exhibit 2 suggests that each media class should be classified according to the effect distractions are likely to have on processing capacity. Exhibit 2 suggests some general classifications, but the key to the model is not the content, but the framework and the principles behind it. Newspapers and television both tend to be cluttered, thus leaving consumers with little excess capacity for processing advertising messages. However, by increasing the size of the newspaper ad or the length of the television commercial, these media can be adapted to higher involvement messages.

Evaluating Media to Address IMC Tasks

Implicit in our discussion of Exhibit 2 is the notion that media exposure is governed by a kind of “psychological program” – a set of mental instructions that conditions media consumption behavior. The placement of media alternatives in the exhibit reflects the fact that some of these “programs” address information processing tasks, while others serve more entertainment-oriented or other emotional uses and gratifications. Similarly, some media consumption “programs” are able to process advertising stimuli in depth, while others grant advertising only cursory attention.

The notion of “media exposure programs” is useful when we address the shortfalls of our media-class evaluation framework. While Exhibit 2 provides some very useful guidelines for matching media with promotional tasks, it falls short of capturing the full variation in media environment represented by the alternatives advertisers now face. For instance, properly viewed, the Internet is not a medium, but a collection of media (see Wells and Chen 2000 for the possible dimensions of these differences), used by different kinds of people (Wells and Chen 1999; Rodgers and Cannon 2000) for very different reasons (Stafford and Stafford 1998; Korgaonkar and Wolin 1999; Rodgers and Sheldon 1999). A study by Cannon (1982) suggests that matching value-profiles of editorial content versus advertising messages might be a way to further refine the match of promotional tasks with the media environments to which they might be best suited. Value profiles capture additional dimensions of the “media exposure programs” that drive different media environments. Of course, media planner judgment provides the most widely accepted method for peforming this matching. This judgment may be informed by experience, or formal analyses of the way consumers process advertising media, such as Rodgers and Thorson’s (2000) analysis of interactive media.

Prioritizing Media Vehicles by Cost/Effectiveness

The vehicle evaluation stage of the planning process (box D in Exhibit 1) is both the most controversial and easiest to address. It is at this stage that most planners despair of finding a common ground between old and new media. First, measurements of media exposures are not available for every media class. Second, it is very difficult to compare media as diverse as outdoor posters and direct mail. Comparing old and new media – television with the Internet, for instance, is even more difficult yet.

Granting that comparisons are difficult, the criteria for evaluating cost/effectiveness nevertheless do not really change across media classes. This is true even when new media are involved. The basics go back to the classical ARF model (Advertising Research Foundation 1961), in which media effectiveness is evaluated in terms of media distribution, audience size, advertising exposure, advertising perception, advertising communication, and advertising response. Gensch (1970, 1973) provided a similar framework in which he proposed to evaluate media based on target population weights, vehicle appopriateness weights, commercial exposure weights, commercial perception weights, and cumulative frequency weights.

Exhibit 3:
Evaluating Media Vehicle Efficiency

Evaluating Media Vehicle Efficiency


We suggest that planners may effectively evaluate alternative media vehicles using a much simpler scheme of modified cost per thousand (CPM), as suggested by Exhibit 3. CPM evaluates the cost of message distribution by considering how much money it takes to purchase 1,000 exposure opportunities. An obvious refinement would be to consider the cost of purchasing 1,000 target market exposure opportunities, or CPM TM. An even more useful index would be to estimate the cost of actually creating 1,000 effective advertising exposures, however that exposure is defined. We can call the end result the cost per thousand effective target market exposures (CPM ETM), the ultimate basis for comparing media vehicle cost/effectiveness, or efficiency, with a given media class.

If M represents the media audience (in thousands), T the target market (also in thousands), T M would represent the intersection of the two, or the target audience. We may use p(e|m) to represent the probability that a given individual, e, has been effectively exposed to the advertising message, given that s/he is also a member of the media audience (event m). The measure of media vehicle efficiency would be as follows (Equation 1):










TM * p(e|m)


In fact, the estimates required for CPMETM are not particularly demanding. However accurate their data, planners must estimate the size of their audience, and better, the size of their target market audience. Purveyors of media space recognize this and generally go out of their way to provide these data, be it from subscriber studies or some independent source. Furthermore, the general logic of “prototyping” provides a useful way for extending existing data sources to “unmeasured” media (Baron 1990/1). It suggests that media can be grouped into categories with relatively similar audience characteristics. One or more of the group are then taken as a prototype, and data are then “borrowed” to help estimate the characteristics of the unmeasured vehicle (Cannon and Boglarsky 1992). In the industry, the term, “prototyping,” is generally applied to magazine media. However, the logic is widespread. For instance, in newspapers, it is common to take circulation figures and multiply them by average readers-per-copy figures from other newspapers, thus yielding an estimate of total audience.

Of course, in Internet media, actual audience figures are generally captured by the web-site protocols. However, the logic of prototyping can be helpful in estimating target market audience. Hit rates and click throughs do not necessarily tell the site manager what proportion are actually in the target market for different products that might advertise on the site. If another site, or perhaps an outside research agency, were to conduct a study of target market concentration among web-site users, the results could be used to “prototype” data for other sites as well. To illustrate, consider an Internet campaign that is targeted to people who have purchased an SUV in the past 12 months. Simmons or MRI conduct large-scale syndicated research studies in which they ask respondents (among other things) to indicate various aspects of their web usage behavior. They also ask questions regarding product and brand usage. Their cross-tabulated reports would indicate the proportion of Internet users who had purchased an SUV. This, in turn, can be used as an estimate of the conditional probability of target market membership (SUV purchase), given media usage (a particular type of Internet activity) – p(t|m). Let M represent the number of people in your Internet audience and T the number in your target market. Equation 2 gives the estimated intersection of the two (TM), or the target market audience:




M * p(t|m)


Of all the estimates required for the analysis of media vehicle efficiency (CPMETM), p(e|m) is the most troublesome. Planners tend to question the meaningfulness of comparing direct mail campaigns, where CPMETM may be $1,000 or more, with mass media campaigns, where the figure may be well below $10. Note, however, that we are arguing that this figure should be used only to compare media vehicles within a given class, where they are assigned the same IMC task. The differences are ironed out when we consider the probability of advertising response, as we will in the next section. Consumers vary dramatically in how they respond to different kinds of media exposue. Outdoor advertising exposures may be inexpensive, but they are ineffective for eliciting consumer purchase responses in most situations. This makes direct mail much more cost/effective. The same is true for the Internet. The fact that CPMETM for Internet campaigns might be higher than for mass media campaigns does not suggest that mass media are more efficient. The objectives are usually quite different.

Adjusting the Schedule to Maximize Frequency Value>

Selecting the most efficient media vehicles does not necessarily produce an efficient media schedule, even within a given media class. The problem is that media vary in the degree to which they overlap, and duplicate exposures have a different impact on consumers than original exposures. We have already noted three generations of tools that have sought to address this issue.

First was reach and average frequency. This recognized that some exposures were duplicates, but it failed to account for frequency distributions – the fact that not everyone reached by a media schedule receives the same number of exposures.

Second, effective frequency planning (EFP) sought to address the problems with average reach and frequency by identifying an effective level of exposure that audience members must receive before they are counted as “exposed.” However, this ignores the fact that there is no single level of exposure required for advertising to be effective. Typically, the first exposure is most effective, followed by diminishing returns for subsequent exposures.

The third generation is only now being discussed. In it, frequency value planning (FVP) seeks to address the flaws in EFP by acknowledging the exposure value of every different level of frequency to which audience members are exposed. As with our models of media class selection (box C of Exhibit 1) and media vehicle evaluation (box D of Exhibit 1), it lends itself equally well to new as old media.

The FVP process is easiest to understand when it is broken down into its component steps. Exhibit 4 does this. Each of steps is discussed below.

Exhibit 4:
The Frequency Value Planning Process

The Frequency Value Planning Process


Step A: Developing Media Strategy

Boxes C, D and E of Exhibit 1 do not constitute media strategy. They are analyses that support it. The strategy itself involves a general plan for targeting of media to people in support of the firm’s IMC program. The target audience’s response to media exposure – the basis for evaluating the effects of frequency — depends on the advertising problem and the interaction of the advertising with the target market. For instance, a campaign that uses the Internet to deliver critical product information to people who are in the market for a new car would have a characteristic pattern of advertising response that is quite different from a campaign using television advertising to create brand image in the minds of future car buyers.

The key to this step is identifying key factors affecting the shape and the slope of the advertising response curve. A number of studies have addressed this topic over the years. Unfortunately, the work has been largely judgmental, with relatively little rigorous classificational research or testing of theorized relationships. Foote & Belding Communications (FCB) developed a comprehensive project to identify the factors that determined the need for more or less frequency (Ostrow 1982; Wray 1985). While the FCB propositions were not anchored in specific studies found in the literature, they represented a comprehensive effort by advertising practitioners who were seeking to develop a valid, workable system for estimating advertising response, addressing marketing factors, copy factors, and media factors. Cannon (1987) reviewed the literature and developed 27 theoretical propositions regarding the relative need for frequency. Presumably, these would be valid for new and old media alike.

Step B: Developing a Trial Media Schedule

Step B in Exhibit 4 involves the construction of an actual media schedule, much as a planner would when using EFP. The only difference is that the planner is working against a different criterion. An effective reach schedule of 3+ means the planner will try to place ads in vehicles that have relatively high audience overlap if they have a relatively small budget with which to work. In contrast, if the planner were working against a frequency value criterion, the plan would generally seek to minimize duplication and extend reach as much as possible (Ephron 1995; Jones 1995). This is because the advertising response curve is typically concave (characterized by continually diminishing returns). Therefore, lower levels of frequency deliver relatively higher value to the schedule.

Step C: Estimating Advertising Exposure

Audience research services typically provide media exposure, or opportunity to see (OTS), data. Obviously, any audience response will come from actual ad exposures, not OTS. Therefore, an adjustment must be made to convert vehicle to advertising exposures, using what we characterized as p(e|m) in our earlier discussion of CPMETM. While a considerable literature exists regarding this subject, very little has been written to provide practical guidance in making these adjustments. This speaks to the difficulty of making meaningful generalizations. It also explains why most media planning models have relied on OTS rather than advertising exposure, notwithstanding the broadly accepted criticisms of the approach.

We will argue that, regardless of how difficult the process of estimation, or how inaccurate the estimates, the issue cannot be ignored. Cannon and Riordan (1994) suggest that it was a failure to consider the problem of advertising exposure that caused the industry to misinterpret McDonald’s classic brand switching study (Naples 1979), thus setting the stage for EFP. The Cannon and Riordan analysis suggests that even the crudest estimates of advertising exposure rates would have unmasked the problem, and helped head off two decades of misguided industry effort. The fact is that we know advertising exposure rates are substantially lower than vehicle exposure rates. To ignore the fact is analogous to financial executives ignoring the time value of money, simply because they have no accurate way to predict future interest rates.

The adjustments become particularly important when we seek to compare conventional with Internet media. Internet media often measure actual advertising exposure, signaled by such indicators as click-throughs. Ignoring the difference would put these media at a disadvantage relative to media where “exposure” is really defined by OTS. Furthermore, Internet media themselves vary in measurement approach, with some measuring web-page exposure (“hits”) while other measurements are based on actual audience response. Again, the adjustments are important if we are to compare the media alternatives.

We can draw on efforts from conventional media to illustrate two different approaches one might take to develop a system for estimating exposure rates. One is to develop a set of norms, or adjustment rules, for discounting the value of OTS. For instance, research that observed television audience members found eyes-on-screen time averaged 32.8% for commercials compared to 62.3% for programs (Krugman, Cameron and White 1995). This suggests that advertising exposure would be only (32.8/62.3=) 52% of OTS.. Abernethy’s (1990) detailed review of observational and survey studies led him to estimate 32% television commercial avoidance, or 68% advertising exposure. These figures provide a basis for adjusting OTS figures downward to represent actual advertising exposures. Beardon, Headen, Klompmaker and Teel (1981) reviewed studies addressing attention levels for daytime television, noting that they varied between 20% and 50% of program ratings. In prime time, attention levels were reported at 76% of program ratings for station-break and 84% for in-program commercials. Again, a planner would adjust the average exposure estimates up or down, depending on whether an ad was placed in daytime or primetime television, in a station-break or in a program. For magazines, Roper Starch publishes a book called Adnorms, in which average advertising exposure rates are given for different categories of products in different magazines. These exposure rates can also be used, much as the television data we have discussed to adjust OTS estimates. For instance, if the ad exposure rate were 50% for a magazine that reached 10% of the target market, the reach used to estimate frequency distribution would be (.5 x .1 =) 5%.

The second approach is to use an estimating formula, such as a regression model (Cannon 1982). Philport (1993) discusses factors that might be used to estimate magazine exposure. Donthu, Cherian and Bhargava (1993) discuss the factors that might be used to estimate exposure rates in outdoor advertising. Again, the same basic approach would work equally well or even better for new media.

Step D: Estimating the Frequency Distribution

As a rule, frequency distribution formulas are based on estimates of the probability that a given person will be exposed to a media vehicle, or the probability of OTS. Our discussion of step C above suggests that planners should use the probability of advertising exposure instead, that is, multiplying OTS by p(e|m). This does not change the actual calculations. It simply yields a more realistic estimate of exposure frequency distribution.

The literature suggests that sequential aggregation methods provide what is perhaps the most practical tool for estimating the distribution, since they strike a balance between theoretical grounding, accuracy and speed of computation (Lee 1988; Rice and Leckenby 1986). Such methods are also inherent in some proprietary packages used by media planners (Lancaster 1993; Liebman and Lee 1974). However, the specific frequency estimation program is not important. Any frequency distribution works equally well in FVP, as long it is accurate. Different approaches tend to work better for some media than others, so planners need to consult the literature to determine which is likely to work best for any give aspect of their IMC program. Leckenby and Hong (1998) evaluated the accuracy of six models for use in Internet media and found that all but the binomial distribution model were accurate within acceptable levels.

Step E: Estimating Advertising Response

Step E seeks to estimate the level of audience response associated with each level of the advertising frequency distribution (step D). It is here that our model really addresses the concerns of critics such as Hoffman and Novak (2000). Media planning begins with media, and ultimately, advertising exposure, because this is what media deliver. However, it is not enough to consider whether to consider whether people see a television ad aimed at creating product awareness. We must ask whether it really makes people aware of the product. Similarly, it is not enough to consider whether people hit on a business-to-consumer direct-response web site. We must ask whether the hit really results in a consumer response.

One way to look at the issue is to think of advertising response value as a type of conditional probability — p(r|ei), where r represents audience response, and ei represents effective exposure to level “i” of advertising. That is, how likely are consumers to respond to the advertising message in the desired manner given 0, 1, 2, 3 and so forth exposures? If we plot these response probabilities, they form an advertising response curve. Returning to our earlier comparison of an outdoor versus a direct mail advertising campaign, the CPMETM for direct mail campaign is much higher than that of the outdoor campaign. However, p(r|ei) for the outdoor campaign is virtually zero at every level of exposure. The direct mail campaign would be likely to elicit positive advertising responses, so it would be much more cost-effective.

As we have noted, the shape of advertising response curve will typically be concave. That is, response to increasing levels of advertising exposure is characterized by continually diminishing returns. Such a curve can be represented by the formula shown in equation [3] (Cannon, Leckenby and Abernethy 1996).




M * (1-e(-a-b*i) )







Advertising response value, or what we have referred to as p(r|ei)




Maximum response value, or p(r|e)




Parameter representing the Y-intercept (p(r|e0))




Parameter representing the slope of the curve




the number of advertising exposures



Note that “e” in the equation represents a natural logarithm, not the event of effective exposure, as in p(r|ei)

Note that the equation has three key parameters: M, a and b. M and a can be estimated directly. b can be determined by using equation [4] with an estimate of the response value to the first advertising exposure (i.e. R1, or the value of R where i=1).







The rationale for using a mathematical curve rests in the fact that advertising response is not capricious. It operates according to principles. If the principle is diminishing returns, as we have suggested, the incremental response value of advertising must be lower for each subsequent exposure. How much lower depends on the situation, which, in turn, is reflected in the slope and magnitude of the curve. If you experiment with different values, you’ll find that there is virtually no variation in the way you can plot hypothetical response values without violating the principle of diminishing returns, given an initial level of response (response level with zero advertising exposures), a minimum response (response to one exposure), and a maximum response (response to an infinite number of exposures). These values are determined by the mathematics of the curve. Again, this holds true regardless of the nature of the medium, be it conventional, Internet-based, or whatever.

Some researchers argue that advertising is often characterized by an S-shaped curve, where low levels of advertising generate minimal returns, increasing rapidly once advertising has reached some threshold level (Ackoff and Emshoff 1975; Pavlou and Stewart 2000). The increase then begins to taper off again with diminishing returns. We have noted that the weight of current evidence from the literature appears to suggest that S-shaped curves seldom characterize real advertising situations (Simon and Arndt 1980; Schultz and Block 1986). Nevertheless, the FVP model will accommodate any pattern of advertising response. One need only supply the advertising response values for each level of exposure.

Step F: Calculating Frequency Value

The response value of a campaign is simply the weighted sum of responses for audience members exposed at each level of the frequency distribution. To illustrate, consider a specialized Internet banner program designed to stimulate click-through to a related web site containing product information. Based on a preliminary test, you estimate that 50% of the target market is likely to visit the web site and see the banner. 40% visit once, 10% visit two or more times. Your study shows that there is a 25% chance that people who visit the site once will click on the banner, while the probability increases to 30% for people who visit two or more times. The “click-through” response value for the one-visit group would be (.40 x .25 =) 10%. The value for the 2+ exposure group would be (.10 x .30 =) 3%. The total response value of the campaign would be (.10 + .03 =) 13% (Exhibit 5). That is, 13% of the target market is likely to click through to the target site, which was the objective of the campaign.

Exhibit 5:
Estimating Frequency Value for a Hypothetical Internet Program

Level of Exposures

Frequency Distribution

Probability of Response

Frequency Value













Total frequency value


Frequency value planning allows us to take our analysis a step farther. The company may estimate that the monetary value associated with 1% of the target market clicking through to the target site is $10,000. The total value of the Internet program, then, is ($10,000 x 13 =) $130,000.

Step G: Evaluating the Schedule

Obviously, the schedule’s frequency value can be increased by simply increasing the media weight. However, more weight costs more money. Furthermore, in most cases, the increase in value will be subject to diminishing returns, resulting from a concave advertising response curve. To address this problem, simple frequency value, whether in percentages or dollars, is not as useful as frequency value per GRP, or in this case TRPs (target market rating points). In the case portrayed in Exhibit 5, we don’t know how many TRPs the schedule includes, because we don’t know how many high-level exposures (more than 2) there are in the distribution. For illustrative purposes, assume that the schedule included 65 TRPs. The value per TRP would be (13%/65=) 20%. That is, 20% of all target market exposures result in the desired audience behavior. Second, the monetary value is ($130,000/65=) $2,000 per TRP. As long as the cost per TRP is less than $2,000, the program will pay out.

Step H: Comparing alternative schedules

Finding that one’s media schedule yields a positive return (referring to our discussion above) does not mean it is the best possible schedule. Step H is shown in Exhibit 4 as a feedback loop, in which planners begin developing a new media schedule, in hope of finding a more efficient combination of media vehicles. Cannon and Riordan (1994) spoke of “optimal frequency planning.” FVP follows their logic, but it recognizes the fact that planners will never be sure the schedule is optimal. Rather, it is only better or worse than other schedules being considered.

Step I: Implementing the Schedule

As with all media plans, the ad space or time should be purchased in the most cost-effective manner possible. Our emphasis on tools for developing effective media strategy in no way detracts from the importance of these tactical issues. Indeed, FVP can be used at the buying level as well to ensure that the actual media buys hold to the principles of value-oriented media scheduling. But the basic parameters of the buy will be determined by the schedule that emerges from the evaluative process portrayed in Exhibit 4.

Summary and Conclusions

The premise of this paper is that convetional media planning procedures, or at least the latest thinking regarding conventional media planning, can also be applied to new media. The term, “new media,” is deliberately vague. New media would include the full range of emerging media alternatives, from global satellite television to wireless interactive media. We have focused on the Internet, both because of its enormous significance in today’s world, and because it has been to focus of allegations that conventional media planning will not work on new media (see Hoffman and Novak 2000). However, the Internet represents only one of a much larger set of emerging interactive media (Katz 2000).

Our conclusion is not only that all media can be held to a common standard, but that they actually should be. In a world where efficiency is the watch word of business, we cannot afford to trust millions of dollars to the whims of creative judgment. Not that creativity is unimporant. It is more important than ever! However, it is a necessary, not a sufficient condition for media planning success. The system we have proposed seeks to put all media on an equal footing, estimating the actual cost/benefit of the programs in which they are utilized.

The traditional argument against cross-media comparisons is that media vary so much in their capabilities that they cannot be meaningfully compared. We address this by suggesting that each media program within a larger integrated marketing communications effort plays a unique and important role. The effectiveness of the program should be evaluated in terms of the specific audience response the program is designed to achieve. Thus, if a direct mail campaign has an average CPMETM of $1,000, this must ultimately be compared to the frequency value the program develops. Doing this would likely show that direct mail is much more cost/beneficial for eliciting actual consumer purchases that billboards, even though the CPMETM for a billboard campaign would be much lower.

The system we have proposed develops measures of value for an entire media schedule. It estimates the actual percentage of a target population responding to the campaign in a desired manner. By extension, establishing the value of this response enables planners to put an actual dollar value on each media program. These can be further evaluated by their cost efficiency, evaluating the total response they are expected to elicit from audience members by the media effort (GRPs) it took to achieve this response.

The system is neither complete, nor perfect. First, implementation requires a great deal of judgment. This, in turn, requires additional theory. For instance, where in the matrix shown in Exhibit 2 do various media actually fall? How much does this vary by market segment? What are the realistic ranges in which different advertising response curves fall? How do we determine which one actually applies in any given situation?

The second problem is that, as we consider more interactive media, it becomes difficult to associate advertising response to a specific number of advertising exposures (Pavlou and Stewart 2000). This, in turn, invalidates the notion of the advertising response curve.

The third problem with the proposed system is that the effects of one type of media exposure might interact with another. Indeed, they should interact with eachother if the concept of IMC is truly being used effectively. After all, what is media synergy if not the interactive effects of the various media. But, when media exposures interact, evaluating the cost/effectiveness of each media program separately does not provide the complete picture. For instance, the value of using magazines to stimulate Internet usage cannot be determined by evaluating the magazine and Internet programs separately. The desired response to magazine campaign would be Internet usage; Internet response would purchase, or some other similar criterion. But if the magazine campaign were cut back for lack of efficiency, the Internet campaign would suffer as well – something that did not show up in the FVP analysis for the Internet program.

None of these problems is insurmountable. They are merely an indication that we need to do more work. In fact, the problems are not new. With respect to the need for judgment, this has always been the case in media planning. The system we are proposing is simply a method of making judgments more systematic.

We may offer a similar response to the second problem. When we examine the actual nature of advertising interactivity (see Heeter 2000), we see that media have always been interactive to some degree. The fact that many new media are more interactive suggests that we have an even greater need to look for ways to better account for the interactive effects, but it is not a reason for discarding conventional response-oriented planning systems.

Finally, with respect to media interaction, we must again supplement our model with judgment. To say that we need to consider how one media program affects another is not to ignore the importance of media programs’ non-interactive value. Media campaigns have always interacted with each other to some extent. The advent of IMC and the incorporation of new types of media suggests that we should redouble our efforts to quantify the effects of media interactions, but, in the meantime, judgment will have to suffice.


Aaker, David A. and Phillip K. Brown (1972). “Evaluating Vehicle Source Effects,” Journal of Advertising Research 12:4 (April), 11-16.

Abernethy, Avery M. (1990). “Television Exposure: Programs vs. Advertising,” Current Issues and Research in Advertising 13, 61-77.

Abernethy, Avery M. (1991), “Physical and Mechanical Avoidance of Television Commercials: An Exploratory Study of Zipping, Zapping and Leaving,” Proceedings of the 1991 Conference of the American Academy of Advertising, 233-41.

Advertising Research Foundation (1961). Toward Better Media Comparisons. A report of the Audience Concepts Committee, Advertising Research Foundation, New York.

Agostini, J. M. (1961). “How to Estimate Unduplicated Audiences,” Journal of Advertising Research 1:3 (March).

Baron, Roger (1990/91). “Using the Profile-Distance Method to Select Unmeasured Magazine Prototypes.” Journal of Advertising Research 30:6 (December/January), 11-18.

Beardon, William O., Robert S. Haden, Jay E. Klompmaker and Jesse E. Teel (1981). “Attentive Audience Delivery of TV Advertising Schedules,” Journal of Marketing Research 18:2 (May), 187-191.

Bower, John (1963). “Net Audiences of U.S. and Canadian Magazines: Seven Tests of Agostini’s Formula,” Journal of Advertising Research 3:2 (March), 13-20.

Caffyn, J.M. and M. Sagovsky (1963). “Net Audiences of British Newspapers: A Comparison of the Agostini and Sainsbury Methods,” Journal of Advertising Research, 3 (March) 21-24.

Cannon, Hugh M. (1982). “A New Method for Estimating the Effect of Media Context,” Journal of Advertising Research, 22:5 (October/November), 41-48.

Cannon, Hugh M. (1987). “A Theory-Based Approach to Optimal Frequency,” Journal of Media Planning 2:2 (Fall), 33-44.

Cannon, Hugh M. (1996). “Beyond Effective Frequency: Getting More from Your Media Schedule through Frequency Value Planning.” MRCC Review (May), 1-7.

Cannon, Hugh M. and Cheryl Boglarsky (1991). “A Framework for Creative Strategy Development. Proceedings of the 1991 Annual Conference of the American Academy of Advertising, April 1991, pp. 102-103.

Cannon, Hugh M. and Cheryl Boglarsky (1992). “Matching Unmeasured Magazines to Target Markets: A Test of Two Methods.” Proceedings of the 1992 Annual Conference of the American Academy of Advertising, April 1992, pp. 95-99.

Cannon, Hugh M, John D. Leckenby, and Avery M. Abernethy (1996). “Overcoming the Media Planning Paradox: From (In)Effective to Optimal Reach and Frequency.” Proceedings of the 1996 Conference of the American Academy of Advertising, March, pp. 34-39.

Cannon, Hugh M., John D. Leckenby and Avery Abernethy (2001). “Beyond Effective Frequency: Evaluating Media Schedules Using Frequency Value Planning.” Working paper, Wayne State University Department of Marketing Working Paper Series, paper 2001.001 [URL: 2001.001.pdf]

Cannon, Hugh M. and Norman Goldring (1986). “Another Look at Effective Frequency,” Journal of Media Planning 2:1 (Spring), 29-36.

Cannon, Hugh M. and Edward A. Riordan (1994), “Effective Reach and Frequency: Does It Really Make Sense?,” Journal of Advertising Research, 34 (March/April), 19-28.

Chandon, Jean-Luis (1986). A Comparative Study of Media Exposure Models. New York: Garland Publishing, Inc.

Claycamp, Henry J. and Charles W. McClelland (1968). “Estimating Reach and the Magic of K,” Journal of Advertising Research 8 (June), 44-51.

Colley, Russell H. (1961). Defining Advertising Goals for Measured Advertising Results. New York: Association of National Advertisers.

Danaher, Peter J. (1988). “Parameter Estimation for the Dirichlet-Multinomial Distribution Using Supplementary Beta-Binomial Data,” Communications in Statistics A17,6 (June), 777-778.

Danaher, Peter J. (1989). “An Approximate Log Linear Model for Predicting Magazine Audiences,” Journal of Marketing Research 26:4 (December), 473-9.

Danaher, Peter J. (1991). “A Canonical Expansion Model for Multivariate Media Exposure Distributions: A Generalization of the ‘Duplication of Viewing Law’,” Journal of Marketing Research 28 (August), 361-7.

Danaher, Peter J. (1992). “A Markov-chain Model for Multivariate Magazine-Exposure Distributions,” Journal of Business and Economic Statistics 10:4, 401-7.

Donthu, Naveen, Joseph Cherian, and M. Bhargava (1993),”Factors Influencing Recall of Outdoor Advertising,” Journal of Advertising Research 3 (June/July), pp.64-72.

Ephron, Erwin (1995). “More Weeks, Less Weight: The Shelf-Space Model of Advertising.” Journal of Advertising Research 35:3 (May/June), 18-23.

Gensch, Dennis H. (1970). “Media Factors: A Review Article,” Journal of Marketing Research 7:2 (May), 216-225.

Gensch, Dennis H. (1973). Advertising Planning: Mathematical Models in Advertising Media Planning. Amsterdam: Elsevier Scientific Publishing Company.

Green, Jerome D. and J. Stevens Stock (1967). “Advertising Reach and Frequency in Magazines.” New York, New York: Market Math and Readers’ Digest.

Hirschman, Elizabeth C. and Morris B. Holbrook (1982). “Hedonic Consumption: Emerging Concepts, Methods, and Propositions,” Journal of Marketing 46:2 (Summer), 92-101.

Hoffman, Donna L. and Thomas P. Novak (2000). “When Exposure-Based Advertising Stops Making Sense (And What CDNOW Did about It).” URL:

Hofmans, Pierre (1966). “Measuring the Cumulative Net Coverage of Any Combination of Media,” Journal of Marketing Research, 3 (August), 267-278.

Holbrook, Morris B. and Elizabeth C. Hirschman (1982). “The Experiential Aspects of Consumption: Consumer Fantasies, Feelings, and Fun,” Journal of Consumer Research 9:3 (September), 132-140.

Jones, John Philip (1995). “Single-Source Research Begins to Fulfill Its Promise,” Journal of Advertising Research 35:3 (May/June), 9-16.

Katz, Helen (2000). “Interactivity in 2000: An Industry Viewpoint,” Journal of Interactive Advertising, 1(1) (accessed on 1/30/2001).

Kim, Hyo-Gyoo and John D. Leckenby (2000). “Internet Research/Frequency Estimation Accuracy by Data Collection Method.” Proceedings of the 2000 Annual Conference of American Academy of Advertising.

Kreshel, Peggy J., Kent M. Lancaster, and Margaret A. Toomey (1985), “How Leading Advertising Agencies Perceive Effective Reach and Frequency,” Journal of Advertising 14:3, 32-38.

Krugman, Dean M., Glen T. Cameron and Candace McKearney White (1995), “Visual Attention to Programming and Commercials: The Use of In-Home Observations,” Journal of Advertising, 24 (Spring), 1-12.

Kuhn, Walter (1963). “Net Audiences of German Magazines: A New Formula,” Journal of Advertising Research, 3 (March), 30-33.

Lancaster, Kent M. (1993). ADplus: For Multi-media Advertising Planning. New York, New York: Telmar Information Services Corporation.

Leckenby, John D. and Marsha Boyd (1984). “An Improved Beta Binomial Reach/Frequency Model for Magazines,” Current Issues and Research in Advertising, 1-24.

Leckenby, John D. and Jongpil Hong (1998). “Using Reach/Frequency for Web Media Planning,” Journal of Advertising Research 38:2 (March/April), 7-20.

Leckenby, John D. and Heejin Kim (1994), “How Media Directors View Reach/Frequency Estimation: Now and a Decade Ago,” Journal of Advertising Research 34:5, 9-21.

Leckenby, John D. and Shizue Kishi (1984). “The Dirichlet Multinomial Distribution as a Magazine Exposure Model,” Journal of Marketing Research, 21 (February), 100-106.

Leckenby, John D. and Marshall Rice (1985). “A Beta Binomial Network TV Exposure Model Using Limited Data,” Journal of Advertising 14:3, 13-20.

Lee, Hae-Kap (1988). Sequential Aggregation Advertising Media Models. Unpublished doctoral dissertation, The University of Texas at Austin, Austin, Texas.

Liebman, Leon and Edward Lee (1974). “Reach and Frequency Estimation Services,” Journal of Advertising Research 14 (August), 23-25.

Marc, Marcel (1963), “Net Audiences of French Business Papers: Agostini’s Formula Applied to Special Markets,” Journal of Advertising Research, 3 (March), 26-29.

Metheringham, Richard A. (1964). “Measuring the Net Cumulative Coverage of A Print Campaign,” Journal of Advertising Research, 4 (December), 23-28.

Naples, Michael J. (1979). Effective Frequency: The Relationship Between Frequency and Advertising Effectiveness. New York: Association of National Advertisers.

Novak, Thomas P. and Donna L. Hoffman (1996). “New Metrics for New Media: Toward the Development of Web Measurement Standards.” URL:

Pavlou, Paul A. and David W. Stewart (2000), “Measuring the Effects and Effectiveness of Interactive Advertising: A Research Agenda,” Journal of Interactive Advertising, 1(1) (accessed on 1/30/2001).

Petty, Richard E., John T. Cacioppo, and David Schumann (1983). “Central and Peripheral Routes to Advertising Effectiveness: The Moderating Role of Involvement,” Journal of Consumer Research 10:3 (September), 135-146.

Philport, Joseph C (1993), “New Insights into Reader Quality Measures,” Journal of Advertising Research 5 (October/November), pp. RC-5-RC-12.

Rice, Marshall D. and John D. Leckenby (1986). “An Empirical Test of a Proprietary Television Media Model,” Journal of Advertising Research, (August/September), 17-21.

Rodgers, Shelly L. and Hugh M. Cannon. “The Many Faces of Web Users: An Exploratory Study of Functionally-Based Web-Usage Groups.” Paper presented to the 2000 Conference of the American Academy of Advertising.

Rodgers, Shelly and Ken Sheldon (1999). “The Web Motivation Inventory: Reasons for Using the Web and Their Correlates.” Proceedings of the 1999 Conference of the American Academy of Advertising, p. 161.

Rodgers, Shelly and Esther Thorson (2000). “The Interactive Advertising Model: How Users Perceive and Process Online Ads,” Journal of Interactive Advertising, 1(1) (accessed on 1/30/2001).

Rossiter, John R. and Larry Percy (1997). Advertising Communications and Promotion Management, 2nd Edition. McGraw-Hill: New York.

Ratchford, Brian T. (1987). “New Insights about the FCB Grid,” Journal of Advertising Research 27:4 (August/September), 24-38.

Rust, Roland T. (1986). Advertising Media Models: A Practical Guide. Lexington, MA: Lexington Books.

Rust, Roland T. and Robert P. Leone (1984). “The Mixed Media Dirichlet-Multinomial Distribution: A Model for Evaluating Television-Magazine Advertising Schedules,” Journal of Marketing Research 21:1 (February), 89-99.

Schultz, Don E. (1979). “Media Research Users Want,” Journal of Advertising Research 19:6 (December), 13-17.

Schultz, Don E. and Martin P. Block (1986). “Empirical Estimation of Advertising Response Functions,” Journal of Media Planning 1:1 (Fall), 17-24.

Schultz, Don E. and P. J. Kitchen (2000). “A Response to ‘Theoretical Concept or Management Fashion?'”, 40:5 (September-October), 17-21.

Simon, Julian L. and Johan Arndt (1980). “The Shape of the Advertising Response Function,” Journal of Advertising Research 20:4 (August), 11-28.

Shimp, Terrence A. (1997). Advertising, Promotion, and Supplemental Aspects of Integrated Marketing Communications, 4th edition. The Dryden Press: Orlando, FL.

Stafford, Thomas F. and Marla Royne Stafford (1998). “Uses and Gratifications of the World Wide Web: A Preliminary Study.” Proceedings of the 1998 Conference of the American Academy of Advertising, pp. 174-181.

Vaughn, Richard (1980). “How Advertising Works: A Planning Model,” Journal of Advertising Research 20:5 (October-November), 27-33.

Vaughn, Richard (1986). “How Advertsing Works: A Planning Model Revisted,” Journal of Advertising Research 26:1 (February-March), 57-66.

Web, Peter H. (1979). “Consumer Initial Processing in a Difficult Media Environment,” Journal of Consumer Research 6:4 (December), 225-236.

Web, Peter H. and Michael L. Ray (1979). “Effects of TV Clutter,” Journal of Advertising Research 19:3 (June), 7-12.

Wells, William D. and Qimei Chen (1999). “Surf’s Up — Differences between Web Surfers and Non-Surfers: Theoretical and Practical Implications.” Proceedings of the 1999 Conference of the American Academy of Advertising, pp. 115-126.

Wells, William D. and Qimei Chen (2000). “The Dimensions of Commercial Cyberspace,” Journal of Interactive Advertising, 1(1) (accessed on 1/30/2001).

Wray, Fred (1985). “A Paradox: Advertising Is Fleeting But Its Residue is Hardrock” in Jack Z. Sissors (Ed.) New Directions in Media Planning. Evanston, IL: Advertising Division, Medill School of Journalism, Northwestern University, 137-146.

Zufryden, Fred S., James H. Pedrick, and Avu Sankaralingam (1993). “Zapping and Its Impact on Brand Purchase Behavior.” 33:1 (February-March), 58-66.

About the Authors

Hugh M. Cannon (Ph.D., New York University, 1979) is the Adcraft/Simons-Michelson Professor of Marketing at Wayne State University, a position he has held since 1988. ; e-mail: [email protected].