Consumer understanding and advertising strategy: analysis and strategic translation of laddering data.
Numerous academic studies have addressed this fundamental issue of meaning from the traditional product-attribute perspective (Bass, Pessemier, and Lehmann, 1972; Bass and Talarzyk, 1972; Lehmann, 1971; McAlister, 1982). Under the attribute perspective, product meaning is the observable physical characteristics of the product. This limiting perspective ignores any type of personal meanings of the product attributes.
Recognizing this deficiency, product meaning has been expanded beyond merely attributes to include benefits those attributes symbolize to the consumer. This orientation concentrates primarily on the direct results the product delivers to the consumer through product purchase or consumption (Haley, 1968, 1984; Myers, 1976). More recently, the definition of product meanings has been expanded yet again to include higher levels of abstraction (Gutman and Reynolds, 1979), namely, personal values (Homer and Kahle, 1988; Mitchell, 1983; Vinson, Scott, and Lamont, 1977).
The ever-broadening focus on understanding the consumer meanings that underlie the decision-making process, from attribute to benefit to personal-value perspectives, is primarily driven by the competitive forces in the marketplace. That is, the dramatic increase in the number of competing brands in most product categories forces marketers to look for positionings that are more directly relevant to the decision-making criteria of the consumer. Clearly, understanding the personally relevant meanings that consumers hold for a product, and the new positioning strategies which may stem from these meanings, is invaluable to marketing strategists.
Means-End Theory (Gutman, 1982) presents an appealing framework to more comprehensively represent the consumer meanings that underlie product-positioning research. Rather than focus on a particular level of meanings, the Means-End framework incorporates all levels into a conceptual model, that additionally focuses on the associations (or derived meanings) between the levels. The associations between concepts offer an explanation of how consumers interpret a product attribute as symbolizing associated benefits. Consumers translate product attributes into the benefits (termed consequences) they produce, and benefits are ultimately translated into the consumer's driving value orientation.
For example, a dog food may have an attribute of being "Dry and Crunchy." To a dog owner, "Dry and Crunchy" means the dog food will help deliver the consequences (benefits) of "Cleaner Teeth" and "Healthier Dog." In turn, these consequences help the dog owner to fulfill a personal value, to feel a sense of "Responsibility as a Good Owner." Put simply, the product, as defined by its discriminating perceptual attributes, is the means which satisfies the more personal ends, represented by values.
Importantly, the Means-End framework adds a much richer understanding of how consumers derive meaning from products. Within this framework, meanings reflect the linear pattern of concepts and associations across levels of meaning, which, taken together, serve to explain the underlying reasons why the consumer considers a given attribute to be salient. The cognitive perspective of Means-End Theory can be seen to incorporate attribute, consequence, and value research paradigms into a framework encompassing all three models. It is the associational aspect of the Means-End model that provides a unique perspective on consumers, personally relevant meaning.
Recently, several aspects of Means-End Theory are receiving increased attention. Several articles address research methodology (Reynolds and Gutman, 1988; Valette-Florence and Rapacchi, 1991). Others have addressed the application to positioning strategy design (Olson and Reynolds, 1983; Reynolds and Craddock, 1988; Reynolds and Gutman, 1984). Still others apply Means-End theory as a conceptual framework for the strategic assessment of advertising (Gengler, 1990; Gengler and Reynolds, 1993; Reynolds and Gengler, 1991; Reynolds and Rochon, 1991). This attention has resulted in the broader realization that significant potential lies in using this consumer-based, strategically oriented research framework. However, three major practical problems emerge: (1) the significant time and cost of gathering individual in-depth, means-end (laddering) data; (2) the time and effort required to perform the content analysis of the qualitative responses (steps in the ladders) and the quantitative summaries of the dominant pathways; and (3) the lack of any detailed framework or system to translate strategic options as represented in the summary HVM into a working format for the agency creative staff.
Laddering Data Issues
Data Collection. Several researchers recently addressed the first of these problems: the issue of data collection. Gengler (1990) used an interactive computer program to assess strengths of associations between concepts. The concepts were derived a priori in focus groups. Vallette-Florence and Rapacchi (1990) used a card-sorting task to group concepts which were related. Both of these techniques relied heavily upon a priori definition of concepts to be associated and are in that respect inferior to the open-ended format of laddering interviews. However, both techniques are quicker and easier to administer than laddering interviews. The main issue is whether or not researchers feel their consumers are homogeneous and predictable enough to use a predetermined set of concepts. Walker and Olson (1991) used a paper-and-pencil technique of data collection. In their technique, a questionnaire was administered to a group of individuals simultaneously. This technique shows promise, but precludes the insightful probing characteristic to laddering. In sum, a number of researchers are attempting to find more cost-effective and efficient methods of data collection, but each of these has potential shortcomings when compared with the laddering technique advocated by Reynolds and Gutman (1988). Indeed, it is difficult to justify any small savings in time or money when compared with the enormous cost of inaccurate or incomplete results.
Analysis of Laddering Data. The second problem offers a better place to increase the efficiency of conducting a Means-End study: streamlining and improving the process of analysis. Analysis of laddering data is a cumbersome task requiring several days of effort by highly skilled analysts for even a medium-sized study. The basic analysis steps can be summarized as (see Reynolds and Gutman, 1988, for a detailed description of these tasks):
A. Breaking the raw, conversational data into separate phrases. These phrases are the basic elements upon which subsequent analysis is based. This involves reviewing the verbatim notes or tapes of the discussion probes for the elements that best represent the concepts expressed by each individual subject.
B. Content analysis of the elements selected in Step A.
C. Summation of associations between the content codes, resulting in a quantitative assessment of all paired relationships, termed implications.
D. Construction of a diagram to meaningfully represent the main implications, termed a Hierarchical Value Map (HVM).
Translation of Means-End Results into Strategy and Creative Copy. The third problem involves the lack of any detailed framework to translate strategic options from a laddering study into a working format for the agency creative staff. Although the issue of divining strategy from results is often discussed (Olson and Reynolds, 1983; Reynolds and Craddock, 1988), the focus generally concentrates on definition of the strategy and ignores the issue of translating that strategy into creative, executional concepts for advertisements. Discussing advertising strategy, Little (1979) states: "Good strategy requires imagination and style and always will. At the same time, strategy emerges best from a foundation of reliable facts and sound analysis." Integration of creativity and Means-End study results to develop a strategy and design creative executions is an art. To become accomplished at any art requires the development of technique. A key issue, then, is the outlining of techniques which can aid creative staff in developing executions from strategies.
Purpose of this Paper. The raw Means-End data is the key building block from which all subsequent analysis is based. Although other approaches to data collection have been proposed and are being pursued, the costly interviewing process of laddering is often a necessity. The process is not subject to streamlining without sacrificing quality of understanding of the meanings that drive consumer decision-making. Hence, rather than focus on data collection, the dual purpose of this paper is to deal explicitly with the analysis and strategic implementation of laddering data. The following sections of the paper deal with a brief background of means-end theory and laddering, a discussion of improvements in laddering analysis, and the use of laddering results to aid in developing potent creative copy.
Specifically, this paper first details the use of an interactive software tool that expedites the rather cumbersome and time consuming analytic steps outlined above (Gengler and Reynolds, 1989). Content analysis is a major portion of qualitative analysis. Although labor intensive and often tedious, great care and skill must be used in the content analysis, as the results are the basis of all subsequent analysis. Often, content analysis is an iterative task, where the analysts may recode data several times, combining categories, splitting categories, eliminating or creating new categories, until they feel they have achieved the optimal solution. This stage of the analysis process is drastically improved by the use of interactive computer software, so that the content analysis can easily be reviewed and modified.
In addition, automating the process allows analysts to develop several separate or aggregate analysis based upon demographic segmentation. After a first analysis, each coding change done by hand can result in many hours of recalculation for the next analysis; whereas an automated tool can help the analyst reach the same point in a matter of seconds. Using the system we describe here, the summation of associations between content codes can be performed almost instantaneously, and analysts can experiment with different HVMs resulting from these summations quickly and easily. This experimentation would require days of repetitive effort by hand. Essentially, the use of an interactive software tool described here moves the analysis of laddering responses from being a rough, one-shot subjective analysis to a thoroughly reviewed and easily revised final analysis that a marketing manager can put confidence in. In this paper we shall go through this process to demonstrate how laddering data can best be analyzed.
Secondly, the paper presents a conceptual approach to translating laddering data into a format for creative ideation sessions. This translation bridges the gap between the abstract understanding of meanings latent in strategy specification to the concrete construction of creative messages. To be useful to creative executives, a laddering study should deliver more than just a few vague terms specifying an overall positioning. It should deliver information on how consumers relate different meanings and a basis for idea and message development--not a dogmatic restriction of exactly what an advertisement should say. It is information to feed the creative process rather than restrictions to suffocate it. Methods for the effective usage of laddering data are discussed in the final section of the paper.
Personal values are theorized to be a basis of attitude and preference (Rosenberg, 1956; Vinson, Scott, and Lamont, 1977; Howard, 1977). Means-end theory (Gutman, 1982) is based upon a personal values orientation, in which personal values are the motivating "end-states of existence" which individuals strive for in their lives (Rokeach, 1973). Personal values, then, represent individuals, internal, self-relevant, goal states, while products are often represented as a bundle of physical product attributes. Means-end theory simply posits that the way in which these physical attributes of products are linked to personal values of individuals defines how products gain personal relevance and meaning. Thus, a physical attribute of a product is important if that attribute, during product consumption, produces a desirable benefit or consequence to the consumer. In turn, the perceived consequence of product purchase and/or consumption, derives its importance through the extent that this consequence is linked to another higher level consequence and ultimately into an individual's personal value system. A fundamental problem facing consumer researchers is how to ascertain this means-end cognitive structure (attributes to consequences to values) for any particular market.
Laddering (Gutman and Reynolds, 1979; Reynolds and Gutman, 1988) is the standard method for assessing cognitive structure consistent with the means-end paradigm. The laddering process is performed through a series of one-on-one, in-depth, personal interviews. In the process of laddering, subjects are asked to perform a choice or sorting task in order to uncover a preference-based distinction which they use to choose between brands in the market. The interviewer then continues to ask the respondent a series of probing questions to uncover the structural relationships between this distinction and the respondent's personal value system. In other words, the interviewer is trying to elicit the cognitive relationships that give personal relevance to the product preference distinction. These questions are designed in a manner that will not lead the subject to respond in any specific answer but will prompt them to give an answer that reflects, in their own words, their own particular perspective and meaning. Typically, these questions will be short and of a form similar to "Why is this important to you?" A thorough discussion of the laddering interviewing technique can be found in Reynolds and Gutman (1988).
Stages of Computer Analysis of Means-End Data
Analysis of the responses gathered through laddering interviews involves several steps. To illustrate this process, data gathered on consumer choices of food for their canines is used as an example. This data set contains responses from 67 laddering interviews, with one to four ladders produced from each subject. The discussion of analysis will be presented in terms of the stages the analyst must go through, the important considerations at each stage, and how these stages can be facilitated using Gengler and Reynolds, (1989) decision-support tool, LADDERMAP.(*)
Stage A: Specifying Elements of a Means-End Chain. First of all, the conversational nature of the raw data from laddering forces the analyst to separate the ladder responses of each individual into "chunks" of meaning. These chunks correspond to the distinct levels of product meaning identified by the analyst within the ladder. Two important decisions must be made by the analyst at this stage. First, since the interviews are open-ended and conversational in nature, response not germane to the topic must be eliminated. Secondly, specifying what composes a "chunk" of meaning is extremely important since these units are the basis of all further stages of analysis. For example, an extract from a typical laddering interview may go something like "I prefer brand J (of dog food) because it is dry and crunchy and it cleans my dog's teeth when she eats it." In this statement, the subject is actually stating two different "chunks" of means-end information, specifically, the attribute "dry and crunchy" and the consequence "cleans my dog's teeth."
The LADDERMAP software is designed to assist in this stage. An interactive data entry feature is provided. Under this feature, analysts can enter multiple ladders per interview subject and up to ten "chunks" of meaning per ladder. As ladders are entered for each respondent, the analyst is prompted for the first stage of content analysis which is classifying each "chunk" as either an attribute, a consequence, or a value. This classification underlies the theoretical basis of the analysis and aids the analyst in discerning what are and are not relevant "chunks" to include from the verbatim responses. An example of the ladder entry screen is shown in Figure 1, with a ladder from the dog food data.
Stage B: Content Analysis of Means-End Data. Next, the "chunks" of meaning must be content analyzed in order to aggregate and generalize across subjects. This process involves two steps. The first step is to define a dictionary of content codes into which classifications can be made. This involves a preliminary review of the data and the development of a comprehensive (and exhaustive) set of categories into which to classify all of the "chunks." The second step is the actual assignment of each verbatim to these codes. In a well-defined product category, where analysts typically have strong insights into consumers, perceptions and motivations, many of the category codes may be defined a priori. More often, however, in laddering data analysis the steps of code definition and classification process are interwoven and the codes essentially evolve during the classification.
To facilitate this highly labor-intensive and recursive task, which inherently requires intensive human judgment and decision-making, the software allows interactive coding in an easy-to-use format. Actual content from the interviews is shown on the screen, grouped under the codes it has been assigned to. An example of the screen is shown in Figure 2a, where verbatim responses have been coded under the categories of "Flavor" and "Taste." If, upon inspection, it is found that some concepts are assigned incorrectly, they can be easily corrected and assigned to the proper content code. Also, if the initial coding scheme is very specific, many content codes may have relatively few actual concepts assigned to each. For example, the codes of "Flavor" and "Taste" above could be combined under flavor.
Similar codes can then be easily grouped hierarchically under a larger code, making reassignment an easy task. For example, in Figure 2b all of the items under "Taste" are easily moved under the major heading of "Flavor" by simply assigning "Taste" under "Flavor" in the coding scheme. This enables the analysts to split, combine, or redefine categories quickly and easily on-line. The content analysis task is truly the heart of laddering analysis. It is the step where qualitative data (the raw, verbatim responses from the laddering interviews) are converted into nominal codes which can be quantified. Because codes can be easily combined hierarchically within each other when using the software, we recommend a large number of very specific codes when first analyzing the data and gradually combining and grouping similar meanings until a manageable number of approximately 50 remain. However, if you are performing analysis by hand, we recommend attempting to reduce the number of codes to approximately 50 immediately. Although this may result in a slightly higher misclassification, the combination of categories by hand would be restrictively difficult and tedious, besides being error prone. The lexical listing reports from the software, which report what verbatim responses are categorized under each content code, can be referenced at any future point to see what exactly was collapsed into the final content codes. Any further analysis of the data is only as good as the content analysis. Hence, this task should not be underrated in importance.
Stage C: Defining Connections between Content Codes. Once the laddering data is classified into codes, it can be quantitatively analyzed to produce a diagrammatic representation of the meaning structure. The end product of a laddering data analysis is a graphical representation of means-end structures aggregated across all subjects, the Hierarchical Value Map (HVM). An HVM consists of the different content codes derived from content analysis arranged on a map and connected with lines. These lines show the common pathways of meanings, representing how product attributes are related to personal values. The main goal of analysis is the construction of the HVM, which is the framework for assessing strategic positionings in the market place. This involves two stages: determining what connections should be represented on the HVM, or directed graph, and constructing the HVM in a fashion which is easily readable.
Reynolds and Gutman (1988) present a straightforward decision rule for determining what associations should be illustrated in an HVM. Each association is compared with a cutoff level. If an association has a strength greater than or equal to that value, then the association or connecting linkage is illustrated on the HVM. The LADDERMAP software implements this decision rule in its HVM construction algorithm.
The algorithm consists of three steps. First, we construct an aggregate implication matrix, which contains the sums of all of the instances where concepts were linked in the laddering interviews. These associations can be counted in two ways--only the direct associations are included, or all the direct plus indirect associations are counted. To illustrate what direct and indirect associations mean, consider a means-end chain of A [right arrow] B [right arrow] C. Here, we have direct association, from A to B and from B to C and an indirect association from A to C. The sum of direct and indirect associations is an indicator of the strength of a given association. Often, multiple ladders will be gathered from each individual in the course of an interview. The number of ladders elicited from each individual respondent is usually dependent upon the respondent's depth of knowledge or involvement in the product category. Also, a respondent may name an association between two concepts several times in different ladders elicited. In this case the association is only counted once per respondent when constructing the implication matrix in order to prevent bias in the aggregate results.
After the implication matrix is constructed, a cutoff value is selected by the analyst to determine which connections should be represented on the HVM. To assist the analyst in making this decision, a bar chart is provided on screen by the software to show how much variance would be explained by different levels of cutoff values (see Figure 3). Typically, this cutoff value is compared against the aggregate of direct plus indirect associations in each cell of the implication matrix. A binary matrix is formed which contains a (1) in each cell for which the corresponding element of the implication matrix is greater than or equal to the cutoff value and a (O) otherwise. These binary flags indicate which associations or connecting linkages should be illustrated on an HVM. However, in the interest of constructing a meaningful uncluttered HVM, not all of the marked associations are actually drawn as individual lines. Some of the connections indicated in the binary matrix are considered redundant and, therefore, do not need to be illustrated. If, for instance, the matrix indicates X [right arrow] Y, X [right arrow] Z, and Y [right arrow] Z, then the direct connection X [right arrow] Z is redundant since it is captured in the X [right arrow] Y and Y [right arrow] Z relationships. Although some may argue that this conceals the X [right arrow] Z connection, the map would quickly degenerate into an unreadable state if all redundant connections were included. After all of the redundant, "pass through" relationships are eliminated, the binary matrix can then be used to draw the final HVM.
The process of summing the implication matrix and determining the connections to be made is a relatively simple task and takes only a few seconds on a microcomputer, whereas it is an extremely long and laborious task by hand. This allows the analyst to go back and make content-analysis-coding changes and subsequent alternate analysis without concern since the quantitative steps can be reproduced rapidly and effortlessly. Furthermore, automation of the task allows the analyst more flexibility and control over the process. This allows for experimentation with different levels of cutoff values and different coding strategies. Interestingly, the total number of 1s accounted for in the HVM which is reported can be considered a measure of the representativeness of the solution. This percentage index serves as a useful summary measure that should be reported in all laddering research.
Stage D: Drawing the Hierarchical Value Map. After the data is analyzed to determine exactly which associations should be illustrated as connections on the HVM, the final stage of producing an HVM must be performed. Two requirements are imposed on the analyst at this stage. First, the finished HVM must represent a significant number of the associations derived from the raw laddering data. From experience in conducting over 100 studies, the minimum threshold value should never be less than 70 percent with an average number typically in the 75 to 85 percent range. To represent any smaller percentage can cause valuable insights to be lost. Secondly, and perhaps more importantly in many business environments, the final HVM must be easily interpretable by management if it is to be a viable tool. Again, this stage involves qualitative judgments made by a skilled analyst to produce an HVM which is both accurate and aesthetically pleasing, hence a tradeoff between validity and parsimony.
The algorithm discussed in Stage C only determines what connections should be made, but does not actually indicate where nodes should be placed to draw an intelligible HVM. Gengler and Reynolds (1989) presented a heuristic-based algorithm and interactive editing software which can aid analysts in drawing an HVM derived from the binary matrix. This facility allows the analyst to quickly and easily view several HVMs based upon different cutoff points and different coding judgments. Since an aesthetic component of readability affects the interpretability of the HVM, the software provides several interactive capabilities to adjust the map and enhance its readability, such as moving nodes about, cutting and redrawing lines, or renaming nodes to be more representative of the data.
In sum, the construction of an HVM from raw laddering data involves several stages of both quantifiable and nonquantifiable analysis. The major task of bridging from qualitative to quantitative analysis lies in content analysis of the subjects, verbatim responses. Inherently, this is a human judgment task but can be greatly facilitated by the use of interactive computer software. Furthermore, the use of software at this stage facilitates the subsequent quantitative stages of analysis, in particular, summarizing the frequency of the codes by subgroups within the total sample. Although this final stage can be (and is) automated to a great degree by the LADDERMAP software, this is not a task which can be entirely put through a black box to produce useful results. Analyst judgment and decision-making at all stages of the process is a critical component of the analysis.
Interpretation and Strategic Use of Hierarchical Value Maps
The resultant HVM from an analysis of the dog food category is shown in Figure 3, with verbatim examples nested under each code. (This HVM accounted for 83 percent of all of the connections or associations in the raw laddering data, which we refer to as a measure of variance.) Again, the HVM represents the patterns of meaning by which individuals give personal relevance to product distinctions. The thickness of the lines connecting the concept nodes similarly represents the varying frequencies of association.
The HVM can be divided into three fairly distinct levels corresponding to the a/c/v codes. The product attributes (Dry, Dietary, Name Brand, Moist, and Flavor Variety) are located at the lower part of the map. The consequences which are basically of two types, functional and psychosocial, represent the immediate outcomes that the consumer perceives to result from the corresponding attributes. In other words, the desired consequences or outcomes are the immediate, tangible reasons a consumer attaches importance to the attributes. The Values (Love, Belonging, Fulfill Responsibility, and Financial Responsibility) placed at the top of the map represent personally relevant goals or objectives achieved by the lower level consequences.
The connections between the nodes represent personal meanings. These links are actually the key to understanding and using an HVM. This is true for two reasons. First, being able to identify the connections between concepts in the mind of the consumer is essential to understanding the perceptual basis for decision-making. This represents the cardinal insights offered by an in-depth understanding of the consumer. Second, once a positioning strategy is determined, the creative task essentially involves developing words, images, and/or symbols that will create the desired connections in the mind of the consumer. Thus, focusing on the connections between concepts is central to both understanding and using laddering research.
A common method for interpreting laddering data (Reynolds and Gutman, 1988) is to consider the unique pathways of meaning from the attribute to the value level as perceptual orientations or perceptual segments. This is useful, but also has its shortcoming. As a segmentation method, this approach is useful only if the analyst takes into account the method in which the HVM is constructed, namely, that all the concept nodes in a pathway need not be included in the perceptual orientation. This is due to the fact that the HVM is constructed to include related "pass through" nodes through the elimination of redundant connections, minimizing the number of connecting lines required. To avoid this problem, one must check the implication matrix to make sure the unique pathways actually represent key defining elements which are significantly interconnected. This method of drawing an HVM assumes that those reading the map will naturally understand that a link from concept A to concept B and from concept B to concept C implies a link from concept A to concept C, even if it is not explicitly drawn. In most cases, drawing in these implicit connections will render a map unreadable due to its complexity and multitude of crossing lines. If one is dealing with a very simplistic knowledge structure or coding scheme, or if interviewers failed to elicit full ladders from subjects, a map including all connections may then be feasible.
Alternatively, the implication matrix can be converted to a triangular distance matrix and used as input to a hierarchical clustering algorithm (see Klenosky, Gengler, and Mulvey, 1993 for an example). Different attributes, consequences, and values are grouped together by the analysis. The LADDERMAP program creates a file for this purpose, which can be easily used with any standard statistical applications package. Each of these clusters could be viewed as a perceptual orientation and a basis for a psychographic segment. An issue, then, is to assess which of these is the appropriate target market for a given brand.
Making Positioning Decisions Based on Means-End Pathways
Each of the perceptual orientations discussed as segments should be evaluated as a potential product positioning. This is accomplished by benchmarking the strengths and weaknesses of the respective products, using a combination of traditional attitude data and subjective judgment. The objective data provides a sound basis for assessing the lower attribute and functional consequence levels. The more personal psychosocial consequences and value levels related to the competing brands, positionings can usually be accurately assessed from their advertising communications.
Combining the segmentation and the competitive positioning analyses results in the strategic framework from which positioning options can be developed. Basically, four options emerge. The first, and least likely, is discovering a significant yet untapped orientation, one that is not currently being used in the competitive environment. Given the sophistication of today's marketer, this is becoming increasingly less likely.
Option two involves grounding a positioning by establishing ownership of a meaning, essentially creating a stronger link between what is at present a relatively weak association. For example, in the HVM in Figure 3, the linkage between "dry" and "clean teeth" is seen to be weak, therefore, one positioning option would be to build a strong association here, in the context of "healthy dog." "Healthy dog" would then need to be defined in terms of another, higher order meaning, like "prolong life." The net result would be a strategic positioning that communicates to the consumer that the meaning of "dry" [right arrow] "clean teeth" is a discriminating characteristic to satisfy the higher order needs they have with respect to their dog.
Option three involves developing new meanings, essentially forming a meaningful connection between two as yet unrelated concepts. Again, using the HVM in Figure 3, one example would be to connect "flavor variety" to "high quality," thereby tapping into the strong (tightly connected) higher order meanings that stem from "high quality." A simple example of this would be to create "flavor varieties" (at least named as such) or descriptors that would commonly be considered or associated with a superior cut of meat by human standards (i.e., filet, choice, or tournedos). Of course, the higher level association from "healthy dog" to the most appropriate values (given the competitive environment) must also be specified in the positioning.
The fourth option involves creating a new meaning by adding a new attribute descriptor to the consumer lexicon. One example of this type of positioning development would be to define a new attribute that could readily be associated with "nutrition," given the central role it plays in the HVM. A possible alternative would be a "medically grounded supplement" such as a unique combination of needed vitamins and minerals, which could be easily linked to superior "nutrition" and ultimately reinforcing to more personal value drivers at the higher levels. An approach like this could offer significant potential if the specifics of canine nutrition could be defined with a unique contrast to human dietary requirements, essentially creating a new knowledge framework the consumer could use to ground the rational component of his or her decision-making.
As is apparent, the HVM offers more than consumer insight. It is a framework to contrast current positionings and to develop "what if" scenarios which ultimately can become strategic options. Similar to the skill required to construct a representative HVM, the development of strategy cannot be done by a black box algorithm: it requires clear and oftentimes creative thinking.
The specification of positioning strategy based in a Means-End framework using the Model (see Figure 4) is well documented in the academic literature (Olson and Reynolds, 1983; Reynolds and Craddock, 1988; Reynolds and Gutman, 1984). However, to date, no specifics have been forthcoming on how to translate the specification into a framework that creative staff can use to develop executional ideas. The abstract nature of the content codes and the HVM, though grounded in consumer meanings, appears more like a logical set of connections between rather simple, lifeless descriptors. The primary reason underlying this surface interpretation lies in the failure to adequately explain in detail the relevance of the concept of meaning. The first of two illustrations of this inadequacy in both explanation and understanding are made in the prior section, where meanings, defined as the connection between two concept nodes, served as the basis for the development of strategic options. Understanding the critical associative aspect of meaning offers significant potential to solve the strategy-to-creative translation problem that currently exists.
To illustrate, consider the "Super Premium" potential strategy that appears in Figure 5, which creates a new linkage between "flavor variety" and "high quality." Note that the specification here is repeated on the far left and the key strategic elements are presented in the center section. This form of strategy specification offered by the two left most sections appears less than bland to the insightful creative. What is missing, again, is the concept of meaning; for it is the creative goal to create meanings that will make the product personally relevant to the consumer.
The simplicity and brevity of strategy specification in this manner, though apparently limiting, actually has the potential to serve the creative process exceptionally well. Not only is it unrestrictive, it also provides a unique structure for ideation. The focal point of this ideation is the associative aspect of meaning between any two given concepts.
To develop meanings, one must focus on the connecting lines between the concepts and explore the possibilities that maximize the probability that the desired meanings (connections) will result. This task directly feeds the creative process. What is required, then, is to develop executional ideas, scenarios, symbols, and/or feelings that will cause the association of the two concepts in the mind of the consumer. Generation of ideas in this way can initially be accomplished by answering the question: "What will cause the connection to be made?" Once ideas are developed for each of the three key strategic connections (see the right-most section of Figure 5 for rough examples), the blending of these ideas can take place by creating specific scenes that serve to deliver the desired meanings, or an overall executional action plot that embodies all of the key meanings.
The initial form of strategy translation seen in Figure 5 represents the basic underpinnings that would create the desired connection. The goal is to generate specific ideas thereby expanding the creative concept. For example, the "product bridge" linking the Message Elements to the Consumer Benefit could be enhanced by considering product names that infer the "high quality" and the "flavor variety" meaning. Using human meat labels such as choice or filet might accomplish this. In addition, combining the product name with the visual of the pet really enjoying the special, and thereby superior, meal may create both of the desired sets of connections.
The "personal relevance" bridge between the Consumer Benefit and the Leverage Point can be exemplified by demonstrating the good disposition of the pet, such as showing it playing with kids or being well disciplined. Tying this idea into the execution, either before and/or after feeding, offers another example of how the strategic concept can be brought to life. For the "value bridge," connecting the Leverage Point to the Driving Force, a visual demonstration of the affection latent in the bonding of the pet and its owner seems like an obvious executional idea.
Clearly, the sample creative ideas presented are merely examples limited by the lack of time spent and real creative insight. However, these ideas serve to demonstrate how the creative process can bring to life the strategy elements provided in a specification. The creative contribution is obviously the ultimate payoff-the tangible result of the positioning strategy and has to be worked every bit as rigorously as the development of strategic options. It is abundantly clear, however, that the creative output is intended to communicate meaning. Thinking specifically in these terms offers significant potential to focus the creative process.
Figure 6 demonstrates how another potential positioning, "Special Needs," can be developed from the HVM. In this example, a new attribute is created, one that gives the consumer a rational reason for grounding their decision-making (Reynolds, Cockle, and Rochon, 1990). The strategic goal, then, is to provide the consumer with a rational basis to believe the food is more nutritious, and thus superior to the competition, which is accomplished by building the appropriate meaning.
Continuing with this line of reasoning, the absence of specific dog-nutrition descriptors in the HVM offers the possibility of defining what dog nutrition is and how it differs from human nutrition. The meaning of nutrition, as defined by whatever "medically grounded supplements" can be delivered by the product, can then serve to positively differentiate the brand. Once grounded, the scenarios needed to convey the higher level connections can be developed similarly to the previous example.
As the two examples illustrate, understanding the HVM is the key to both specifying strategic options and to translating the options into grist for creative development. The central tenet and primary contribution of the point of view offered here can be summarized as this: the successful implementation of the Means-End approach to strategy is the realization that meaning is everything. Positioning is about meaning. Analysis of consumer perceptions of the reasons that drive decision-making behavior should be framed as a study of meaning. Therefore, the development of strategic communications involves understanding how visual and verbal elements contribute to generate the desired meanings in the mind of the target consumer.
Laddering is one of the most useful qualitative research techniques available to advertising researchers. It provides an opportunity for consumers to respond to choice situations in their own words and express their own feelings, yet provides enough structure to keep the conversation focused exactly on what the consumer thinks about the product category. The analysis of laddering data can be a laborious task, fraught with myriad classification decisions. The analytic tool described in this paper provides a methodology for analyzing laddering data in a more-organized, less error-prone, and less opinionated fashion. This is done without suppressing the amount of information communicated to those who will eventually use the results of the analysis. The results of qualitative research should not be the biased opinion of the researcher. They should not be a selected amusing vignette or two from the best communicators in the sample interviewed. They should represent all of the perspectives of all of the individuals interviewed. Only through a careful analysis process such as we have discussed here can this be achieved.
Yet, even if an analysis has yielded valuable insights into consumer or industrial buyer psyche, these insights are worthless if they are not put into action. The strategic statements and positionings derived from a laddering study must be communicated to creative staffs developing advertising for the product. Furthermore, they must be communicated in a framework that stimulates creative ideation around the chosen positioning rather than restricting the creative staff to an overly specified message content. Such an over-restriction can be a fantastic formula for dry, unexciting advertising. For this reason, we have outlined how strategies derived from a laddering study can be successfully used as a source of ideas for creative staffs. Finding new ways to translate a product's tangible features into customers, key benefits, or to translate benefits into personally relevant feelings and values, is vital to creating advertising which is exciting and cohesive with a brand's chosen positioning. The focus of any communication with customers must be on the lasting product/brand related meanings formed in the customer's memory. This focus will not only help to build messages that contribute to a stronger brand image and positioning but will also help to preempt the creation of messages which, in isolation, may be "good ads" but in a holistic perspective dilute the positioning of the brand and confuse the brandimage.
[Figures 1-6 ILLUSTRATION OMITTED]
(*) LADDERMAP is a registered trademark of Means-End Software. LADDERMAP software can be obtained through Means-End Software, P.O. Box 3364, Camden, New Jersey 08101, ph. (609) 858-6781.
Bass, Frank M.; Edgar A. Pessemier; and Donald R. Lehmann. "An Experimental Study of Relationships Between Attitudes, Brand Preference and Choice." Behavioral Science 17, 6 (1972): 532-41.
--, and W. Wayne Talarzyk. "An Attitude Model for the Study of Brand Preference." Journal of Marketing Research 9, 1 (1972): 93-96.
Gengler, Charles E. "An Architectural Perspective on Advertising Strategy." Unpublished doctoral dissertation, University of Texas-Dallas, 1990.
--, and Thomas J. Reynolds. "A Structural Model of Advertising Effects." In Advertising Exposure, Memory, and Choice, Andrew Mitchell, ed. Hillsdale, NJ: Lawrence Erlbaum, 1993.
--, and --. "Means-End Structural Analysis: Computer Generated Hierarchical Value Maps." Paper presented at EIASM Workshop on Consumer Behavior: Extending the Cognitive Structure Perspective, Brussels, Belgium, November 1989.
Gutman, Jonathan. "A Means-End Chain Model Based on Consumer Categorization Processes." Journal of Marketing 46, 2 (1982): 60-72.
--, and Thomas J. Reynolds. "An Investigation of the Levels of Cognitive Abstraction Utilized by Consumers in Product Differentiation." In Attitude Research Under the Sun, J. Eighmey, ed. Chicago: American Marketing Association, 1979.
Haley, Russell I. "Benefit Segmentation: A Decision-oriented Research Tool." Journal of Marketing 32, 3 (1968): 30-35.
--. "Benefit Segments: Backwards and Forwards." Journal of Advertising Research 24, 1 (1984): 19-25.
Homer, Pamela M., and Lynn R. Kahle. "A Structural Equation Test of the Value-Attitude-Behavior Hierarchy." Journal of Personality and Social Psychology 54, 4 (1988): 638-46.
Howard, J.A. Consumer Behavior: Application and Theory. New York, NY: McGraw-Hill, 1977.
Klenosky, David B.; Charles E. Gengler; and Michael S. Mulvey. "Understanding the Factors Influencing Ski Destination Choice: A Means-End Analytic Approach." Journal of Leisure Research 25, 4 (1993): 362-79.
Lehmann, Donald R. "Television Show Preference: Application of a Choice Model." Journal of Marketing Research 8, 1 (1971): 47-55.
Little, John D. "Aggregate Advertising Models: The State of the Art." Operations Research 27 (1979): 629-67.
McAlister, Leigh. "A Dynamic Attribute Satiation Model of Variety Seeking Behavior." Journal of Consumer Research 9, 3 (1982): 141-50.
Mitchell, Arnold. Nine American Lifestyles. New York: Warner, 1983.
Myers, James H. "Benefit Structure Analysis: A New Tool for Product Planning." Journal of Marketing 40, 4 (1976): 23-32.
Olson, Jerry C., and Thomas J. Reynolds. "Understanding Consumers, Cognitive Structures: Implications for Advertising Strategy." In Advertising and Consumer Psychology, Vol. 1, L. Percy and A. Woodside, eds. Lexington, MA: Lexington Books, 1983.
Peter, J. Paul, and Jerry C. Olson. Consumer Behavior: Marketing Strategy Perspectives. Homewood, IL: Irwin, 1987.
Reynolds, Thomas J., and Alyce Byrd Craddock. "The Application of the Model to the Development and Assessment of Advertising Strategy." Journal of Advertising Research 28, 2 (1988): 43-54.
--, and Charles E. Gengler. "A Strategic Framework for Assessing Advertising: The Animatic vs. Finished Issue." Journal of Advertising Research 31, 5 (1991): 61-71.
--, and Jonathan Gutman. "Advertising Is Image Management." Journal of Advertising Research 24, 1 (1984): 27-36. --, and --. "Laddering Theory, Method, Analyses, and Interpretation." Journal of Advertising Research 28, 1 (1988): 11-31.
--; Brian C. Cockle; and John P. Rochon. "The Strategic Imperatives of Advertising: Implications of Means-End Theory and Research Findings." Canadian Journal of Marketing Research 9 (1990): 3-13.
---, and John P. Rochon. "Means-End Based Advertising Research: Copy Testing Is Not Strategy Assessment." Journal of Business Research 22 (1991): 13142. Rokeach, Milton. The Nature of Human Values. New York, NY: Free Press, 1973.
Rosenberg, Milton. "Cognitive Structure and Attitudinal Effect." Journal of Abnormal and Social Psychology 53 (1956): 367-72.
Valette-Florence, Pierre, and Bernard Rapacchi. "A Cross-Cultural Means-End Analysis of Perfume Purchases." In Proceedings of the Third Symposium on Cross-Cultural Consumer and Business Studies, 1990.
---, and --. pacchi "Improvements in Means-End Chain Analysis: Using Graph Theory and Correspondence Analysis." Journal of Advertising Research 31, 1 (1991): 30-45.
Vinson, Donald E.; Jerome E. Scott; and L. M. Lamont. "The Personal Role of Values in Marketing and Consumer Behavior." Journal of Marketing 41, 2 (1977): 44 50.
Walker, Beth, and Jerry C. Olson. "Means-End Chains: Connecting Products With Self." Journal of Business Research 22, 2 (1991): 111-18.
CHARLES E. GENGLER is assistant professor of marketing at Rutgers University in Camden, N.J. His research focuses on improvements in laddering data collection, analysis, and testing of advertising according to the Means-End paradigm. He has performed consulting work and given executive development seminars on marketing and positioning strategies in the banking, pharmaceutical, casino, and resort industries. Recently, he completed a stay with Grey Advertising in New York through the Advertising Education Foundation's Visiting Professor Program
THOMAS J. REYNOLDS is president and chief executive officer of Wirthlin-Reynolds, a marketing and communications strategy consulting company He is also currently a partner and chief marketing strategist for Richmont Corporation, a private merchant bank. He has served as a professor in the Graduate School of Management and as Director of Marketing Studies at the University of Texas at Dallas where he was recently named Professor Emeritus. He has published more than 50 articles on the development and implementation of values-based strategy As a consultant he has worked with over 40 corporations in over 20 countries operating on marketing strategy problems including such companies as Procter & Gamble, Coca-Cola, AT&T, Intel, and Mary Kay Cosmetics.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Special Issue: Research Input into the Creative Process|
|Author:||Gengler, Charles E.; Reynolds, Thomas J.|
|Publication:||Journal of Advertising Research|
|Date:||Jul 1, 1995|
|Previous Article:||Applying laddering data to communications strategy and advertising practice.|
|Next Article:||Seeing the voice of the customer: metaphor-based advertising research.|