Anda di halaman 1dari 24

FROM MULTI-COUNTRY CONCEPT TESTING/

OPTIMIZATION TO CORPORATE DATABASE AND BEYOND


Johannes Hartmann Howard Moskowitz Jeff Ewald
This paper provides a case history for multi-country conjoint measurement. The project began as an exercise to understand the features and communications of a food product across many countries, but turned into an organizing principle to understand consumers, worldwide, and at the same time evolved into a unique database that could be used by the company again and again for product development. The authors present the pros and cons of the approach and describe how the thinking has evolved.

Johannes Hartmann, Howard Moskowitz, Jeff Ewald

INTRODUCTION: THE EXISTING SITUATION FROM THE CORPORATE BOARDROOM


The existing belief in many companies is that concepts for FMCG (fast moving consumer goods) need to be individualized for each culture. This point of view does not necessarily find itself based on fact, but rather on belief. Management recognizes the growing internationalization of products, the increasing trade among countries, and the emergence of brands that can be called world brands. At the same time there is the perennial reluctance to abandon the local point of view and simply agree that the world is becoming homogeneous. Current trends neither point to a complete homogenization, nor do they point to complete country-to-country differentiation. They point to some in-between reality (Trout and Rivkin, 2000). In actuality, even when a case for globalization can be made on the basis that consumers in many countries declare the same motivations and expectations, a closer look will often reveal slight differences that must be taken into account (Kapferer, 1998). The attractive but vague notion of internationalization doesnt deal with the obvious difference between the what and the how. Similar product formats and product designs can be used across countries. However, the way consumers discover and incorporate new products into their lives is to a high degree a function of culture (Bacon and Butler, 1998). As a consequence, companies need to build a systematic approach to the how they structure the process of developing connecting logic with consumers. This how will vary by cultural background (Mitchell, 1983; Wells, 1975). It must be recognized that, within the organization, a move towards globalization can mark a profound cultural revolution a change that necessarily entails a change in decision-making procedures. Even more so, the change signals in general an evolution in the underlying structure of the organization (Kapferer, 1998). Ideally, these structural changes and new procedures should provide an opportunity for members of the multi-national corporation to collaborate, learn from each other, and cross both the geographic and organizational boundaries separating each other to fulfill the task (co-creativity). The changes should make both corporate marketers and product developers aware of similarities and differences among different countries and cultures. It is apparent that these approaches must take into account regional and even global learning curves. Of course, good methodology is key. But good methodology requires perfect pre- and post work. The process character offers the opportunity for regional teams to learn from each other. If the region teams make conscious use of the

From multi-country concept testing/optimization to corporate database opportunity they can transcend cultural differences that are rife both internally and externally (Cooper, 1999).

IF WHAT NEEDS TO BE DONE IS SO OBVIOUS, THEN HOW CAN CONSUMER RESEARCH CONTRIBUTE?
The job of consumer research is to identify opportunities for products, whether local or international, and where possible develop the knowledge base for these products. The traditional approach to answer this problem follows a series of well-choreographed steps known to most consumer researchers, and by now embedded in the knowledge-development practice of most corporations (Hoban, 1998): 1. Opportunity identification: Identify a product opportunity, either by monitoring trends or by working with consumers and in turn identifying their unmet needs. Occasionally the product opportunity presents itself through technical developments by R&D, such as a new product, new process, or even a new ingredient. At this stage the product opportunity is relatively unformed. It exists, it can be sensed, occasionally its magnitude can be guessed, but there is nothing specific about the product that will fill this opportunity (Haeckel, 1999). 2. Concept development: At this stage the opportunity has been identified and accepted. The concept development stage now seeks to create an idea that will encompass this opportunity. The product concept will tell the developers what the features of the product should be. The marketing concept will tell the agency how to talk about the product. The concept development stage can be short, long, involved, or relatively simple. Occasionally the development stage is skipped entirely, based upon management fiat. It is at this stage that there are opportunities for transnational research. It is furthermore at this stage that the different frameworks for multi-national development make themselves known. There are two ways that companies can build concepts. One way is topdown development, and the other is bottom-up development, respectively. Top-down concept development, the traditional method, waits for a corporate mandate, creates a variety of concepts through ideation sessions, and then selects the best concept. This traditional method is useful in one country, but runs into problems when the concept must be transported from one country to another. The changes in the concept are often done on an ad hoc manner, so in the end the single concept may become two, three or more disparate concepts (Fuller, 1994).

Johannes Hartmann, Howard Moskowitz, Jeff Ewald As described earlier, this top-down approach doesnt create an internal learning curve across cultures. In actuality it effectively proves the obvious truism that all cultures are different. Internally people say this is not working here and they can prove because, it is likely that the connecting logic is not sufficiently culture-specific.

THE BOTTOM-UP APPROACH: STRUCTURE PLUS FLEXIBILITY


A more rational approach for multi-country, multi-cultural concept development builds concepts from the bottom-up using a structured approach. In this case the approach begins with a set of product, positioning, and package features, combined into test concepts through conjoint analysis. Testing these concepts in different countries allows the marketer, product developer, and researcher to create a global model as well as individual country-by-country models showing how each of the concept elements drives interest at the respondent level. Furthermore, by using the same elements across countries it becomes straightforward to segment the respondents by commonality of response to concept elements, producing true trans-national concepts. A key aspect of the bottom-up approach is that it occurs early in the development process, where there are many options open. Rather than development proceeding along a path governed both by opinion and by ideas that might be in either overt or covert conflict, bottom-up development ensures that opinion need not sway the business decisions. There can still be opinion, but scientific corroboration of the opinion goes hand in hand with the opinion, rather than perhaps one advocate in the corporation being proved right and the other advocate by necessity being wrong. Opinion and fact synergize to produce effective, knowledge-based development (Rosenbaum, 1987; Veganti, 1997; Zaltman, Duncan and Holbeck, 1973). The beauty of the bottom-up approach is that it is inherently a collaborative exercise that subtly but inexorably shapes both the organizations structural and procedural dynamics. Different people from different cultural backgrounds have to decide on the input of the study. Hobby horses, apparent logic and my-bosss-favorites could be easily added into the approach. In addition, people are forced to define one set of inputs, a requirement which aligns the methodology so it is agreed to at the start of the project. The leadership team using the bottom-up approach collaboratively should experience and generate the following:

From multi-country concept testing/optimization to corporate database 1. Common understanding and expectations: All participants of the team will need to understand the process and the methodology of the tool used. As participants gain an understanding of the research paradigm and terminology, they easily find individual ways that allow them to own the process. In our experience, conducting a comprehensive project briefing at the beginning establishes a common syntax leading to a methodologycentric common language. The common language is critical. It binds people together. The shared process/methodology become the center of gravity for a group identity. On occasion the country representatives will initially resist the idea of cooperation, but after they are briefed and then realize the new tool offers them full contribution, most contrarians start to align themselves with the process. 2. Understandable outcomes that are anticipated and welcomed: Bottom up methods create a base for accepting the outcomes. Corporate participants and external associates in the ideation project impatiently look forward to, and accept the outcomes. The participants dont need to defend their positions as their positions are intrinsically recognized and respected by the methodology. 3. Innovative approaches create a positive hype in the organization that can breed organizational change. This outcome is especially important when innovations have to create momentum. A new and/or advanced method, embraced with a common purpose and understanding can produce an esprit dcorps and lead to outcomes that never would have been achieved with traditional, well accepted, and therefore less thought provoking, approaches. 4. New language, new bonding, new positioning: The new method automatically creates new language for the team, which creates a bond among the members. The bonding leads to the positive perception of the team and its contributions by other members of the corporation. The value to a company of doing bottom-up concept development lies in the following four outcomes: 1. Enhanced, unique learning at the early development stage, where the options are open. The data encourages cross-cultural learning and acceptance about what features, benefits and positionings work in different countries, and what does not. 2. Harmonization: The resulting output can be immediately harmonized because the concept elements (inputs) are similar or the same across countries. This provides the connecting logic an empirically based hierarchy of ideas based on the degree to which the same inputs (stimuli) produce the same effect (response).

Johannes Hartmann, Howard Moskowitz, Jeff Ewald 3. Trans-national segmentation. The segmentation is truly transnational, independent of country, because one only needs the response of the individual, not the country from which the person comes from. Buzzell and Quelch (1988) recognized that despite the corporate view of globalism with respect to a given brand or product, it is ultimately the individual consumers whose choices remain individualistic and egocentric, and in reality who determine the degree to which a similar or differentiated strategy will be successful. 4. Databasing. The data provides an integrated database that allows the company to create products over time, for the total panel, for a transnational segment, for the country or even for a trans-national segment subsequently particularized for a specific country. The richness of the data creates a de facto product database that has sustaining value for the company. Thus one can use the data to create different products for consumers in a country, or fine-tune the product across countries. This capability solves some of the issues of segmentation by creating a single base idea that can be differentiated as needed.

THE UNILEVER STRUCTURE AND APPROACH TO TOP-DOWN VS. BOTTOM-UP DEVELOPMENT


Unilever is one of the leading, fast-moving consumer goods companies in the world. It has been in existence for more than one hundred years. Through the years Unilever developed a corporate genetic code for development that constituted the foundations for its success. In the same way as other companies, however, Unilever has had to struggle through the years to change its views to keep competitive advantage in its core categories and brands. These struggles were not always easy and not always successful. Unilever has a very unusual organizational structure. It is led by two head offices, one in the United Kingdom and one in the Netherlands. This structure has had advantages as well as disadvantages to the overall decision making performance of the company: 1. Diversity generally breeds caution: This unusual structure created a lot of diversity in terms of views on issues. As a natural outcome, therefore, decisions have always been very balanced and well-thought through. 2. Separated offices can create conflicts: On the other hand: Unilever has had its share of politics which sometimes made it difficult to arrive at any decision.

From multi-country concept testing/optimization to corporate database Unilever always valued the independence of its local companies. This belief was and is an intrinsic part of the Unilever culture. Local independence proved to work very well until the point was reached when in the local country Unilever had to face the increasingly competitive environment. The mass competition forced a re-focus on core brands and core business at the more general level rather than at the local level. It was necessary to lose sight of some of the country specifics in order to become stronger worldwide, and to achieve more efficiencies and economies of scale. The Unilever brand-portfolio broadened over the years. In the 1970s and 1980s almost every new business thrust was linked to the launch of a new brand. As local companies werent disciplined at all by the center, each country developed its own set of brand names for the comparable positionings and categories. In the 1990s Unilever tried to reverse this trend and still works at this reversal. What had evolved into an enormous number of brands has nowadays been reduced to a small set of global leading brands and local jewels. This transformation process has been a painful one, requiring a dramatic change in the way Unilever was doing business and in the way the countries had to be aligned. In the beginning of this transformation process Unilever experimented with different organizational models. Each model was intended to create synergies between countries. Here we talk about Europe specifically, a market where Unilever is exceptionally strong. After a couple of organizational restructurings Unilever finally installed a model in Europe that put the responsibility for the value-creation process, i.e., all brand- and innovation developments, into several European innovation centers. These centers were responsible for the innovation agenda and for the harmonization of the brand portfolio in Europe. The management in the different countries had, of course, a huge say on these agendas. Finally management from the countries was required to introduce the newly developed concepts from these centers into the market place. The very problem of this structure is obvious: countries and centers need to agree what is good for the countries and at the same time good for the regional approach. The situation paralleled what was so clearly and profoundly said in the movie A Beautiful Mind: You have to find out what is best for you individually whilst you have to find out what is the best for the group! One of the consequences of the shift to innovation centers was the necessary evolution of concept development. Top down approaches, which had been regularly used in the past, never delivered a real break-through because they were too vulnerable during the final negotiation step among the different parties. It was impossible to legislate the acceptance of a new idea, no matter how good, because there were too many vetoes in the hands of parties with

Johannes Hartmann, Howard Moskowitz, Jeff Ewald different agendas. The innovation center had to change the concept development paradigm by involving the different countries in the very early phase of the project. The Ideamap approach with the related process steps was one of the approaches that Unilever was testing to fit in with this innovation process and the harmonization through cooperative collaboration at the early stages. IdeaMap helps mold and even in some cases force very diverse teams into community by offering a bottom-up structure. In this structure everyone played an integral role, had a key part, and could not be overlooked. No one had to fight for his personal advantage to win because of what the process might produce in the way of a concept. Everyones opinions and pet ideas could be incorporated into the research, so no matter what happened the group would win, rather than any individual prevailing at the expense of any other individual.

A CASE HISTORY ILLUSTRATING THE APPROACH, AND COMMENTS ABOUT THE ISSUES FROM THE VANTAGE POINTS OF BOTH RESEARCH AND A BUSINESS PROCESS
Our case history involves the creation of an intermediate product positioned to be between margarine and oil, capturing the advantages of both worlds. Creating these products that stand astride two different categories is one of the Unilever approaches to innovation. The melding of two product categories into a new category is a standard innovation strategy. Margarine is one of Unilevers most prominent products. Unilever creates and markets margarine products worldwide, so that the experience with creating a new, healthoriented margarine is both new because of the new-product development, but also a well-trodden path. The primary objective for this particular margarine was to create a European product that would take advantage of the growing concern with health, and position the product as an alternative to cooking oil. The concept had to be ownable by the current Van den Bergh franchise, and work in the four different countries: Germany, Netherlands, Sweden and the United Kingdom as the initial markets. The secondary objective was to learn how to create a product concept from the bottom-up. We created a large-scale database for margarine from the bottom-up, across four countries, in order to demonstrate the usefulness of the approach. In doing so we developed both knowledge of the features that would lead to a good concept, and knowledge about the process of concept creation using this new-to-Unilever approach.

From multi-country concept testing/optimization to corporate database

IDEATION: MIXING OLD AND NEW STEPS


The Van den Bergh group of Unilever had extensive experience in concept development because of its history of introducing European products. Thus, at a general level the notion of creating concepts for the new margarine product was already well respected. In its role as a multi-national company, Unilever was also accustomed to having its operating units think in multi-national terms so that the creation of the concept elements for the IdeaMap conjoint study elements was done with sensitivity to multi-national requirements. Furthermore, the notion of ideation was also well accepted within the Unilever companies, which had early on spearheaded the consumer research techniques, and ensured their use worldwide. The traditional way for concept development was top-down. At the time of this innovation project the management at Unilever was struggling to fit the traditional way of doing things into this development objective. This new oil-margarine interface product represented one of the first forays outside the traditional approach into bottom-up development. The ideation was done with members of the team from the different countries in one room. The ideation encouraged each of the members to contribute nuggets of ideas, or single-minded phrases, rather than fully formed ideas. With members from the different participating countries present, and with the ultimate goal of creating a conjoint study having 300+ elements, the ideation session allowed members of the team from each country to offer an array of ideas without feeling constrained. The qualitative brainstorming was accompanied by qualitative research among consumers in each country, with secondary analysis to deconstruct the ideas used by both Unilever and competitor companies in the general area of margarine, oil and spreads. A sense of the richness of the output can be appreciated by looking at table 1. A total of 316 elements were finally selected from more than 500, after the array of possible elements was reviewed, edited, and culled to remove redundant items. The elements fell into 12 categories, or general groups. The conventional conjoint approach uses equal categories, limited in size. The alternative conjoint method IdeaMap allowed many dozens of elements in a category. A category (e.g. Usage Occasion or Packaging Description) contains related elements. In multi-country work, where there are different cultures with differing sensitivities, and where the database approach may yield similar but modified concepts on a country basis, this capability to deal with many elements is key. It is important to cast a wide net and assess many elements for the new margarine product ... because we dont know what will work (Moskowitz and Martin, 1993).

10

Johannes Hartmann, Howard Moskowitz, Jeff Ewald Table 1 SIX OF THE TWELVE CATEGORIES FOR THE FOUR-COUNTRY STUDY OF MARGARINE, NUMBER OF ELEMENTS IN THE CATEGORY AND AN EXAMPLE OF THE ELEMENT
Category Visuals Product name Packaging description Packaging benefits Product descriptions Functional Benefit Elements 25 28 21 27 27 32 Bag-in-Box visual Product name is: Magic Touch Comes in a bottle similar to olive oil The packaging is sealed to guarantee freshness It's a cooking liquid with a great buttery taste All the natural goodness of oil Example

STRUCTURING THE CONCEPT CREATION APPROACH TO CREATE A MULTI-COUNTRY DATABASE


Ideation of whatever sort usually leads to a large set of elements for the new product, which is the happy outcome of most of the new product efforts at Unilever as well as other companies. It is not so much in the ideation but rather in the formal, structured, yet open approach to work with this ideation output that the interest lies. In order to make the most of the data for current and future work, and in order to embed the way of thinking and the data into the Unilever system, it was necessary to follow a structured approach. The structure followed seven steps, which ensured cross-country cooperation during the course of the up-front work prior to research, standardized field work during research, and a database approach to concept development (concept informatics) after the research: 1. Basic experimental design: We used large-scale-conjoint measurement followed by a large-scale classification study, with both parts of the structure similar across all countries (Moskowitz and Martin, 1993). The approach was designed to be the prototype for future multi-national, crosscultural product studies. It was necessary in this design structure to accommodate several hundred elements in order to represent the contributions of the different cultures to the same product. Conjoint analysis is a well-respected research method with a long and successful history (Bateson, Reibstein and Boulding, 1987; Wittink and Cattin, 1989).

From multi-country concept testing/optimization to corporate database This happy state of affairs ensured its acceptance by Unilever, even among those project participants who had no prior experience with the method. 2. Element bank (raw material) for subsequent databasing: For the conjoint measurement portion, participants in the ideation and subsequent in-house editing by Unilever created a very large bank of concept elements that dealt with different brands, benefits, packages, product features, health features, etc. This element bank would serve as a resource for the study. As much as possible, we held these elements identical across the different countries. The study itself comprised 316 concept elements. Since the goal was to create a corporate database that transcended countries, and could be used both at the time of the research as well as for years afterward, it was important to use as many different concept elements as possible. It became increasingly clear that a great deal of value would lie in the richness of the concept elements and in the analysis of subtle differences of certain descriptors. One or two words stated a little differently might make a huge impact. The more elements to be studied meant more insight and more possibilities for applying the research results to different cultures. a) From these categories and elements, experimentally designed concepts were created, using the principles of experimental design. b) Most concepts contained four or five elements out of the 316. c) Each respondent evaluated 80 elements, embedded in 100 different concepts. d) The combinations rated by each respondent comprised their own experimental design. This design at the individual level allowed us to create a model for the individual respondent showing how the individual elements tested by that respondent drove ratings 3. Execution: The PC-based interviews ensured that the study could be executed anywhere, in an identical fashion, in each country, with 200+ respondents per country (Germany, Netherlands, Sweden, United Kingdom). Each interview lasted approximately one hour. PC-based interviews (and now Internet-based interviews) allowed us to develop a system that could be transported worldwide, without concerns about the ability of the local researcher to execute the study. This capability would prove very valuable for subsequent work in languages that do not feature Latin characters. The PC-based interview meant that many controls could be placed in the interview, such as unique combinations rated by each respondent, rather than having every respondent test the same set of stimuli. These computer-based features ensured that the results of the interview would have less bias. Furthermore, user-friendly PC-based interviews ensured that the approach could be used in countries where

11

12

Johannes Hartmann, Howard Moskowitz, Jeff Ewald face-to-face interviewing might produce covert biases with respondents trying to please the interviewer. The interviewer effect would be minimized. Figure 1 shows the orientation page for the English study. Figure 2 shows the orientation page for the Swedish study. In each study the respondents rated the test concepts on three attributes: interest, fit to a local brand (e.g., I Cant Believe Its Not Butter for the UK; Milda for Sweden, etc.) and use (in place of oil versus in place of butter/margarine). Each respondent evaluated 100 systematically varied concepts on each of the three rating scales. The IdeaMap computer program created the individual and unique combinations from a subset of the full set of 316 elements, presented the combinations to respondents, and collected the ratings of each concept on the three attributes.

Figure 1 ORIENTATION PAGE FOR THE ENGLISH STUDY

From multi-country concept testing/optimization to corporate database Figure 2 ORIENTATION PAGE FOR THE SWEDISH STUDY

13

4. Analysis: The ratings for interest for each respondent were re-coded, so that the low ratings (1-6) were recoded as 0, and the high ratings (7-9), denoting strong interest, were recoded as 100. The ratings for the other attributes were not recoded. The ratings assigned by each respondent were then analyzed by ordinary least squares, using a dummy variable model. The analysis showed the part-worth contribution of each of the 80 elements for the particular respondents. The regression model for an individual is expressed by the simple equation: Interest = k0 + k1(Element 1) + k2(Element 2) k80(Element 80) An algorithm then estimated the part-worth contributions of untested elements, so that in the end each respondent had a full set of utilities, showing the contributions of each of the 316 elements to interest. The additive constant, k0, is technically the conditional probability that the respondent will rate the concept between 7 and 9, denoting high interest, if there are no elements present. Clearly no concepts were rated absent elements. However, the additive constant is an estimated parameter emerging from the regression modeling. It provides a sense of the basic interest in the margarine products. Additive constants below 20 correspond

14

Johannes Hartmann, Howard Moskowitz, Jeff Ewald to products that are not themselves particularly interesting, but can be made interesting by the judicious selection of particular product features and sales messages. Additive constants of 60 or higher denote products that are intrinsically interesting, so the concept elements need not do much work to increase concept acceptance. The utility values k1 k80 show how each of the 80 different elements tested by a single respondent drive interest ratings. Technically, the utility values are the coefficients of the dummy variable regression equations, and emerge from the dummy variable regression analysis. For example, a utility value of 10 can be interpreted as an additional 10% of the respondents may be expected to rate the concept as highly acceptable (7-9) if the element is introduced into the concept. High utility values (ki > 10 or more) correspond to strong performing elements. Utilities of 5-10 represent good performers. Utilities 0-5 represent elements that do modestly at best. Utilities below 0 correspond to elements that do poorly. Most of the elements in the study do poorly, with utilities of 5 or lower (see figure 3). Figure 3 SMOOTHED FREQUENCY DISTRIBUTION, SHOWING THE NUMBER OF CONCEPT ELEMENTS OUT OF 316 THAT ACHIEVE SPECIFIC UTILITY VALUES

The subgroups are listed in descending order of their peak counts, so that the Total is the top-most distribution, and the UK is the bottom-most distribution.

From multi-country concept testing/optimization to corporate database 5. Subgroups: For this paper, we will look at the data by total across all countries, by concept response segment and then by country. A sense of the type of data that emerged from this multi-country study can be seen in table 2, which shows the winning elements for the total panel, three concept response segments and the four countries (see below). At the level of country there were some observations made that described the data, but there did not appear to be the unifying theme that would allow Unilever to use the data both for the current project, and also let it become the foundation of a knowledge system. Some of the observations made from looking at the same data across different countries (itself a major leap forward), were the following: a) At the country level, there are some differences in utility ranges and the nature of the winning elements. b) The constants vary slightly, with that of Sweden rather low, meaning low basic interest in the idea. For Unilever this ability to understand the basic attraction of an idea by country, and later by segment, was to become very important because it provided a new way to understand country-to-country differences in reactions to concepts. c) The score ranges vary slightly, with the Netherlands and the United Kingdom somewhat more responsive. Again this would become important because there were no similar normative data on responses to concept elements by country. d) The preferred themes vary slightly, with the Netherlands more open to a variety of themes, while the United Kingdom and Sweden are dominated by health and Germany by convenience. e) Nonetheless, at country level there are few major or consistent differences, and preferences are not strongly differentiated at country level. There are few elements with high utility values (figure 3). f) The story is not a purely national one. There must be a deeper pattern in the data.

15

16

Johannes Hartmann, Howard Moskowitz, Jeff Ewald Table 2 WINNING ELEMENTS FOR THE TOTAL PANEL, THE COUNTRIES, AND THE THREE KEY CONCEPT-RESPONSE SEGMENTS THAT EMERGED FROM THE STUDY
Segment Tot Conv Health Cooks UK Base size Additive constant Maximum scoring element Minimum scoring element Range of elements Average of elements Total Panel FA20 Is made from vegetable oils, provided by nature and essential to your body EN2 Live a longer healthier life Convenience Segment FA28 Because it's liquid, it's easy and handy to use FA25 It can do everything a solid margarine can - but it's quicker, handier and cleaner Health Segment FA9 Has virtually no cholesterol PD23 Its a well balanced mix of 30% oil and 70% premium margarine 4 3 8 4 8 -1 4 4 3 12 0 -1 2 2 2 4 833 44 4 -7 11 1 253 41 12 -10 21 1 244 44 8 -10 17 1 223 42 12 -8 20 1 213 43 8 -5 13 1 Country GR 201 46 5 -8 13 0 SW 204 36 6 -10 16 1 NT 215 50 7 -14 21 1

4 4

4 3

6 6

6 7

5 6

0 0

6 3

5 7

10

-1

-1

-2

-1

From multi-country concept testing/optimization to corporate database TABLE 2, CONTINUED


Segment Tot Conv Health Cooks UK Cooking Interested Segment PK19 Comes in a bottle similar to olive oil VS5 Glass bottle UK Respondents FA9 Has virtually no cholesterol FA5 Contains calcium for strengthening bones and teeth German Respondents PB23 The packaging stays clean - every time. So does your fridge. And so do you. PD16 It's a well balanced mix of 70% oil and 30% butter German Respondents FA20 Is made from vegetable oils, provided by nature and essential to your body TG12 For the health-conscious cook Dutch Respondents FB28 It doesn't spit and spatter because of its special formulation EN2 Live a longer healthier life 4 3 8 4 8 -1 4 4 1 1 -2 -2 -2 -1 12 11 3 2 0 3 0 -1 1 1 Country GR SW NT

17

1 3

9 2

-1 6

-2 3

2 2

5 4

1 1

-2 3

4 4

4 3

6 6

6 4

5 5

0 1

6 5

5 4

3 4

6 3

4 6

2 7

4 6

0 0

1 3

7 7

(GR = Germany, SW = Sweden, NT = Netherlands; Conv= convenience oriented)

18

Johannes Hartmann, Howard Moskowitz, Jeff Ewald 6. Concept-response segmentation: Conjoint analysis models at the individual respondent level. This individual modeling allows segmentation of respondents, independent of country (Green and Krieger, 1991). The data from the four countries were analyzed, independent of country, to generate three trans-national segments. The systematic generation of such trans-national segments was key to the objectives of this study, and to the objective to database multi-national concept information. We had set up the design to allow us to do this segmentation, following the approach previously suggested for other products such as coffee (Moskowitz, 1996). The stronger performance of some elements emerging out of the transnational segment, based upon patterns of utility values independent of country, appears clearly in figure 4. Figure 4 SMOOTHED FREQUENCY DISTRIBUTION FOR TOTAL PANEL AND CONCEPT-RESPONSE SUBGROUPS, SHOWING THE NUMBER OF CONCEPT ELEMENTS OUT OF 316 THAT ACHIEVE SPECIFIC UTILITY VALUES

The subgroups are listed in descending order of their peak counts. The segments show greater numbers of high scoring elements than does the total panel.

The segments transcend country, and appear in different proportions in each country. The data from the total panel shows the weak performance of the elements, but the segments have quite a number of stronger performing elements. International data by itself can become confusing if the analysis is kept to a description of what wins and what loses. It was critical for Unilever to be able to identify these winning segments, their

From multi-country concept testing/optimization to corporate database themes, and the organizing principles giving rise to the segments, because these results suggest larger-scale organizing themes that increased fundamental knowledge of the margarine-oil business beyond the studies of individual countries and individual products. The insights for Unilever afforded by the concept-response segmentation took on additional value because of the large number of concept elements (316) that repeated across the different countries. Some of the insights appear in table 3. Table 3 INSIGHTS BASED ON WINNING THEMES FROM THE CONCEPT-RESPONSE SEGMENTATION
1. Each of these segments represents a viable but quite distinct target. 2. Their presence in each country, and across all user groups explains the relatively neutral data - the demographic scores are averaging out three quite diverse attitudinal segments. 3. It is still not possible to reach everybody. Convenience Seekers and Cooks tend to be diametrically opposed in terms of preferences. 4. By targeting segments instead of demographics, it is finally possible to create a product that excites real interest Convenience Segment respond to six themes Its cleaner Its quick and easy It has some sort of packaging design which makes it quicker It wont spit, spatter or burn, causing mess and skin burns Its a liquid or spray, not a solid Its a modern product for a busy lifestyle Health Seekers respond to five themes Contributes to a healthy lifestyle Includes specific health benefits in terms of calories, etc. Has some kind of oil mix Tastes good Versatility

19

20

Johannes Hartmann, Howard Moskowitz, Jeff Ewald TABLE 3, CONTINUED


Cooks respond to five themes It comes in a bottle like olive oil Its made from vegetable oil Its all about good home-cooking Its the best Can be used for baking

7. Databasing: The data permitted product management and R&D at Unilever to understand consumer demands across four different European countries. The database thus created was reported in and of itself. However, beyond listing the database of elements by key panel subgroups, Unilever worked on creating a concept optimizer and simulator for this data. The optimizer sorted through the different concept elements, and created new combinations that had specific features, and performed well in a number of countries. The concept optimizer furthermore took into account several dozen pair-wise combinations of elements that could not appear together (constraints). The optimizer accompanied the database as an integral part of the knowledge base, and was used separately in each country for decision-making, as well as by an international team representing the different countries. All individuals were using the same data, and could coordinate their development and marketing strategies, using this common database.

EIGHT LEARNINGS AND OBSERVATIONS EMERGING FROM EXERCISE TO CREATE CONCEPTS FROM THE BOTTOM UP
The eight key learnings about the role of market research and the concept development process are the following: 1. The Leadership: A demonstration of the power of bottom-up development. Applying a process and new methodology, Unilever market research stimulated break-through thinking with a very diverse team of participants. The success of the project was an important concrete demonstration for Unilever leadership about the power of bottom-up development, at the early stage of development.

From multi-country concept testing/optimization to corporate database 2. The Corporation: A chance to change the corporate culture involved in concept development. It is worth noting here, parenthetically, that by using a bottom-up approach for the first time in a European context, Unilever was to discover the strength of that approach. Many projects are run topdown, but the record is beginning to show that that they are not particularly successful. 3. The Corporation: An opportunity to systematize the concept development processes that lead to databases. The approach presented here systematizes concept development, and improves the chances of success because it combines a broad scope of raw materials, a way to understand consumer reactions, and a segmentation approach that uncovers new, and hitherto unexpected segments that transcend different countries and cultures. 4. The Corporation: Transcend differences among professionals in the corporation responsible for developing new ideas and advancing into new areas. Internally we learned that we had to define concepts close to culture and had to combine conservatism with expansiveness. It became clear that initially anticipated differences across countries were not that overwhelming that they hindered the process. In the subsequent development processes the coordination between country representatives became much easier. 5. The Corporation: A platform for open negotiation among participants. The process offered a platform to negotiate without losing ones own position. People dont need to defend their inputs. The methodology defused the personality ownership and position issues during inputgeneration. 6. The Research Professional: Empowered market research professionals. Up to now a great deal of concept development has relied on experts, on intuition, or on serendipity. Market researchers have been relegated to the job of evaluating the new concepts. When called to be members of the creative team they have provided input, but have not been as vocal as the advertising agency, marketing, or the external creative/marketing consultant. The use of research here provides the professional with a way to be heard, and with the ability to contribute winning ideas to the concept development process. 7. The Marketing Research Field: Advanced the market research discipline. The use of segmentation across cultures provided a new dimension for consumer researchers who continually seek better ways to understand consumer behavior.

21

22

Johannes Hartmann, Howard Moskowitz, Jeff Ewald 8. The Marketing Research Field: Consumer research emerged into an effective regional leadership tool. Consumer Research defined as a process has the clear chance to define aligning principles in an environment where so-called cultural differences are used to defend local perspectives.

IMPLICATIONS FOR OTHER ORGANIZATIONS


It is well recognized that there are many barriers to effective globalization, some external in nature. However, many barriers are internally erected. Examples include ineffective organizational design, legacy paradigms and over-reliance on ad hoc, disjointed information, and the like. For many organizations, globalization is still a voluntary step for the local operating manager. The notion of relying on specific, ethnocentric learning and experience creates a strong bias against full implementation of a global strategy backed by the well-known byword not invented here (Kapferer, 1998). Even when controlled by highly centralized management, an effective global implementation requires the buy-in and cooperation of regional and national managers to coordinate and control the myriad marketing elements across regional and national boundaries. The systematic, empirical process described in this paper not only provides the global manager with the information necessary for improved decision making, it also enhances buy-in and cooperation across geographic boundaries by instituting a collaborative procedure early in the idea process. With objective data providing the roadmap for which elements are best controlled centrally and which need to respond to local conditions, we think becomes we know. The difference in results can be dramatic. Corporate energies can be channeled against the right things in the right places. Global ideas are allowed to span their potential the promised economies of scale can be realized. Important local differences are recognized and funded appropriately. Petty arguments are discontinued. From a common understanding of what needs to be done comes an uncommon commitment to getting the job done.

From multi-country concept testing/optimization to corporate database REFERENCES


Bacon Jr., F.R. and Butler Jr., T.W. (1998). Achieving Planned Innovation. New York: The Free Press. Bateson, J.E.G., Reibstein, D.J., and Boulding, W. (1987). Conjoint analysis reliability and validity: A framework for future research. Review of Marketing, 451-481. Buzzel, R.D. and Quelch, J.A. (1988). Multinational Marketing Management. New York: Addison Wesley. Cooper, R.G. (1999). The Invisible Success Factors In Product Innovation. Journal of Product Innovation Management, 16, 116-133. Fuller, G.W. (1994). New Food Product Development: From Concept to Marketplace. CRC Press, Boca-Raton, FL. Pp. 69-124. Green, P.E. and Krieger, A.M. (1991). Segmenting markets with conjoint analysis. Journal of Marketing, 55, 20-31. Haeckel S. (1999). Adaptive Enterprise, Creating and Leading Sense and Respond Organizations. Harvard Business School Press, Boston. Hoban, T.J. (1998). Improving the success of new product development. Food Technology, 52, 46-49. Kapferer, Jean-Noel. (1998). Strategic Brand Management. London: Kogan Page Limited. Mitchell, A. (1983). The Nine American Lifestyles. New York, Macmillan. Moskowitz, H. (1996). Segmenting consumers world-wide: An application of multiple media conjoint methods. Proceedings of the 49th ESOMAR Congress, Istanbul, 535-552. Moskowitz, H.R. and Martin, D.G. (1993). How computer aided design and presentation of concepts speeds up the product development process. Proceedings of the 46th ESOMAR Congress, Copenhagen, Denmark, 405 - 419. Rosenbaum, H.F. (1987), Redesigning Product Lines with Conjoint Analysis: How Sunbeam Does It. Journal of Product Innovation Management, 4, 120-137. Trout, J. and Rivkin, S. (2000). Differentiate or Die: Survival in Our Era of Killer Competition. New York: John Wiley & Sons, Inc. Veganti, R. (1997). Leveraging on systemic learning to manage the early phases of product innovation projects. R&D Management. 27(4):377-392. Wells, W.D. (1975). Psychographics. A critical review. Journal of Marketing Research, 12, 196-213. Wittink, D.R. and Cattin, P. (1989). Commercial use of conjoint analysis: An update. Journal of Marketing, 53, 91-96. Zaltman, G., Duncan, R., and Holbeck, J. (1973). Innovations and Organizations. New York, John Wiley & Sons.

23

24

Johannes Hartmann, Howard Moskowitz, Jeff Ewald THE AUTHORS


Johannes Hartmann is Vice President, Consumer Understanding, Unilever Bestfoods, Singapore. Howard R. Moskowitz is President, Moskowitz Jacobs Inc., United States. Jeff Ewald is President, Optimization Group, Inc., United States.

Anda mungkin juga menyukai