This note is a shortened and simplified version of an article in the *Journal of the Operational Research Society* April 2013, Volume 64, Issue 4, pp 597–610, written Shipley, M., Johnson, M., Pointer, L. et al. For full details, including contact details and references, please see the full article.

### Abstract

A Fuzzy Attractiveness of Market Entry (FAME) model is developed to address the decision-making problem of product introduction into alternative markets. FAME is a market entry selection model that is specifically designed to handle situations when information is limited and/or ambiguous, and a high level of uncertainty exists. As such, the FAME model is an easy to implement tool that supports a reasoning approach to market selection decisions. The model uses expert opinions regarding four factors: (1) fit of the firm's marketing mix in each market; (2) the fit of its key competitor's marketing mix in each market; (3) environmental conditions in each market; and (4) the strategic importance of each market to the firm. Application of the model algorithm is conducted for a small, Bulgarian winery's market selection decision. Ease of use is relevant for small to mid-size companies since a spreadsheet is sufficient to complete the algorithmic calculations.

### Introduction

The selection of the appropriate market in which to sell a firm's products can be a difficult multi-criteria decision-making problem. When considering the attractiveness of any market, factors such as the firm's marketing mix (product, price, distribution and promotion) as contrasted to competitors’ marketing mix, the environmental variables, market share, potential profitability, and even the strategic importance to the firm's management of a particular market impact the decision. Thus, the assessment of potential markets is very knowledge-based, and yet, generally stochastically driven where data are frequently unavailable or of such quality that uncertainty and ambiguity are inherently present. Extant methods that utilize classical logic or statistics are often inadequate for effectively dealing with situations where information is limited. In fact, for over 20 years, behavioural and expected utility theorists have argued that probability theory and other traditional quantitative techniques are not equipped to consider the uncertainty that exists in most personal judgments.

Yet, small firms and new ventures frequently rely on expert judgments in multi-criteria marketing decision-making situations since data collection and statistical analyses modelling applications are often cost prohibitive. In addition, Sarabia reported that about one quarter of marketing professionals did not formally evaluate segments. This could be because, marketing professionals did not have access to the necessary data and/or that the managers viewed the data or the findings from the data to be uncertain or ambiguous with respect to predicting the success of the next venture. Sarabia also found that expert opinion was seldom used in the evaluation and market selection process. Fuzzy set-based modelling offers an advantage to all companies for dealing with uncertainty and ambiguity of data, but is particularly relevant to small to mid-size organizations (SMOs), such as the Bulgarian winery which forms the basis of the illustrative example of the FAME model's algorithmic process. The Fuzzy Attractiveness of Market Entry (FAME) model developed in this paper is knowledge-based rather than data driven. As such, the model provides a multi-criteria decision-making aid (MCDA) that allows the manager to explicitly and logically consider the certainty of expert opinions and obtain a ranking of the attractiveness of different markets for a particular product.

### Theoretical background

Research shows that managers approach marketing decisions in one of four ways: optimizing, reasoning, analogizing and creating. Of these approaches, optimizing, reasoning and analogizing are best suited for making market selection decisions.

Models that support the optimization approach are quantitative in nature and include portfolio and regression models. With the portfolio model the firm gathers data about each market under consideration. Typical data include growth potential, sales revenue, ROI, and competitor intensity. These data are acquired from secondary sources, surveys or managerial estimates and are used to build attribute profiles for each market. After the market profiles are created, mathematical programs are written to: (1) find the market that best matches a desired market portfolio attribute profile; (2) identify the optimal market; or (3) identify the market that is best with respect to one of the attributes in the portfolio. The results from the portfolio model show each market's attributes on quantifiable measures. Managers can examine each market option on its ranking as a composite of desirable attributes as well as its ranking on each desirable attribute.

Regression-type models including time series analysis have been useful for selecting retail markets based on historical sales data. Secondary data for seasonally adjusted monthly sales are used to approximate a realized rate of return for each market. By plotting the mean (average) realized rate of return against the standard deviation (risk), the manager has a visual representation of the risk *versus* the rate of return for each market. Regression models can also be expanded and enriched by including additional variables such as market potential, competitive resistance and interactions between markets.

Decision making by analogy is supported by the case-based reasoning model for market selection. These models utilize a firm's internal databases. Key information on prior company projects are retrieved based on parameters identified by the manager. These parameters could include country risk, project size, type of client, attitude of government, and cultural similarity. In creating the database of cases, managers use their judgment to assign a numeric value on each parameter within each individual case. The result is a case library which can be searched using a number of key words. When faced with a market entry selection decision, the manager looks for analogous situations from the company's case library. The manager retrieves from the database the cases with the greatest similarities. Similarity can be defined as feature counting, weighted feature counting or inferred feature computation. Using these similar cases, the predicted profitability and the level of competitiveness for potential projects are derived.

The third approach to market selection decision making is reasoning. Index models were designed to evaluate the attractiveness of different countries and provide an effective way to summarize and consolidate quantitative information. The process involves collecting country data on several dimensions including market size, market growth rate, market intensity, market consumption capacity, commercial infrastructure, economic freedom and market receptivity from secondary data sources. Multiple measures on each dimension are incorporated into an index score for that dimension for each country. Managers can look at rankings on each dimension or multiple dimensions. This information provides insights for the manager in understanding each market's relative attractiveness at a macro level which can be added to micro level information that the manager may possess.

Although market researchers have generally used uncertain and ambiguous data as input into decision models based on probability theory, Ellsberg and others have demonstrated that the fundamental rules governing probability are rendered invalid when the likelihood of events cannot be expressed in binary terms. In view of the fact that the traditional two-valued logic of probability judgments is inadequate to handle the combined presence of uncertainty and ambiguity, the calculus of fuzzy sets has been used in marketing models to allow effective and systematic handling of imprecise and imperfect information.

One of the characteristics of qualitative marketing research is a degree of fuzziness because it involves open-ended descriptions of complex phenomena. Classification of items into categories after data collection traditionally required judgments that place an item with 100% membership into the category. In the early 2000s, fuzzy logic gained acceptability as a mathematical model for allowing items to have membership to varying degrees in multiple categories. A fuzzy latent class model (FLCM) was developed that extended the sDillon and Mulani's seminal model, allowing crisp items to be attached to one category but introducing fuzzy membership functions for other items.

Finally, most recent fuzzy set-based marketing models have returned to focus upon the uncertainty in human judgments. Specifically, a fuzzy analytical network process was developed to accommodate the interactions, dependencies and feedback and to counter the vagueness and uncertainty in consumers’ judgments. Importantly, researchers have begun to employ more comprehensive intelligent systems for international marketing planning. The system was further expanded by linking web-enabled hybrid intelligent software systems which combined web-based knowledge automation, fuzzy rules and online databases to aid in a variety of international marketing decisions from market entry, entry mode, marketing and competitive.

### Development of the FAME model

The FAME model utilizes human judgments for data input values and fuzzy logic-based computations to produce a score by which a rule-based decision can be executed and specific products to a set of potential markets can be rank ordered. The basics of fuzzy set theory necessary for understanding of the FAME model and the FAME algorithm are presented below.

### Fundamentals of fuzzy set theory

The major premise behind a fuzzy set is the generalization of the concept of the characteristic function of a set. To illustrate, let *X* denote a universal set and denote by *A* some non-empty subset of *X*. Then the characteristic function of set *A* is

With an ordinary (ie, crisp) set, the characteristic function has values that are either 0 or 1. That is, any element of a universal set *X* clearly is in set *A* or is not in set *A*. In fuzzy set theory, it is no longer certain whether an element is or is not in a certain set. Thus, any element can belong to some degree to any set and hence gives rise to the concept of a ‘perfect fit’. Suppose, for example, that *A* denotes the set of products that perfectly meet the wants and needs of a particular group of buyers. The meaning of ‘perfect fit’ is itself subject to interpretation. Therefore, whether a particular firm's product is included in set *A* is necessarily a matter of opinion.

To deal with such uncertainty, the generalization of the concept of a characteristic function becomes a fuzzy membership function. Associated with every *x**∈**X*, there is a ‘membership value’ *m*_{A}(*x*), which indicates the degree to which it is believed that element *x* belongs to fuzzy set *A*. This membership function can take any value on the closed interval [0, 1]. Membership equal to zero indicates that element *x* is definitely not in *A*, while membership equal to one indicates that *x* definitely is in *A*. Values between zero and one represent varying degrees of uncertainty as to whether or not *x* is a member of set *A*. In mathematical terms, *m*_{A}: *X* → [0, 1] where *X* is called the universe of discourse of *m*_{A}.

The usual set operations of complementation, union and intersection are extended to fuzzy sets through the membership function. Specifically, the membership function of the complement of fuzzy set *A*, designated by ¬*A*, is *m*_{¬A}(*x*)=1−*m*_{A}(*x*); the membership function of the union of fuzzy set *A* with fuzzy set *B*, designated by *A* ∨ *B* is *m*_{A}_{∨}_{B}(*x*)=Max{*m*_{A}(*x*), *m*_{B}(*x*)}, and the membership function of the intersection of fuzzy set *A* with fuzzy set *B*, designated by *A* ∧ *B* is *m*_{A}_{∧}_{B}(*x*)=Mix{*m*_{A}(*x*), *m*_{B}(*x*)}.

When fuzzy sets are defined on a finite support (ie, finite universal set), a commonly used notation is *A*=*a*_{1}/*x*_{1}+*a*_{2}/*x*_{2} +⋯+*a*_{n}/*x*_{n}. This statement is read, element *x*_{1} is in fuzzy set *A* with membership *a*_{1}, element *x*_{2} is in fuzzy set *A* with membership *a*_{2}, etc. Fuzzy sets can, of course, have non-finite support, but for this model, finite support will suffice.

### The theoretical FAME model

The FAME model evaluates market attractiveness in terms of three key variables which are derived from a limited number of industry experts. These variables are (1) the fit between the firm's products and a market, (2) the market environment including key competitors, and (3) the strategic value of the market to the firm. These three variables are consistent with the four stages of analysis recommended for market evaluation and selection. The strategic value of the market to the firm incorporates profitability and sales volume. Finally, the three key variables in the FAME model build on the synergy effects identified in Levy and Yoon's decision framework for global market entry (1995) by explicitly considering key marketing decisions in the area of promotion, pricing and distribution. This intentional evaluation of each aspect of the marketing mix provides information that can be the basis for tactical decisions if a market is selected. The FAME model uses expert beliefs and the strengths of those beliefs on these three key factors. The FAME model, therefore, specifically recognizes that a firm's ability to compete within a market is determined by more than a competitive product. It must have the distribution, pricing and promotion needed to support the product within a particular market, the market environment must be favourable and the market must have strategic value to the firm.

Given these considerations, the FAME index is represented by the following function:

where *A*_{t} =market attractiveness at time *t*; *X*_{t} =the firm's marketing mix variables (product, price, distribution, and promotion) in terms of fit with the preferences of buyers in the market at time *t*; *C*_{t} =competitor's marketing mix variables (product, price, distribution, and promotion) in terms of fit with the preferences of buyers in the market at time *t* ; *E*_{t} =environmental variables at time *t*; *M*_{t –}_{1} =market share at time *t−*1 (if new market, then projected market share for time *t*); *P*_{t –}_{1}=profitability at time *t−*1 (if new market, then projected profitability for time *t*); *S*_{t+}_{1}=strategic importance of market at time *t+*1; *ɛ*_{t}=stochastic error.

Within *A*_{t}, each variable is specified as a fuzzy set based on experts’ assessments of customer perceptions with additional accumulated data such that they can evaluate the fit of each component that comprises that variable. Assimilating customer preferences with other observable data, each knowledgeable expert provides his/her belief in the ability for the company's product to fit the target market. Unlike crisp sets, the belief in a linguistic state does not have to sum to one. However, should a choice be made to allocate 100% between the states of the parameters, it can be done without loss of validity of the model.

Given a representative number of fuzzy sets defined by the experts’ perception of *X*_{t}, the beliefs across all input sets are combined. Various procedures can be undertaken to accomplish this merger, from averaging the beliefs of all experts to accepting the minimum belief or the maximum belief of any assessor. A weighted combination of minimum and maximum beliefs can be determined as *α* (*X*_{tmax})+(1−*α*) (*X*_{tmin}) where the value of *α* reflects the optimistic or pessimistic attitude of the decision makers (0⩽*α*⩽1). Obviously, if they are neither optimistic nor pessimistic about the product, *α*=0.5 would be the average of the minimum and maximum beliefs.

Next, *C*_{t} is considered as the Best Market Fit for the Firm's Marketing Mix (BMFMM) which is the degree of fit of the decision makers’ perceptions of their competitor(s)'s position with respect to each state of product, price, distribution, and promotion at the same time, *t*, for which *X*_{tweighted} was determined. The degree to which the *X*_{tweighted} fits *C*_{t} (defined as the BMFMM) for Market *i* would be determined by cardinality rules and fuzzy logic operational methods.

The key parameters that can impact market attractiveness, *A*_{t}, are investigated where *E*_{t} would be based on the infrastructure of the market with respect to both external and internal environmental factors, *M*_{t–}_{1} or *M*_{t} (for a new market) would be the percentage of market share captured previously or anticipated to be captured with the product introduction, *P*_{t–}_{1} or *P*_{t} (for a new market) would be the ratio or percentage of profitability, and *S*_{t+}_{1} would be a measure of strategic importance. While *M*_{t–}_{1} or *M*_{t} and *P*_{t–}_{1} or *P*_{t} would be crisp values as percentages, *E*_{t} and *S*_{t+}_{1} would be judgments made by experts such as market researchers, functional area managers, CEOs, and/or boards of directors which can be converted to fuzzy sets.

The concept of Centre of Gravity can be used to defuzzify the sets. Assuming that the scale for Favourable and Highly Important is 8–10 (on a scale of 1–10), that Moderate refers to 5 to 7, and that Unfavourable or Unimportant equates to 2 to 4, the midpoint of each range can be used to calculate the Centre of Gravity (or average). Thus, a relatively strong belief in the environmental factors and infrastructures may be diminished by the lack of strong strategic initiatives for the market under consideration, thereby decreasing its market attractiveness.

### FAME algorithm and decision rule

Consolidating the information above yields the following algorithm that will be referenced in the application of the Bulgarian winery's market selection decision.

*Step* 1:

Determine *X*_{t} where each expert indicates his/her belief in the firm's product, price, distribution, and promotion in terms of fit with the preferences of buyers in the market segment at time *t*

*Step* 2:

Combine results across all experts’ opinions for the Minimum and Maximum fit then calculate *X*_{weighted} as *α* (*X*_{tmax})+(1−*α*) (*X*_{tmin}) where the value of *α* reflects the optimistic or pessimistic attitude of the decision maker (*α*=0.5 is average of highest and lowest beliefs). Note that if a statistically sufficient number of experts are available, a simple average over all experts’ beliefs can be used for *X*_{weighted}_{.}

*Step* 3:

Determine *C*_{t} where each expert indicates his/her belief in the major competitor's product, price, distribution, and promotion in terms of fit with the preferences of buyers in the market segment at time *t*

Step 4:

Average across all expert opinions applying a weighted value for *C*_{t} as in Step 2 above or a simple average

*Step* 5:

Determine the belief in the firm's product in comparison to *C*_{t} considered as the Best Market Fit with respect to Market Mix (BMFMM) using fuzzy set intersection definitions and calculating

where *X*_{t} and *C*_{t} are the weighted averages calculated from Steps 2 and 4.

*Step* 6:

Determine *E*_{t} from expert opinions of the infrastructure of the market with respect to both external and internal environmental factors, such as economy, political climate and social conditions. For each linguistic variable, such as excellent, good, fair or poor, the maximum and minimum beliefs weight the average based on some value of *α* and assuming no more importance is attached to either environmental factor, a simple average is taken across all infrastructure variables. Centre of Gravity, weighted by scores or ranges of scores attributed to each linguistic variable, is used to derive *E*_{t} for the market being evaluated.

*Step* 7:

Determine *S*_{t+}_{1} from expert opinions based on the perceived strategic importance of the market with respect to both potential sales growth and profit margin. As for Step 6, experts’ beliefs for each qualitative variable are weighted by *α* and then averaged for the linguistic variables. Centre of Gravity is used to derive *S*_{t+}_{1} for the market being evaluated.

*Step* 8:

Use any statistical data available for other variables and calculate as multiplicative function:

If no data are available for *M*_{t−}_{1,} and *P*_{t−}_{1}, these variables are assumed to be 1.0.

Note that the error adjustment *ɛ*_{t} can be proposed by the firm's management.

After calculation of *A*_{t}_{,} a rule-based decision-making procedure could be based on management's determination of the knowledge contribution inherent in the scores and the minimum level of overall attractiveness for market entry. An example of rule-based system might be:

**R**_{1}: *If market selection index is greater than 0.20, then strongly pursue the market, changing price, promotion, and production to do so*.

**R**_{2}: *If the market selection index is greater than 0.15 but less than 0.20, then moderately pursue the market considering changes in price and promotion.*

**R**_{3}: *If the market selection index is less than 0.15 then give only slight consideration to the market, perhaps considering production changes and market as applicable to over production supply.*

### Expert opinion survey

A survey was designed for assessment by experts from Tsenov Academy and a local Svishtov, Bulgaria winery to which the experts served as grape suppliers and process consultants or managers. For this illustration of the FAME model, three experts responded to all statements. Specifically, two of the experts were vineyard owners and academicians with extensive research experience in the Bulgarian wine industry. The third expert was a wine producer who had several years experience selling in both markets under consideration. These were identified as TD, NK, and SA.

The survey was titled: Expert Opinion Regarding the Wine Market. The questionnaire was divided into two parts. Part 1 covered opinions regarding wine in a local market while Part 2 inquired about the national market. Each part contained six sections. Sections 1–4 contained four opinion statements about four different wines, a cabernet sauvignon and a chardonnay from the Svishtov winery and the same from the dominant competitor in each market, local then national. Each opinion statement contained three different linguistic variables. A block of questions concerned: (1) buyer preferences for product quality and taste (high, medium, low); (2) buyer's evaluation of price (too high, about right, a bargain); (3) advertising and promotion in attracting buyers (very effective, effective, not effective); and (4) buyer access (ie distribution) to the wine through retailers (very good, good, poor).

The respondent was asked to rate his/her confidence on a scale of 0 (no confidence) to 5 (absolute confidence) that the wine buyer's preference would fit within the market according to the statements proposed. For the sake of ease of explanation to non-native English-speaking experts, the term ‘confidence’ was used to represent the degree of belief by the expert that the statement was to some degree true and, thus, the opinion that to varying degrees a wine fit the needs of a buyer in a particular market. For example, the respondent was asked to indicate his/her degree of confidence in the following three statements regarding the advertising and promotion of the Svishtov winery's product and then a comparable competitor's product by brand name.

- The advertising and promotion of this wine in attracting buyers is
*very effective.*
- The advertising and promotion … is
*effective.*
- The advertising and promotion … is
*not effective.*

A respondent might choose a confidence level of 3 on the first statement, a 5 on the second statement and a 1 on the third statement. He/she might even check the line on a box indicating 0.5, 1.5, etc. Thus, the degree of uncertainty in the respondent's agreement with each statement was captured and was not mandated to be perfect certainty and sum to any specific total (where sum of 5 would be belief of 1.0). In order to use the typical fuzzy set membership, the scores were, thus, normalized such that for the above example, membership would be: 0.6/Promotion Fit Very Effective+1.0/Promotion Fit Effective+0.2/Promotion Fit Not Effective.

All belief scores for fit of the product to the market based on buyer preferences were calculated as detailed, including those for the final two sections of the survey. Section 5 contained statements regarding the market conditions for wine consumption including the (excellent, good, poor) general economy, infrastructure, political climate and cultural structures. Section 6 contained two statements on the strategic importance of the market to the firm in terms of (above average, average, below average) sales growth and profit potential.

### Analysis

The best marketing fit scores for the firm's chardonnay are also very similar. The local market eases out the national market with a 0.83 compared with a 0.81. The experts’ beliefs regarding the firm's chardonnay are consistent across the local and national markets on pricing, distribution and the product itself. The firm's chardonnay has better distribution than the benchmark brand in both the local and national market. Its promotional tactics are perceived as weaker in both markets and its pricing competitive in both markets. In the experts’ view the key difference between the local and national markets concerns the product itself. The firm's chardonnay does not compete as well in taste in the local market against the benchmark brand as it does in the national market. In this case, the higher scores in distribution and promotion in the local market are compensating for slightly lower scores in the taste of the chardonnay. The result is that the local market is considered a better marketing fit for the firm's chardonnay.

### Market environment

Four aspects of the market environment are considered. These are economic conditions, infrastructure, political climate and social conditions. The experts were asked to indicate the strength of their belief that each of these elements was excellent, good or poor

### Strategic importance

The strategic importance variable contained two elements: sales growth and profit. For each of these elements, the experts indicated the strength of their belief that the element was above average, average or below average.

### Belief in market attractiveness

Step 8 of the Algorithm is now executed to calculate an overall market attractiveness score. Since this is a new venture for the firm, market share or profitability from the previous year, as statistical measures were not available therefore the attractiveness score is based on the best marketing fit for the marketing mix, the market environment and the strategic importance of the market. Table 3 summarizes the data used to calculate the attractiveness scores.

### Results

For the study of the Bulgarian winery the following is known. The Bulgarian winery chose to expand into the national market. The tactics implemented included increased advertising and promotion of the wines through various media including print and digital. The firm also expanded its distribution network to include key supermarkets that operated throughout the country. The quality of the wine was generally perceived as good and has been recognized as some of the best in national competitions. The winery was able to slightly increase the price of the wine to offset the increased costs associated with the expansion. During the period of expansion into the national market, the firm also began expansion into international markets. The cabernet sauvignon wine sales are distributed currently as follows: 10% locally; 40% nationally; and 50% internationally. The chardonnay wine sales are equally split (40% each) between the local and national market with the remaining 20% being sold internationally.

The belief in market attractiveness values from the application of the FAME model indicated that the national market was more attractive than the local market. The accuracy of the experts’ opinions about the national infrastructure as well as the economic, political, and social conditions was confirmed by the results which followed. The shift in the distribution of wine sales for both the chardonnay and the cabernet sauvignon support the belief in the attractiveness of the national market. It is also noteworthy that the FAME model had alerted the winery to the challenges of entering this market, specifically the need for improved distribution and promotion. Both of these weaknesses were specifically addressed as the firm invested in expanding distribution into key supermarkets as well as investing in promotion.

### Conclusions

Traditional models such as factor-weighting, AHP, and regression have limitations in determination of a better screening approach for decision making. In particular, scoring models do not take into account uncertainty with respect to mapping one's judgment to a number. Also, subjective judgment and preferences of the evaluators can create bias in scoring. While AHP can reconcile differences in managerial judgment and perceptions, the scale has the same limitations as the scoring model. There is also danger of selecting the best of a bad lot (Lin and Chen, 2004). Regression models attempt to eliminate bias and uncertainty through statistical analysis which as stated in the Introduction has been unsatisfactory for addressing these qualities of human judgment and/or available data.

Fuzzy AHP has seen increasing usage in all areas of multi-criteria decision-making research, but it has generally moved beyond theoretical development to become more application driven in its orientation. By contrast, the theoretical development of the FAME model is a unique marketing index model. Its capabilities are illustrated in the study of the Bulgarian winery by which expert opinions can be captured and analysed for a decision without having extensive data available. At each step in the process managers gather information that is logically relevant to the decision yet contains a recognized uncertainty factor and ambiguity of data. The FAME model directly considers uncertainty and ambiguity. Since the experts state not only an opinion but also the strength of that opinion, the firm can intentionally factor uncertainty into their reasoning. In many ways, as has been determined by researchers comparing traditional *versus* fuzzy-set-based approaches, the FAME results based on the combinatorial approach using fuzzy set theory provide a more logical analysis of relationships as opposed to just statistical relationships usually used for market selection decisions.

Although, fuzzy set-based modelling offers an advantage to all companies for dealing with uncertainty and ambiguity of data, it is particularly relevant to SMOs. While both small and large firms can implement the FAME model with little difficulty, the ability to conduct all the analysis in a spreadsheet is a major benefit to small firms. The collection of expert opinions can be obtained relatively quickly. A simple survey instrument similar to the one used for the Bulgarian winery can be created and distributed with minimum effort, time and cost. The key to ensuring quality is in the identification of experts. Once the experts are identified, the data can be collected and stored in a spreadsheet. In that format, the manager can conduct a series of scenarios in which they can give different weights for different experts or adjust for optimistic or pessimistic viewpoints.

Content-wise, the FAME model utilizes the three variables that are consistent with the four stages of analysis recommended by Sarabia for market evaluation and selection. Beginning with an assessment of how well the firm's product, pricing, promotion and distribution meet buyer preferences, the manager begins to identify arguments supporting or negating the attractiveness of different markets. The model directs the manager to consider how its fit with customer preferences compares to its key competitor's product, pricing, promotion and distribution. These two steps in the model provide insights into the strengths and weaknesses of the firm's marketing mix in each potential market; adjusting *α* to indicate an optimistic to pessimistic decision maker viewpoint.

The FAME model also considers macro-level factors, such as infrastructure, economic, social and political conditions. Understanding each market's relative attractiveness at a macro level is added to micro level information that the manager and other experts possess. Thus, the model allows multiple measures on each dimension to be incorporated into a rule-based index score so that managers can look at rankings on each dimension or multiple dimensions. The decision to then pursue market entry becomes knowledge-based. Finally, the FAME model directly incorporates expert opinion on the strategic importance of the markets in terms of sales and profits The FAME model takes managers through the logical steps of evaluating markets in terms of: (1) the best marketing fit of the firm's products, (2) the characteristics of the market environment, and (3) the strategic importance of the market to the firm's overall viability.

Overall, the FAME model is a unique multi-criteria decision-making aid suited for helping managers reason through a decision where the information about a market or set of markets is difficult to obtain. If language or cultural barriers make data collection difficult, then the manager can still use the FAME model so long as there are available experts. A major strength of the FAME model is its usefulness in evaluating market attractiveness without regard to size of company, or the quality or availability of secondary data.

### Limitations

The model developed for assessing market attractiveness for product introduction is rigorous and unique to the literature. However, the application to a Bulgarian winery cannot be defined as a case study in the strictest sense of case research. The present research was part of a larger research project funded by the United States National Science Foundation grant (Award # INT-0207141). The primary focus of that research was the creation of tools and methods for facilitating Bulgaria's transition from a planned to a market-driven economy. The wine industry was identified as important to Bulgaria's entry into the European Union and therefore, some time was spent on understanding the current state of that industry. It should be noted that the time spent in the country was limited and the distance between participants necessitated work outside of the country. Thus, although the experts were strongly associated with the winery's operations and included a manager within the company, representation was limited for our survey. In addition, our access to management at the winery was strictly limited and made more difficult when the management of the winery changed during the implementation period. Although we do not have the detailed data needed for a case study, we were able to verify that the recommendations from the FAME model were implemented with good results.

### Future research

The FAME model could be implemented again in an actual case study setting. This would require the identification of a company considering expansion into a new market that would be willing to accept and implement the recommendations of the model. Such a case study would provide additional validation of the model. Another possibility for future research is to modify the current index model into a fuzzy AHP model and compare the two approaches. A comparison of market attractiveness rankings from the alternative fuzzy approaches could answer questions about the relative benefits of each approach.