essay

Managing Global Brand Investments at DHL Author(s): Marc Fischer, Wolfgang Giehl and Tjark Freundt Source: Interfaces, Vol. 41, No. 1, 2010 Franz Edelman Award for Achievement in Operations Research and the Management Sciences (January-February 2011), pp. 35-50 Published by: INFORMS Stable URL: https://www.jstor.org/stable/23016178 Accessed: 28-04-2019 18:01 UTC

REFERENCES Linked references are available on JSTOR for this article: https://www.jstor.org/stable/23016178?seq=1&cid=pdf-reference#references_tab_contents You may need to log in to JSTOR to access the linked references.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide

range of content in a trusted digital archive. We use information technology and tools to increase productivity and

facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at

https://about.jstor.org/terms

INFORMS is collaborating with JSTOR to digitize, preserve and extend access to Interfaces

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Interfaces Vol. 41, No. 1, January-February 2011, pp. 35-50 issn 0092-21021 eissn 1526-551X1111410110035

doi 10.1287/inte. 1100.0533 ©2011 INFORMS

THE FRANZ EDELMAN AWARD

Achievement in Operations Research

Managing Global Brand Investments at DHL Marc Fischer

University of Passau, D-94032 Passau, Germany [email protected]

Wolfgang Giehl Deutsche Post DHL, D-53113 Bonn, Germany, [email protected]

Tjark Freundt McKinsey & Company, Inc., D-20457 Hamburg, Germany, [email protected]

In this paper, we introduce the customer-insight based approach that Deutsche Post DHL adopted to improve its global express delivery business. DHL has used the operations research based brand assessment tool in more than 20 large countries on four continents since 2004. The tool supports local brand managers in allocating marketing resources to activities that grow the global brand in their country market. Its application led to an estimated increase in brand value of USD 1.32 billion over five years. This corresponds to a return on investment of 38 percent and an internal rate of return of 24 percent. The tool's implementation also had a major impact on DHL's strategy and organization.

Key words: brand management; buyer behavior; estimation-statistical techniques; choice models; OR/MS implementation; service industries; nonlinear optimization.

Deutsche Post DHL is the largest logistic ser vice provider worldwide. It employs more than 510,000 people in more than 220 countries and terri tories, and moves 5 percent of the global trade vol ume. In 2008, it had global sales of USD 76.3 billion (Deutsche Post DHL 2008). Operations are at the heart of the organization, which delivers three million items and over 70 million letters each day, coordinates over 150,000 vehicles, and serves over 120,000 zip code areas worldwide.

Following the European Union's political initiative to deregulate monopolist industries, the state-owned German postal service, Deutsche Post, was trans formed into a private corporation in the early 1990s. Deutsche Post senior management was aware that it needed to develop a vision for the new, publicly listed company to address the challenges from declin ing growth in its existing business and to find oppor tunities from globalizing trade. It decided to build

a global logistics business to complement the tradi tional domestic mail services business. The acquisi tion of DHL in 2001 marked the first milestone of

this plan. DHL, founded in the United States in 1969, was a leading provider of offshore and intercontinen tal deliveries, with a strong presence in Asia. Other acquisitions followed, such as Airborne in the United States in 2003 and Exel in the United Kingdom in 2005; the result was a complex portfolio of several regional and specialist brands.

International logistics is a global business by def inition. Many customers operate on a global scale and expect to deal with one provider. They are likely to become confused when faced with different local

brands offering national and international delivery services. To ultimately become the largest logistics company worldwide, DHL had to follow an ambi tious international expansion strategy. Management realized that a strong, truly global DHL brand could

35

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL 36 Interfaces 41(1), pp. 35-50, © 2011 INFORMS

serve as an umbrella for the numerous acquisitions of regional and specialist brands. Establishing this brand would be the key to achieving its ambitious goal. In 2002, DHL began the brand-building effort that would fundamentally transform the organization.

Building a Truly Global Brand— The Challenges Building a global brand almost from scratch is a chal lenge for any firm. It requires defining and imple menting a brand concept that balances the needs for international standardization and localization.

According to marketing theory, brands are impor tant to customers because they provide important functions in the purchase decision and consumption process. For example, the brand serves as a signal of quality, reducing information costs and risks per ceived by the customer (Aaker 1996). Recent research shows that brands are relevant to customer deci

sion making in the international express delivery markets (Fischer et al. 2010). Strong brands promise higher future cash flows because of improved cus tomer acquisition and retention and/or higher-price premiums. The success of a brand, however, depends on a company's ability to create high brand awareness and superior brand image or reputation, respectively. Extant literature on brand management offers useful insights into the development process for a sustain able, superior brand image (Aaker 1996).

A major advantage of a global brand concept is that it aligns objectives and marketing activities across employees from different national operations. It also facilitates communication and provides guidance to customers. Management's great challenge, however, is to determine brand investments that are effective in

developing the global brand at the individual coun try level and are respectful of budget constraints. For a global logistics services provider such as DHL, this challenge means that the unique promise of the brand must be present at each of its million touch points with its customers worldwide. To achieve this brand presence, thousands of service employees must be educated to understand, live, and execute the

brand promise. When thinking about how to shape DHL's global brand positioning and how to sup port it with appropriate marketing investments at the

country level, management raised several important questions:

—What is DHL's brand-driven organic growth potential relative to its competitors in the focal country?

—Which stages of the purchase process offer the largest growth opportunities from brand investments?

—What are the key brand-positioning attributes? Which attributes of the global brand concept are most important for improving the purchase process in the focal country?

—How does DHL perform on these attributes rel ative to its competitors?

—How can DHL best allocate marketing resources across the different investment categories to differen tiate its brand from those of its competitors in ways that are most relevant to customers?

The answers to these questions are likely to dif fer across countries because customers and the com

petitive structure differ. However, these answers help local brand managers determine the actions that will develop the brand most consistently with the global brand concept, while adapting it to the specific requirements of their local market.

Many marketing experts perceive brand manage ment more as an art than a science. Creative, judg ment-based brand management is no doubt neces sary to the development and communication of a powerful brand. These traditional soft management practices, however, reach their limits when they con front the questions that need to be answered to make the right investment decisions. Consequently, the former CEO of Deutsche Post DHL, Dr. Klaus

Zumwinkel, felt that a fact-based approach that com bines proven operations research (OR) and statistical methods, marketing concepts, and in-depth customer insights from market research, would be the keys for building a global brand. In 2003, he initiated a senior management project under the leadership of DHL's department of corporate brand marketing and with support from Deutsche Post DHL's proprietary mar ket research service center (MRSC), academic experts, and McKinsey & Company, Inc. As a result of this project, DHL implemented an innovative approach of global brand assessment and development in more than 20 of the largest countries on four continents.

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL Interfaces 41(1), pp. 35-50, ©2011 INFORMS 37

In the next section, DHL's Global Brand Assessment

Tool—Understanding the Purchase Process to Make the Right Brand Investments, we describe the global brand assessment tool in detail. We illustrate its applica tion with reference to the UK market, which is a

key mature market characterized by fierce compe tition. In the Implementation and Impact on a Truly Global Scale section, we outline the challenges encoun tered during the implementation within DHL's inter national organization and discuss the tool's impact on DHL. Finally, in Lessons Learned, we elaborate on lessons learned that might provide insights to other industries.

DHL's Global Brand Assessment

Tool—Understanding the Purchase Process to Make the Right Brand Investments

The brand-management tool is theoretically based on the notion that customers pass through a stylized, hierarchical purchase process that represents different stages of customer development and value to DHL. The stages of this process are, in our express deliv ery case:

brand awareness —»■ brand consideration —> brand

usage -*■ choice of main provider —> choice of sole provider.

The idea of a hierarchical customer purchase deci sion process is a well-established concept in market ing theory (Peter and Olson 2008). Logically, there must be a hierarchy in the customer's behavioral steps to reach a decision. Customers can only con sider and evaluate a product or service, respectively, if they are aware of it. Purchase requires that the prod uct has entered the consideration set, which could

include other competitive offerings. Customers can pass through several stages at once. During visits by sales representatives, customers might learn of a new service, consider and evaluate it, and decide to try it.

Thus, the logical hierarchy that defines each stage as a subset of the preceding stages remains. This con cept of a decision hierarchy is universal; therefore, it is transferable to other industries. However, the num

ber and the definition of stages depend on the specific product or service category.

Breaking the purchase process into several stages offers three important advantages.

First, management gets a precise picture of the attractiveness and penetration of its brand among (potential) customers beyond traditional key per formance indicators (KPIs) such as awareness and market share. The picture is precise because we can measure how many customers are at a specific stage with respect to the brand.

Second, management can evaluate its customers in terms of economic relevance to the firm. Customers

generate different returns depending on the stage, with "sole-provider" customers contributing the largest share to sales and profits. In our express deliv ery context, a (business) sole-provider customer is a customer that generates at least 80 percent of its busi ness with one brand. If at least 30 percent of ship ments are allocated to a specific brand, we call this brand a main provider. Brand usage requires that the brand must have been chosen at least once during the last year.

Third, because the stages of the process have a behavioral meaning, the impact of specific marketing actions on changing that behavior can be analyzed. For example, emotional messages might increase the interest for a service so that the customers include it

in their consideration set. In contrast, reliability and fast delivery may be crucial for the choice of a main provider. Market research provides the data to dis close the brand attributes and marketing actions that drive customer decisions along the purchase funnel.

Technically, the hierarchical behavioral process can be specified as a sequence of conditional probabilities at the individual customer level. For customer i, the

unconditional probability of whether or not he or she chooses brand k as sole provider is given by

P-(sp,mp,u,c,a) = P-(sp\mp,u,c,a)-P-(mp\u,c,a)

■P*(u\c,a)-P*(c\a)-P*(a), (1)

where Pf(-) denotes the conditional or unconditional probability, respectively, for being in stage a (aware ness), c (consideration), u (usage), mp (main provider), or sp (sole provider). The utility-based theory of prob abilistic choice that goes back to the Nobel Prize winning work of Daniel McFadden (1974) provides the basis for estimating the power of marketing activ ities to change these probabilities. In the appendix, we

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL 38 Interfaces 41(1), pp. 35-50, ©2011 INFORMS

show more detail about the specification and estima tion of these models.

Market Research Provides the Empirical Facts Market research provides the input to the global brand assessment tool. To evaluate the purchase fun nel, we must collect information that is representa tive of a country's customers. For this purpose, we draw a random sample of at least 800 business cus tomers per country via computer-assisted telephone interviews. In addition to general information about customer businesses and express delivery usage, the questionnaire includes questions on actual brand per formance along the funnel (e.g., awareness, considera tion, usage). We cross-validated these easy-to-answer questions with external information when we devel oped the survey. The questions refer to DHL and up to six competitive local express delivery brands (e.g., Fedex, UPS, TNT). These competing brands are also evaluated in terms of 17 to 20 brand-perception items that cover global brand-positioning attributes, as well as customer perceptions of price performance and basic dimensions of service quality.

More than 20 countries from four continents

have been involved to date. In 2006, for exam ple, the set included the United States, Mexico, Brazil, Germany, France, the United Kingdom, Italy, Spain, the Netherlands, Sweden, Greece, Russia, South Africa, the United Arab Emirates, Turkey, Saudi Arabia, China, India, Japan, Singapore, and Hong Kong. For empirical analysis, we applied sampling weights to correct for variations in the probability selection of respondents and to account for differences in the total usage of express delivery services.

The market survey provides rich data that enable management to derive fact-based decisions about brand investments. Specifically, the global brand assessment tool uses this information according to a five-step process of analysis (see below). Each step answers one important question and represents the output of the tool. The first wave of data collection across the world began in 2004. It provided the basis for recommendations for developing the brand glob ally and adapting it to country-specific needs. These recommendations resulted in a reallocation of market

ing expenditures and refinement of communication campaigns and targets. In 2006, a second global wave

of market research was carried out to confirm that the

initiated action was producing the planned effect and to identify adjustment needs. After the initial calibra tion, the tracking cycle was extended to five years. The next global wave is planned for 2010-2011. Each country's management may, however, decide to col lect data in-between the global tracking cycles to mon itor its brand-development progress. Management in several countries did collect data in other years; for example, in Japan, market research was conducted each year from 2004 to 2009.

In the following sections, we describe the analysis process that produces the manifold insights for brand management and recommendations on investment allocation. The description follows the five-step pro cess; in each step, one central question is answered. We illustrate the process using data from the UK mar ket in 2006.

Step 1. How does the DHL brand perform? The fundamental assumption of the brand-manage

ment tool is that customers pass through a stylized, hierarchical purchase process whose stages corre spond to different stages of customer development and value to the firm. Analysis of the funnel pro vides the basis for measuring the performance status quo of the brand in terms of customer development. It produces two important insights. First, it shows the stages at which the brand is strong and at which it is not strong. The overall objective is to move as many customers as possible to the last stage, in which customers choose DHL as their sole provider (i.e., they use DHL for at least 80 percent of their express delivery business). The descriptive analysis across all potential customers in the market reveals the criti cal stages at which the brand loses most of the cus tomers. Second, it relates the performance of DHL to a competitor benchmark. Depending on the structure of competition and the strategic objective of the analysis, this benchmark may represent the performance of a single strategic competitor, the market average, or the efficient line that reflects the best performance in the market at each stage. For DHL, a strategic competi tor is defined in each country market as the relevant benchmark. This strategic competitor is the leading competitor in terms of market share.

Figure 1 shows the performance of DHL and TNT, the competitive benchmark in the UK. The numbers

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL Interfaces 41(1), pp. 35-50, ©2011 INFORMS 39

Values in percent (rounded) EXAMPLE FOR UK

Gap with respect to TNT

Percent

11 17 3

Figure 1: DHL lost more customers than TNT, the market leader in the UK, along the purchase funnel in the United

Kingdom in 2006. Source: DHL, MRSC.

represent percentage figures. The lower percentage figure within a brand row measures the proportion of customers that achieved a certain stage in the total market. For example, 100 percent of customers know DHL and 67 percent consider it a relevant choice. The upper percentages within a brand row represent conversion rates. For example, 62 percent of the customers that considered DHL actually chose its services at least once during the last year. As a result, 42 percent of all customers use DHL. At the individual level, which is the level of analysis for discrete-choice behavior, stage percentages repre sent the unconditional probability that a customer has achieved a certain stage. The conversion rates reflect the conditional probability that a customer moves to the next stage, provided that it has achieved the given stage. We assume that these conversion rates are influenced by brand-related marketing activities and idiosyncratic customer and market characteristics.

DHL has an advantage at the very beginning of the funnel when it comes to including DHL in the relevant set of express delivery providers. DHL's conversion rate is 67 percent, compared to 57 per cent for TNT. However, DHL loses its advantage across the following stages. The third row shows

the gap in conversion rates with respect to TNT. The largest gap arises when usage customers convert into main-provider customers. Here, DHL has a con version rate of only 49 percent compared to TNT's rate of 66 percent—a gap of 17 percentage points. Although both brands attract the same number of cus tomers (42 percent) at the usage stage, TNT is able to keep more customers at the main-provider stage (28 percent versus 21 percent) and the sole-provider stage (16 percent versus 11 percent). Because cus tomers at these stages contribute by far the largest share to sales, we can explain the difference in market shares of the two brands.

Step 2. How large is DHL's potential for organic growth?

Figure 1 shows DHL's customer conversion rate gap with that of TNT, its benchmark competitor. A natural question arises from this finding. How much of this gap can be closed by improving the brand? To answer this question, we must identify the gap between DHL and TNT in terms of brand perceptions. We also need to understand how impor tant the various dimensions and attributes of brand

perception are for the customer's decision. Knowing the gap in brand-attribute perceptions and the impact

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL 40 Interfaces 41(1), pp. 35-50, ©2011 INFORMS

Values in percent (rounded) EXAMPLE FOR UK

Awareness Relevant set

DHL 2006

Brand gap

Objective

Additional customers

Percentage points Estimated additional

revenues per year EUR millions

Total

r 100

Usage Main provider

Sole

provider

+1 Jjjl +3 II \\

67 43 26 1 15

0.7 37.8 257.1

295.6

Figure 2: Closing the gap in customer conversion rates by improving the brand perception reveals a revenue growth potential of ca. EUR 300 million for DHL in the UK 2006. Source: HC DHL, MRSC.

of these attributes on customer decision making, we can infer how much of the gap in Figure 1 can be closed by improving the brand.

Figure 2 outlines the findings for our case example. The second row of numbers shows that by improv ing the brand, we can increase customer transfer (in terms of conversion rate) from the relevant set to the usage stage by 1 percent, from usage to main provider by 12 percent, and from main provider to sole provider by 3 percent. The following two rows display the resulting number of customers and asso ciated gain at each stage in percentages. DHL has the potential to win 1 percent of additional customers at the usage stage, 5 percent at the main-provider stage, and 4 percent at the sole-provider stage. When we compare the improved customer transfer rates to the identified total gap in Figure 1, it becomes clear that the brand's potential to close the gap is only 9 percent for the first gap (relevant set -» usage), increases to 71 percent for the second gap (usage -» main provider), and achieves a maximum of 100 per cent for the last gap (main provider —► sole provider).

To calculate the improvement potential for cus tomer conversion, we take a quantitative approach to

model a customer's choice sequence, which underlies the funnel in Figure 2. Basically, we adopt a principle that is fundamental to many OR applications: opti mizing an objective function in a multidimensional space for an economic agent. Specifically, we assume that a customer maximizes utility when making deci sions along the funnel. For example, the customer must decide whether to include DHL in the relevant

set of express delivery service providers, whether to choose DHL from the relevant set, and whether to

choose DHL as main provider. Utility maximization implies rational economic behavior. Because our cus tomers are predominantly businesses, this assumption should be meaningful. Utility maximization is a proxy for economic goals such as profit maximization.

The utility is influenced by brand attributes that reflect the tangible and intangible benefits of the ser vice, and customer-specific characteristics. In addi tion, we include a random utility term to account for unobserved influences that we assume to be extreme

value distributed. We derive the likelihood function

that relates observed previous customer choices to the underlying theoretical model of utility maximiza tion. Maximizing the likelihood function by using

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL Interfaces 41(1), pp. 35-50, ©2011 INFORMS 41

advanced maximum-likelihood techniques produces the importance weights for the brand attributes. Based on the estimated importance weights and the required attribute improvement to close the gap to the TNT benchmark, we can compute the marginal effects of conditional choice probability at the individ ual level. Aggregating these marginal effects across all customers in the market gives the conversion rate improvement, and thus the potential gain in cus tomers stage by stage (see Figure 2). The appendix provides more details on how we specified and esti mated the probabilistic choice model and marginal effects.

As a final step in this analysis, we convert the potential gain in customers into expected revenue and profit figures. We start with the observation that cus tomers vary in their value depending on the stage. A customer at the usage stage generates approxi mately 10 times less revenue than a customer at the main-provider stage. This customer in turn gener ates only 50 percent of the revenue of a customer at the sole-provider stage. To calculate the share of each customer group in total revenues, we also must know the relative size of the groups. Forty-two per cent of all customers who are aware of DHL are at

the usage stage (see Figure 2). Approximately half of them do not move to the next stage. Hence, 50 per cent (= 21 /42) generate revenue at this stage. Sim ilarly, we obtain proportions of 24 percent for the main-provider and 26 percent for the sole-provider stage. Triangulating these figures with the spending multipliers 5, 50, and 100 yields the revenue shares of the three customer groups: 6 percent for usage customers, 30 percent for main-provider customers, and 64 percent for sole-provider customers. Deduc ing the revenue growth potential for each customer group from the improvement in customer conversion because of the brand is now a straightforward calcula tion. We note that double counting of new customers must be avoided. For example, a fraction of the new customers at the main-provider stage turn into sole provider customers. In total, we obtain a growth potential for DHL of 25.6 percent, which translates into EUR 295.6 million based on the brand's annual

revenues in the United Kingdom in 2006. The last row of Figure 2 shows how this potential is distributed

across the three stages. The sole-provider poten tial, EUR 257.1 million, is obviously largest. Using the average earnings-before-interest-and-taxes (EBIT) margin for the express business in 2006 (Deutsche Post DHL 2006), the profit-improvement potential because of the brand is EUR 5.6 million. This poten tial is substantial and refers only to one country and one year.

Step 3. What are the strengths and weaknesses of the DHL brand?

The gap analysis reveals significant potential for brand growth. This in turn suggests that the DHL brand has soft spots compared to its benchmark competitor. A detailed analysis of the brand image's strengths and weaknesses reveals the specific areas in which the improvement potential lies.

Brand image is a multidimensional construct mea suring brand performance through the customer's eyes. It includes intangible and tangible attributes as well as nonfunctional and functional benefits (Keller 1993). Figure 3 shows the image items identified by qualitative market research as best characterizing DHL. We present the items that were used in the 2006 UK survey, and since then. The intangible side of the DHL brand (see the left column) is represented by three facets that reflect core brand values as defined

by DHL: personal commitment, local strength world wide, and proactive solutions.

These values comprise several brand attributes or brand associations, respectively. "Likable" is an asso ciation reflecting personal commitment to the brand, for example, whereas "takes the initiative" expresses the ability of the brand to provide proactive solu tions. The tangible face of DHL (see the right col umn of Figure 3) comprises two facets. CEP (courier, express, and parcel) basics embrace the fundamental performance categories of an express delivery brand. The pricing facet simply measures the perceived price performance of the brand. It is important to break down the perception of DHL's image in the market into single items. These items define the substance of the brand promise and provide the links to develop the brand from a creative perspective. The advantage of the brand assessment tool is that it identifies the

needs for creative development that may be turned into briefings for an advertising agency, for example, from a rigorous quantitative approach.

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL 42 Interfaces 41(1), pp. 35-50, ©2011 INFORMS

Intangible attributes and non functional benefits

(7) Personal commitment • Strong customer relationships • Can-do attitude

• Likable

• Easy to use • Makes one feel confident and in control

(2) Local strength worldwide • Genuine worldwide service

• Comprehensive range of services and products

• Locally present

(3) Proactive solutions • Understands my business • Takes the initiative

• Excellent quality

Tangible attributes and functional benefits

(4) CEP basics • Is fast

• Is reliable

• Is trustworthy

(5) Pricing • Good value for money

• Has low prices

Figure 3: Several intangible and tangible brand attributes summarize the customer's perception of the DHL brand. Source: DHL, MRSC.

These brand attributes are not equally important for customers in different countries. Their decision rele

vance may also change across the stages of the pur chase funnel. Our statistical analysis for the United Kingdom found that the first two conversion steps are dominated by intangible brand attributes. A "gen uine worldwide service" and a perceived "can-do atti tude" are most relevant in drawing new customers to try DHL. When it comes to attracting customers to choose DHL as their main or sole provider, the tan gible attributes of the brand increase in importance. We found that express delivery services should be fast

and have low prices. It is not surprising that intan gible attributes play a larger role at the beginning of the funnel, and tangible attributes gain importance toward the end, when customers make purchase and repurchase decisions. The process reflects the behav ior of rational economic agents and can be gener alized across many markets. However, because rev enues are generated at the final stages, we should not be tempted to focus only on drivers relevant to these stages. The success at the end of the funnel

inevitably depends on the pass-through of customers at the beginning.

Figure 4 shows the brand image item ratings for DHL and the UK benchmark, TNT, in comparison to the market average. For 10 of 17 image items, DHL is perceived better than the market average. This is especially true for intangible attributes such as genuine worldwide service or can-do attitude, which are important for the first two funnel conver sion steps. Here, the brand is quite successful and shows a performance similar to TNT in terms of customers maintained at the relevant-set and usage stages (see Figure 1). TNT is the clear market leader, because all its image items are perceived as better than the market average. It outperforms DHL on sev eral items, including "strong customer relationships." More importantly, TNT's perception is better for "easy to use" and "low prices," which are especially crucial to acquiring customers at the main-provider and sole provider stages. This result is consistent with the gap analysis of Figure 2, which showed that the potential to close the gap by improving the brand is highest for the last two stages.

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL Interfaces 41(1), pp. 35-50, ©2011 INFORMS 43

Deviation in percent from market average EXAMPLE FOR UK

■ <98©

Brand attributes "10 ~5 0 5 10 15 i i

Makes me feel confident and in control

Known for excellent quality Reliable

T rustworthy

Provides a genuinely worldwide service Fast

Has a can-do attitude ■

Takes the initiative k%

Understands my business

Builds a strong customer relationship Likable

Offers a comprehensive range of services and products w'

Environmentally aware company

Offers a good value for money \ Easy to use p

Locally present v

Low prices

Figure 4: TNT outperforms DHL on many brand attributes. Source: DHL, MRSC.

Step 4. Which positioning attributes should receive focus?

Actionable management implications can be de rived from a simple but powerful 2x2 matrix (see Figure 5). The ordinate measures the overall relevance of a brand attribute for customer decisions along the

purchase funnel. We compute the overall relevance by aggregating the normalized importance weights of a brand attribute across the four stages. The x-coordinate measures the difference between DHL's

brand perception and the market average with respect to each brand attribute. The northeast quadrant con tains attributes that are highly relevant for customer decisions and represent a DHL strength compared to the market average. These attributes must be nur tured. Figure 5 shows that the DHL strength in the United Kingdom lies predominantly in its percep tion as a genuine worldwide service provider with fast deliveries. DHL is also stronger than the market

average for brand attributes located in the southeast quadrant. However, the relevance of these items for purchase, and thus revenue generation, is lower. As

a consequence, they should be used selectively for differentiation.

The matrix also reveals two soft spots for DHL (see the northwest quadrant). DHL is perceived below market average with respect to "low prices" and "easy to use." Both attributes are highly relevant in the purchase process. Moreover, TNT's performance on these attributes is above market average, and thus much better than DHL's. To close the gap to the benchmark in an optimal way, DHL must focus on improving these two attributes in the country of focus, in this case the United Kingdom.

Managerial recommendations derived by means of the matrix in Figure 5 are generally qualitative. Such recommendations play an important role in brand management. Marketing and sales managers, external agency partners, and employees at the millions of cus tomer touch points need to know where the brand's soft spots are. This enables them to focus their mes sages and customer interactions on the relevant con tent. The objective is to enhance DHL's image with respect to the crucial brand attributes. The innovation of the brand assessment tool is that it identifies these

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL 44 Interfaces 41(1), pp. 35-50, ©2011 INFORMS

EXAMPLE FOR UK

Strengths/weaknesses against market average*

Weaknesses (-10%)

Relevance

for purchase

High Expand

Low prices

Easy to use

Low

0 Strengths (+10%)

Provides a genuinely worldwide service

-<38©

Low prices

Utilize

. Fast

Can-do attitude

Easy Trust to use worthy

-05© JR.

Reliable jm

Known for

excellent

quality

Understands

my business

Use tor differentiation

* Deviation in percent

Figure 5: The brand attributes "low prices" and "easy to use" offer the largest growth potential because their purchase relevance is high and DHL's performance is weak. Source: DHL, MRSC.

attributes by using a quantitative approach to exploit information from the demand side.

Step 5. What is the optimal allocation of the mar keting budget across media and activities?

The quantified impact of each brand attribute on the purchase funnel is also the key input for the last step of our analysis: the optimal alloca tion of a marketing budget across media and activ ities. In our B2B express-delivery context, a mar keting budget can be spent in different ways. It could be used to launch campaigns in communi cation channels, including print media, direct mail ing, or online media (e.g., Internet, e-mail). It could also be used to develop and enhance the interac tion of DHL sales representatives, employees, and other company representatives with their customers. Such activities might require investments in sales folders and internal training for sales representa tives and employees. The choice of the medium or activity, respectively, is largely determined by the choice of which brand-positioning attributes are to

be improved, and by the process performance that determines overall marketing objectives (e.g., acqui sition versus retention). For example, communication campaigns in mass print media could be effective in changing the perception of items such as "likable" and "environmentally aware company." In contrast, strengthening DHL's brand perception as a service that is "easy to use" and has "low prices" would involve very different activities. Apart from improve ments to the service processes and pricing structure, these changes would require training and communi cation efforts that are targeted inward. The perception of an "easy to use" service provider, as an example, is greatly influenced by customer experiences with com mon processes such as billing, invoicing, and com plaint handling.

Because financial resources are usually limited, optimal marketing budgets must be determined under such a constraint. This is also true for DHL

where annual budgets for country units are planned and fixed the previous year. We know from both

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL Interfaces 41(1), pp. 35-50, ©2011 INFORMS 45

theory (Mantrala et al. 1992) and implemented work (Fischer et al. 2011) that the profit-improvement potential from optimizing the allocation of a fixed total budget is far higher than from optimizing the total budget. We developed an allocation rule that provides exact results of how to allocate the bud get across the brand attributes. Note that we already know from the discussion of the implications matrix in Figure 5 that marketing activities should focus on the attributes "low prices" and "easy to use." However, concentrating all expenditures on activities that strengthen these attributes would be inappropri ate. Items in the other two quadrants should not be neglected. The allocation rule helps by finding the optimal balance between investments in these image attributes.

Technically, we solve a constrained-profit-maxi mization problem to derive the rule (see the appendix for details). The solution is easy to implement in a spreadsheet environment. It suggests allocating the total budget across the brand attributes proportional to their effectiveness, measured in terms of elasticity. By elasticity, we mean the relative increase in revenue (in percentages) that is because of a relative increase in the brand attribute rating (in percentages). Elas ticities are obtained from the purchase-funnel anal ysis. They are composed of the estimated marginal effects for the various stages of the purchase funnel. The higher the total marginal effect (i.e., the overall purchase relevance of an item), the higher its elastic ity. However, elasticity decreases with the rating score already achieved for the attribute. Therefore, elastic ity is usually higher for items located in the upper left quadrant in Figure 5 than for those located in the upper-right quadrant. As a result, the marketing activities required to improve these items should be allocated a larger share of the total budget. In our case example, the available budget was spent almost equally across the different attributes in the United Kingdom. Assuming a 10 percent improvement in attribute rating, revenues are estimated to increase by 8.2 percent. Our allocation rule, however, suggested spending 54 percent (30 percent before) of the bud get on the two attributes in the upper-right quad rant of Figure 5, 38 percent (37 percent before) on the three attributes in the upper-left quadrant, and 8 per cent (33 percent before) on the four attributes of the

lower-right quadrant. With this better allocation of the budget, sales are estimated to increase by 13 percent. Because the budget is the same, the difference of 4.8 percent in sales increase (13 percent-8.2 percent) is fully profit relevant.

Implementation and Impact on a Truly Global Scale

Challenges for Implementation Implementing the global brand-assessment tool in the DHL organization, a company with its nearly 300,000 employees located around the world, required tremendous effort. Challenges arose from many sides.

Market research. The first challenge was to obtain the necessary buy-in and support from the central market research unit and the local business units.

High-quality data are a key input for the tool. Tra ditionally, market research was specialized on coun try topics rather than global brand topics. The market research units had to be convinced to support a large research effort that is focused on achieving one global DHL brand. A global research effort that involves more than 20 countries on four continents requires careful rollout planning to handle time, cost, and pro cess complexities. It also poses an extraordinary chal lenge for the development of the survey material to produce results that are comparable across countries. Organization. The organization also faced sev

eral challenges. First, management at the various hierarchical levels had to be inspired to act as role models and embrace the brand as a success factor.

Motivating employees of acquired local brands, such as Danzas and Airborne, and convincing them to accept the new global brand and follow its brand promise was also crucial. This also applied to business units and corporate center service departments, which must adhere to the recommendations of the tool for

success to become generalized. Finally, the manage rial implications of the global brand-assessment tool can only be successful if thousands of service employ ees "live" the brand values and act accordingly at the myriad customer touch points.

Implementation Efforts DHL took several measures to promote brand build ing and consistent brand delivery.

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL 46 Interfaces 41(1), pp. 35-50, ©2011 INFORMS

Senior management support. Historically, brand management receives less management attention in a logistics company such as DHL, which focuses primar ily on operations. Senior management involvement is therefore crucial for a successful implementation. The former CEO, Dr. Klaus Zumwinkel, and the current

CEO, Dr. Frank Appel, committed heavily to initiating and pursuing the global-brand building effort.

Brand delivery. In terms of external brand build ing, DHL harmonized its global design, revising the appearance of 150,000 vehicles across the worldwide DHL organization. The "DHL Brand World" (http:// www.dhl-brandworld.com) was set up as a brand communication platform for both customers and employees. Specific internal brand-building initiatives were developed to improve understanding, create excitement, and eventually change the mindset and behavior of employees. These initiatives included the development of a "brand book" that illustrates the core values of the DHL brand strategy-—proactive solutions, personal commitment, and local strength worldwide. DHL distributed 1,500 copies to managers in its global organization. "Golden rules for brand delivery" were furthermore derived from brand positioning workshops with employees across busi ness units in multiple countries. Internal service units organized training and events for employees.

Standardized market research. DHL also adopted several actions to ensure that global market research is executed in a cost-effective way without sacrific ing quality standards. A standard process tool was developed and implemented within DHL's research and reporting systems. This tool provides guidelines for market research and data analysis, and a sum mary of managerial implications. Dedicated person nel in Deutsche Post DHL's proprietary MRSC own this research.

Strategic Impact on DHL The tool's implementation had a strong impact on DHL's global brand strategy. Data analyses from numerous countries were extremely helpful in sharp ening the understanding of brand attributes that drive customer behavior. As a result, the company devel oped and refined its global brand positioning. Market research showed that countries differ in their needs to

strengthen the core values of the global brand.

The model was very helpful in establishing the relevance of brand strategy and fact-based brand management throughout the organization. Brand management and brand value became a permanent part of strategic discussions at the global board. The use of instruments such as the brand book was estab

lished among all senior and middle managers. The global brand-building effort resulted in an im

proved perception of DHL as a global partner for sev eral customers. Bernie Ecclestone, President and CEO of Formula One Management, attested to the suc cess of DHL's brand strategy: "We are obviously very proud and very happy to be partners with a very suc cessful brand" (B. Ecclestone, personal communica tion). Pascal Eymery, Vice President of Supply Chain and Logistics at Airbus, highlighted trust as a core perception of the brand, resulting in joint advertising campaigns: "DHL has earned our trust because they are committed to very good performance standards in terms of quality, service, and costs" (P. Eymery, per sonal communication).

Financial Impact on DHL Given the dynamic environment, unforeseen competi tor moves, and the long-term horizon to realize brand benefits, evaluating the financial impact of the global brand assessment tool is difficult. A meaningful way to measure its impact is to track the development of the brand asset's value over time. By definition, the financial value of a brand captures the differ ence in future incremental cash flows that result from

the brand when compared to an unbranded busi ness. Therefore, changes in brand value over the years reflect the financial effects of brand investments. We

adopt the valuation approach suggested by Fischer (2007) to compute DHL's brand value for 2003-2008. The model is based on the notion that brand value

is part of the enterprise value for the total branded business, which is similar in spirit to models used by commercial firms such as Interbrand or Millward Brown. It requires determining the brand equity share from market research and the enterprise value or discounted cash flows for the brand, respectively. The global brand-assessment tool provides country specific data on DHL's brand image that needs to be weighted with overall brand relevance in the express delivery category. We obtained this informa tion from Fischer et al. (2010) for several countries.

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL Interfaces 41(1), pp. 35-50, ©2011 INFORMS 47

We used industry-specific sales multiples provided by the Bloomberg proprietary database to estimate the enterprise value for DHL. To control for short-term stock market fluctuations, we took the average for 2003-2008. We did not consider 2009; because of the

extraordinary (financial) crisis, it is not representative. Controlling for the effects of brand acquisitions

in a conservative approach, we estimate a signifi cant brand-value increase of USD 1.32 billion: from

USD 4.19 billion in 2003 to USD 5.51 billion in 2008.

As a result, brand value grew by 32 percent, or 5.6 percent per annum. Assuming a cost of capital of 6.2 percent, building up this brand value would require an annual investment of USD 269 million. However, average actual brand expenditures in 2003 2008 were lower at USD 200 million. Note that these

expenditures refer only to the improvement of brand perception; total marketing and operations expendi tures were much higher. As a result, we compute a brand return on investment of 38 percent and an internal rate of return of 24 percent.

If we do not control for the effects of brand acquisi tions, the estimated brand value amounts to USD 8.22

billion in 2008. This value agrees with the result of the independent valuation by Millward Brown (2009), which included DHL in its top 100 global brand list in 2008. It estimates the value for DHL at USD 9.72 bil

lion in 2008. It is indisputable that brand is now a key DHL asset. In internal discussions, this achievement is

attributed to the rigorous brand-management process during the last years. Millward Brown (2009, p. 29) adds: "Sometimes it looks like an overnight success. But it rarely is. The brand [... ] achieved [its] position due to hard work over many years, developing and communicating clear and sustainable benefits."

Organizational Impact on DHL Improved communication. The model had a consid erable impact on the understanding and application of brand management within DHL and, more impor tantly, on its overall culture. The application of a fact based approach and the connection of (qualitatively defined) brand attributes with hard sales data make brand management tangible for DHL management, functions other than marketing, and employees. The valuable insights that the brand-assessment analysis has provided on customer purchase drivers, competi tor strengths and weaknesses, and ultimately on the

impact of brand on sales, have triggered a new con sciousness among DHL's employees of the impor tance of brand management for business.

Improved controlling. The brand-assessment tool is also used as a controlling tool. It enables man agement to define and monitor clear responsibilities. It is the vehicle to transport and communicate the three core values of the global brand concept. The tool helps align objectives and metrics for brand manage ment among managers from different operations of the multinational company.

Improved employee identification. Finally, the establishment of DHL as the second major brand, beside Deutsche Post, strengthened the identity and culture of the DHL organization overall. DHL employees feel proud of being members of a single global DHL organization and recognize the contribu tion their actions as DHL employees can have on the overall success of the Deutsche Post DHL Group.

Lessons Learned

Several lessons that we learned from this long-term effort might be valuable for other firms interested in establishing a fact-based brand-assessment and man agement tool.

First, although brand management requires creativ ity and qualitative insights to grow the brand, using quantitative techniques to guide this process is pos sible. Specifically, OR offers powerful methods and concepts for generating fact-based decision options in marketing, and particularly brand management; these options might sometimes appear counterintuitive to common beliefs.

Second, a key lesson we learned is that senior man agement attention and support are absolutely crucial. Precisely because brand management is often seen as an intuitive discipline whose results are difficult to measure, it also attracts less attention than it deserves

in many organizations. Third, management must align the organization to

ensure "delivery of the brand promise" in opera tions throughout the organization. Providing enough financial resources for workshops, training, and other supportive activities is also crucial.

Finally, consequent tracking instruments should be implemented to ensure that the right measures are in

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL 48 Interfaces 41(1), pp. 35-50, ©2011 INFORMS

place to capture the appropriate effects. These metrics also inform the employees about the impact of their brand delivery on sales and earnings. To summarize, we believe that OR principles work

well for brand management, a discipline perceived traditionally as qualitative. The brand-assessment tool is innovative, and one of the first of its kind to ana

lyze the impact of brand investment decisions in a logically consistent framework that connects quali tative brand decisions with economic consequences. Because the hierarchical customer decision process and the theory of utility maximization are univer sal theories of the behavior of economic agents, the approach can be generalized across industries. Many companies in diverse industries worldwide have implemented this concept. Hence, the approach should help other companies to capture the value potential of superior brand management.

Appendix Brand Choice Model

In this section, we show details on the specifica tion of the sequential probabilistic choice model that provides the theoretical base of the purchase fun nel (see Equation (1)). Conditional probabilities are determined by a heterogeneous (across customers and brands) constant preference term, brand asso ciations including perceived marketing-mix perfor mance, and a customer-specific error term that is assumed to be independently distributed, each with an extreme value distribution. Assuming utility maximizing behavior, customer i's conditional proba bility of considering brand k is

»!■) = , <2> Exp(v, ) + Exp(iA*) v? = acki + pxki,

where v°k measures the value to consider brand k and

vnck to not consider the brand. Xki denotes the vector of brand associations, and acki and (J' are parameter vectors to be estimated. Specifically, measures the importance weight with respect to brand association I. For identification purposes, we normalize vn°k to zero. Conditional choice probabilities for the other stages of the purchase-process funnel are specified in the same fashion.

Based on the estimated importance parameters, |i, and the observed conditional probability, we compute the marginal effect for brand association I as follows:

dP? = fi,I*(l-P*)dXu. (3)

The associated elasticity bc£ is given by

e%=(3,(l-P*)Xktl. (4) Estimation

The specification of the conditional choice models accounts for customer and brand heterogeneity in the constant preference term, Equation (2). The heteroge neous parameters to be estimated are summarized in vector ak, k = l,... ,K. We impose the following het erogeneity structure on these parameters:

aki = ak + Ahvu + Xk2v2i, where vu, v2i ~ N(0,1)

and Cov(ui;, v2i) = 0. (5)

In the above expression, aki represents an unknown brand customer-specific preference parameter ak; Atl and \k2 are heterogeneity parameters to be esti mated; vh and v2i denote variance components that vary by customer; and K is the number of hetero geneous brand preference constants (k = 1,2,..., K). The implied variance of aki is (Aj^ + Aj^). The variance covariance matrix for a, is given by X = AA'.

We maximize the likelihood function for the

choice model, Equation (2), in combination with Equation (5):

Here, P(Ycki | Xki, ak, p, X, v,) denotes the probability density of the conditional choice to consider brand k, Yki, for all customers, given the brand associations Xki and the vector of unknown parameters ak, (}, and X. Vj is the standard normally distributed latent vari able, which we need for the estimation of the vari

ance of (heterogeneous) brand preference parameter distributions.

Because the likelihood function includes a multidi

mensional integral that does not provide a closed-form solution, it must be evaluated by using approximation methods. We follow the procedure suggested by Ver munt and Magidson (2005) and apply Gauss-Hermite

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL Interfaces 41(1), pp. 35-50, ©2011 INFORMS 49

numerical integration. Following Bock and Aitkin (1981), the multidimensional integral is replaced by multiple sums and approximated by

[ /K-)P(ycki | Xkj, ak, 0, \, v,) dvj vi

* E E 1xti,«t,p, X, vbl)PbPh, (7) fc1=l h2=l

where B denotes the number of quadrature nodes per dimension, vb and vbi are the locations, and Pbi and Pbi are the weights corresponding to quadra ture nodes bx and b2. The procedure uses nodes and weights as provided in quadrature tables by Stroud and Secrest (1966).

The maximization problem of the likelihood function is solved by employing the expectation maximization (EM) algorithm and the Newton Raphson algorithms in combination (Vermunt and Magidson 2005). The advantage of EM is its stabil ity in approaching the optimum, whereas Newton Raphson is faster than EM when it is close to the optimum. We start with 250 EM iterations at maxi mum and switch to the Newton-Raphson algorithm to obtain the final solution. We use 500 random sets

of start parameters to minimize the danger of finding a local optimum.

Allocation Rule

Denote (/[X(S), Z] as the DHL brand's unit sales that depend on X, the row vector containing the perceived brand associations, and other variables (e.g., compet itive marketing) that are summarized in the row vec tor Z. Brand associations are influenced by marketing expenditures denoted by S. Let pre denote the average price, mc denote marginal cost, and R be the market ing budget, which is fixed for the fiscal year in a spe cific country. The country-level profit-maximization problem of Deutsche Post DHL is

max IT = (pre — mc)q[X(S), Z] — S (8)

L

subject to R = J2S: and S, > 0. (9) 1=1

Note that it is sufficient to maximize profit contri bution before marketing cost because these are fixed by the total budget; thus, they are not relevant to

optimization. To solve the problem, we set up the Lagrange function

L = (pre - mc)q[X(S), Z] - h(r - £S;),

which leads to the following first-order condi tions (FOC)

dL / xdqdX r „ 0 and

dL ^ _R + £S; = 0, (10)

^ /=i

where /a is the Lagrange multiplier and I is the unit vector. We transform and expand the FOC to obtain

dq X* dX) S* 1 (pre - mc —- — q —— — — — a* VK y dX, q* ^ dS, X* Sf ^

E,Ex*,s*,; (11)

^ Si=~^^rc~mc^*er,X',ieX',s',i' VZ = 1,...,L,

and l

R = J2si- (12) ;=i

The star indicates that variables ji, q, X, and S are in their optimum. We insert Equation (11) into Equa tion (12) and derive the optimal budget share for mar keting activity I:

S*m (1 /V*)(prc - mc)q*sfis«, m R ELi(1/m*)(FC - mc)q*Bp x* tex, j'

V/ = l m L. (13)

We assume that there is a constant marginal cost for increasing brand attribute perception, which may vary across attributes, but no fixed cost. As a result, the cost elasticity sx s is equal to 1. Using this result and simplifying Equation (13), we obtain the follow ing general allocation rule:

s;= " R, Vl = l, (14) z2i=i sf,x»,i

References

Aaker, D. A. 1996. Building Strong Brands. The Free Press, New York.

Bock, R. D., M. Aitkin. 1981. Marginal maximum likelihood estima tion ot item parameters. Psychometrika 46(4) 443-459.

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms

Fischer, Giehl, and Freundt: Managing Global Brand Investments at DHL 50 Interfaces 41(1), pp. 35-50, ©2011 INFORMS

Deutsche Post DHL. 2006. Deutsche Post World Net annual report 2006. Bonn, Germany.

Deutsche Post DHL. 2008. Deutsche Post World Net annual report 2008. Bonn, Germany.

Fischer, M. 2007. Valuing brand assets: A cost-effective and easy to-implement measurement approach. MSI Report No. 07-107, Marketing Science Institute, Cambridge, MA.

Fischer, M., F. Volckner, H. Sattler. 2010. How important are brands? A cross-category, cross-country study. J. Marketing Res. 47(5) 823-839.

Fischer, M., S. Albers, N. Wagner, M. Frie. 2011. Dynamic marketing budget allocation across countries, products, and marketing activities. Marketing Sci. Forthcoming.

Keller, K. L. 1993. Conceptualizing, measuring, and managing customer-based brand equity. J. Marketing 57(1) 1-22.

Mantrala, M., P. Sinha, A. A. Zoltner. 1992. Impact of resource allo cation rules on marketing investment-level decisions and prof itability. J. Marketing Res. 29(2) 162-175.

McFadden, D. 1974. Conditional logit analysis of qualitative choice behavior. P. Zarembka, ed. Frontiers in Econometrics. Academic Press, New York.

Milward Brown. 2009. BrandZ top 100 most valuable global brands 2009. Retrieved February 12, 2010, http://www.brandz.com/ upload/brandz-report-2009-complete-report(l).pdf.

Peter, J. P., J. C. Olson. 2008. Consumer Behavior and Marketing Strat egy, 8th ed. McGraw-Hill, Boston.

Stroud, A. H., D. Secrest. 1966. Gaussian Quadrature Formulas. Pren tice Hall, Englewood Cliffs, NJ.

Vermunt, J. K., K. Magidson. 2005. Technical Guide for Latent GOLD 4.0: Basic and Advanced. Statistical Innovations Inc., Bel mont, MA.

This content downloaded from 128.6.218.72 on Sun, 28 Apr 2019 18:01:30 UTC All use subject to https://about.jstor.org/terms