Website Features and Functions as predictors of business performance

(1)

Website Features and Functions as Predictors of

Business Performance

Master Thesis

Programme: MSc Business Studies

University of Amsterdam Department of Economics and Business

Author: Valentina Barzakova Student Number: 5896193 Email: V.Barzakova @student.uva.nl

Supervisor: Thomas Adelaar Information Management Amsterdam Business School

University of Amsterdam

(2)

2

Figures

Figure 1: Five-step purchasing model. ... 20

Figure 2: iPACE purchasing model. ... 22

Figure 3: Purchasing model derived from consumer behavioral studies. ... 25

Figure 4: Pre-purchase phase - Problem Recognition. ... 26

Figure 5: Pre-purchase phase – Information Search & Evaluation of Alternatives. .. 28

Figure 6: Purchase phase – Purchase decision, intention and implementation. 29 Figure 7: Post-purchase phase – positive/negative reinforcement. ... 30

Figure 8: Three-Step Purchasing Model Phases. ... 31

Figure 9: Conceptual model and hypotheses. ... 32

Figure 10: General framework. ... 35

Figure 11: Variables definition. ... 41

Figure 12: Residual Plot.. ... 58

Figure 13: Adjusted R squared. ... 62

Figure 14: R2 change. Hierarchical Regression, 4 levels.. ... 63

Figure 15: R2 change. Normal Regression, 2 levels ... 64

Figure 16: Hierarchical Regression Coefficients. ... 67

(4)

4

Figure 18: Residual Plot. Different industries. ... 73

Figure 19: Adjusted R squared. Different Industries. ... 75

Figure 20: R2 change. Different industries. 77 Figure 21: Regression Coefficients. Different industries.. ... 79

Figure 22: Final Framework.. ... 86

Tables

Table 1: Literature on Causal Relationship Approach.. ... 11

Table 2: Literature on Benefit Relationship Approach. ... 14

Table 3: Literature on generalized e-measurements. ... 16

Table 4: Literature on website metrics. ... 18

Table 5: Literature on consumer behavior in online and offline environment. ... 23

Table 6: Hypotheses – Purchasing process. ... 32

Table 7: Hypotheses – Merchant type and industry.. ... 34

Table 8: Website features in the Pre-purchase phase. . ... 42

Table 9: Website features in the Purchase phase.. ... 43

Table 10: Website features in Post-purchase phase. ... 45

Table 11: Correlations Diagnostics, Model 1.. ... 51

Table 12: Correlation Diagnostics, Model 2. ... 52

Table 13: Collinearity Diagnostics, Model 1. ... 55

Table 14 Collinearity Diagnostics, Model 2. ... 55

Table 15: Regression Summary, Model 1 (Adelaar 2007). ... 59

Table 16: Regression Summary, Model 2 (Top 500 Retailers Guide 2008). ... 60

Table 17: Hierarchical Regression Coefficients. ... 66

Table 18: Normal Regression Coefficients.. ... 68

Table 19: Hypotheses status. ... 71

Table 20: Number of Observations per Industry. . ... 72

Table 21: Regression Results. Different industries. ... 74

Table 22: Regression Coefficients. Different industries. ... 78

(5)

5

Abbreviations

BRM Belief Reinforcement Model BSC Balanced Scorecard

CRM Customer Relationship Management

iPACE Information, Price, Assortment, Convenience, Entertainment KPI Key Performance Indicatior

(6)

6

Abstract

Today’s economy is made up of old and new elements and is essentially a hybrid.

Today’s companies need fresh thinking about how to operate and compete in the new economy.

Philip Kotler, 2006

The virtual marketplace is one of the outcomes of the technological advance and globalization. This new form of business has changed the traditional marketplace to an elusive, dynamic and information-intensive system. These trends forced companies to rethink and adapt their strategies and to develop a new dual online-offline business model. This task requires companies to develop a completely new metric system, able to enhance and measure both online and offline efforts and their contribution to the business performance.

The need of online metrics is a widely discussed problem. However, none of the current studies suggests a reliable and easily-applicable e-metric system. This paper strives to fill this gap by constructing a new measurement model, based on the classical consumer behavior theories combined with more recent e-commerce studies. The new model describes the purchasing process in three consequent phases: pre-purchase, purchase and post-purchase, and highlights the website features which propagate each of these phases. The validation of this model is based on a statistical analysis of two datasets based on two independent website metrics.

The model suggests that there is positive relation between relation between the purchasing phases and the business results, and that companies could optimize their performance by a thoughtful planning and implementation of their website features. This paper proves the existence of a relationship between company‟s website and online performance, and provides some insights into the online business environment. Further, it develops a model which could be refined and used as an online metrics.

(7)

7

1 INTRODUCTION

The Web is the most measurable medium in the history of marketing. Now all that’s left is figuring out how to measure it.

McKinsey Quarterly, October 2008

Online measurement models are quite often discussed in the practical and scientific literature. Companies need to measure and control their online performance and to scale their operations to investments. Scientists also struggle with this question, since it would help them to create a complete theoretical framework of the online and offline business environment.

According to a McKinsey digital-advertising survey many businesses are failing to exploit the complete potential of the e-markets due to their inability to measure and control performance (Bughin, Shenkan and Singer 2008). The McKinsey report states that the majority of companies change the ratio of offline/online activities on favor of the later (55% lower expenses on traditional media in order to increase online marketing; Bughin, Shenkan and Singer 2008). They expect that the new medium is more efficient than the traditional one. Although, spending on the e-channel is growing, the measurement of internet activities and performance has not increased (43% do not measure the online performance; Bughin, Shenkan and Singer 2008).

Another finding of the survey says that companies rapidly increase their effectiveness by studying the effect of online activities and applying innovative metrics to the new form of business.

This survey provides a vivid evidence of the need for a sophisticated e-metrics which enhances the dynamics between business activities and their results (Barnes and Hinton 2007). This is important for major business activities (Barnes and Hinton 2007).

Companies require a new performance measurement system which reflects the specifications of the e-business environment and conceptualizes new business models in the “digital economy” (Lohse and Spiller 2006). The available e-commerce literature could not suffice this urgent need for a relevant e-business measurement. Most of the studies focus only on a particular business aspect, evaluating the isolated effect on overall performance. Only a few studies provide more comprehensive measurement models.

(8)

8

However, these are solely theoretical frameworks without practical validation and analysis details. Hence, the literature and the business practice reveal a lack of appropriate and easily applicable measurement for website performance.

The present paper strives to fulfill this scientific gap and to provide a practical business solution. The main research question of this paper is:

How the various website features and groups of features affect on online business performance?

A significant part of the analysis is dedicated to clustering variable into groups and measuring their contribution to online sales. For this purpose, this study builds a framework which defines the purchasing process in three main phases – Pre-Purchase, Purchase and Post-Purchase. These phases represent consumer behavior from the business‟s point of view, i.e. they focus on the business activities employed by companies in attempt to satisfy customers‟ needs and bring them to the next phase. In this case, the consistent and satisfying customer experience is perceived as a major driver of companies‟ performance.

The website serves as a representative of the companies‟ e-commerce activities. Its features are analyzed and classified into three main categories which represent the purchasing phases. Further the features within each category (phase) are grouped into functional subsets, e.g. features which provide information, enable transaction etc.

These subsets are treated as basic triggers of performance. Their effect is measured against the online performance represented by online sales. This measure is used because the online revenues best reflect the direct effect of the website on overall profitability and are correlated to other performance parameters such as, online visitors, customer satisfaction index etc. A distinction is made for the website performance in different industries. This additional step provides a clear and broad view of the online channels‟ performance in the contemporary economy.

This research provides a conceptual framework which could be tailored by the companies to serve as a benchmark for companies‟ planning, monitoring and control activities. The model could also serve as a valuable tool for matching website activities with the performance targets. Moreover, this study complements to the total body of e-commerce literature and would clearly preview the cause-and-effect relationship between online strategy and performance.

(9)

9

2 LITERATURE REVIEW

“Everything that can be counted does not necessarily count; everything that counts cannot necessarily be counted.”

Albert Einstein (1879-1955)

The issue of measurement and accountability of the business performance is a “common concern” in business communities (Barnes and Hinton 2007). Barnes and Hinton (2007) established that it is necessary to identify the features of an “effective e-business performance measurement system.” They reveal that the majority of companies have a narrow range of e-business performance measures.

Due to this fact many organizations apply an “ad hoc approach” in measuring e-business performance and do not reveal “any sense of balance between the measures used” (Barnes and Hinton 2007). Businesses need to be able to plan, monitor and evaluate their own actions, and to predict their effect on the overall performance. An adequate e-metrics would enable businesses to develop “innovative solutions to emerging problems” (Straub 20027).

As noted by Barnes and Hinton (2007), there are not many studies on online performance metrics, which could be explained by the fact that the e-commerce phenomenon is comparatively new to the global marketplace.

The literature review revealed several main approaches in measuring performance. One of these different methods, referred in this paper as Direct Causal Relationship Approach is focusing on the direct causal relation between company‟s efforts and financial performance. It strives to identify how various business activities boost financial performance.

Another method enhances an alternative perspective of indirect performance measures. This approach focuses on interim business objectives such as reduced transaction cost, differentiation, improved communication, or higher customer satisfaction etc. They online efforts are measured against these parameters rather than to sales. All these intermediate measures are considered to have high importance for the overall business success.

(10)

10

Therefore, this approach is referred to as Indirect Benefit Relationship Approach since it considers the benefits delivered indirectly by the online channel. This approach discerns two benefits‟ categories – Business-centric and Consumer-centric. The former considers benefits received by the company (better communication, lower transaction costs etc) while the later emphasizes on the consumers gains (less search efforts etc).

It is important to note that both approaches enhance the business perspective since they perceive business and consumer benefits as a contribution to the companies‟ performance. For instance, consumer satisfaction is considered as the main factor if influence for business success since it stimulates online traffic and sales.

An alternative approach represents the consumers‟ view of the online purchasing process. It highlights the benefits received by the consumers but does not translate them into a business value. This approach is related to behavior studies and does not deal directly with business performance.

2.1 Direct Causal Relationship Approach

Some researches focus on isolated cause-and-effect relationships between various business parameters and performance. They examine particular aspects e-commerce strategy and try to estimate their effect on performance. The starting point for many studies is to identify the type of business with respect to the online environment.

For instance, the level of online integration and industrial specification are considered as major forces affecting performance (Lin and Chen 2009). Based on the website functionality and the availability of a physical store they identify three main levels of online integration: an information provision model (almost no integration, the online channel provides only information), a pure Web model (the typical “dot.com” companies without actual offline channel) and a hybrid model (with a high degree of integration between online and offline activities) (Lin and Chen 2009). Based on this model, Lin and Chen (2009) prove that there exists a specific relation between the level of integration, industrial characteristics and the business performance.

Indeed, this view is supported by other scientists who analyze channel integration with respect to different marketing activities. There exists a vast amount of literature discussing how typical offline activities such as segmentation (Jayawardhena, Wright and Dennis

(11)

11

Direct Causal Relationship Approach

Definition Paper Specifics Author

This approach is refered to as Direct Causal Relationship approach, because it recognizes the direct cause-and-effect relationship between offline/ online and business

performance. The authors try to measure the strenght of this relationship and to build theories based on these findings. Most of the researchers consider only one influencing element

Develop a measure of the level of online integration and industrial specification an prove that there exists a relation between the level of integration, industrial characteristics and business performance.

Lin and Chen 2009

Examine the effect of segmentation in online environment oppose to the offline market

Jayawardhena, Wright and Dennis 2007; Doherty and Ellis-Chadwick 2003 Compare the online and offline brand

loyalty. Prove that in the online commerce additional variable such as the brand share affect highly brand loyalty andthus business performance

Danaher, Wilson, Davis 2003

Study the effect of branding on business performance

Degeratu, Rangaswamy and Wu 2000 Examine the influence of sales process

and distribution on business performance

Forman, Ghose and Goldfarb 2009

Effect of sales process and distribution business performance

Hitt and Frei 2002

2007; Doherty and Ellis-Chadwick 2003), branding (Danaher, Wilson, Davis 2003; Degeratu, Rangaswamy and Wu 2000), sales process and distribution (Forman, Ghose and Goldfarb 2009; Hitt, and Frei 2002) are influencing online performance. All these activities have a crucial role in the traditional offline business model, therefore it is important to adapt them correctly to the new online environment. The online channel has created a new type of “empowered” customers who do not fit in the old segmentation models (Jayawardhena, Wright and Dennis 2007) and require different communication and branding strategies (Danaher, Wilson and Davis 2003). Table 1 summarizes the concept and the literature sources on Direct Causal Relationship Approach.

(12)

12

2.2 Indirect Benefits Relation Approach

The Indirect Benefit Relationship Approach measures the company‟s performance by scaling it to the benefits sought by either consumers or businesses. It analyzes the online activities and their contribution to the business process. This contribution is perceived in terms of achievement of intermediate objectives, which could be related strictly to the business process (i.e. reduced costs, improved communication) or might represent consumers‟ gains (i.e. lower search cost).

2.2.1 Business-Centric Approach

Considerably large amount of scientific papers investigate the online activities in terms of the indirect value added to the company. This approach is widely propagated because of accountability issues. The economic value added through the online channel has an obvious effect on business operations. It is easier to relate an online campaign to online traffic than to actual sales because traffic is the direct outcome of a particular business action while the sales might be influenced by other variables.

Many authors examine online initiatives in a sense of reduced costs (Steinfield, Bowman and Adelaar 2001; Dazhong, Gautam, Xianjun, and Whinston 2004; Brynjolfsson, and Smith 2000, Lee 2001), value, added via differentiation (Steinfield, Bowman and Adelaar 2001; Clemons, Hann, and Hitt 2002) and geographic/product market expansion (Steinfield, Bowman and Adelaar 2001). The technological aspect has also received scientific attention. Some authors highlight on the positive effect of ICT on company internal and external communication and product customization (Vries 2006), and location-awareness (Fielt 2000). They claim that the interface features need to be compatible and user-friendly in order to produce a positive effect of performance (Baty & Lee, 1995; Hoffman, Novak, & Chatterjee, 1995; Jarvenpaa & Todd, 1997, Lohse & Spiller, 1998; Ridgon, 1996).

A more general model of economic benefits from online commerce is developed by Lee (2001). He provides a framework, in which transaction theories and switching costs are considered as the “critical factors of success.” The model explores the economic perspective of online commerce and redesigns the old revenue and cost models to fit the new environment. This analysis is a useful tool for developing business strategies and

(13)

13

monitoring performance but it does not consider the many other non economical aspects of the online-offline dynamics.

2.2.2 Consumer-Centric Approach

On the contrary, the consumer-centric approach scales the performance through the customers‟ point of view. A lot of authors strive to evaluates effect of Web practices on consumers‟ searching efforts (Wu, Ray, Geng and Whinston 2004; Johnson, Moe, Fader, Bellman and Lohse 2004), price sensitivity (Chu, Chintagunta and Cebollada 2008; Donthu and Garcia 1999; Degeratu, Rangaswamy and Wu 2000) and technology acceptance (TAM, by Venkatesh and Agarwal 2006; Gefen and Straub 2000). Their main goal is to uncover which features and functions influence positively on customer behavior and satisfaction.

The studies reveal that on the Web the e-consumer is more price sensitive due to the lower search costs and the easier access to information on the Web (Wu, Ray, Geng and Whinston 2004, Degeratu, Rangaswamy and Wu 2000). However, sometimes the Web places obstacles in the form of new, complicated technology or complex applications which resent consumers from online markets (Venkatesh and Agarwal 2006).

Althought the literature enhances both approaches evenly, the modern business practice is more inclined towards the consumer-centric view. The main implication, carried by the consumer-focused view was introduced in the end of the 20th century by Philip Kotler (2006).

He stipulates the importance of the consumer in the “globalized economy” characterized by virtualization and “hypercompetition”. The online marketplace empowers consumers and gives them an unlimited access to information and technology. In this new environment, not the organization, but the individual has the dominant role (Kotler 2006). Moreover, this consumer perspective is more comprehensive because it probes deep into the marketing theory and presents not only economical but also physiological and demographic factors.

This approach provides a precise view of the situation since it focuses on the human behavior and perceptions and translates them into a market value. Table 2 contains the definition and main literature sources on Indirect Benefit Relationship Approach.

(14)

14

Table 2: Literature on Benefit Relationship Approach.Source: own illustration.

Indirect Benefit Relationship Approach

Definition Main Types Paper Specifics Authors

The Indirect Benefit Relationship Approach is focused on the diverse set of benefits gained by the online channel. Opposite to the previos approach, these benefits are not direct financial or performance measures, but other paramenters which are supposed to boost performance. For instance, reduced transaction cost, differentiation, improved communication, or higher customer satisfactiona are such intermediary targets Business respective takes into account the effect of online features on business related beefits. Such are the value added via differentiation, cost reduction, internal communicatons etc. Online initiatives in a sence of reduced cost geographic/product market expansion, and value, added via differentiation Steinfield, Bowman and Adelaar 2001; Dazhong, Gautam, Xianjun, and Whinston 2004; Brynjolfsson, and Smith 2000;Clemons, Hann, and Hitt 2002 Focus on the positive

effect of ICT on company internal and external communication and product

customization

Vries 2006; Fliet 2000

Interface features need to be compatible and user-friendly in order to establish an efficitent working process

Baty & Lee, 1995; Hoffman, Novak, & Chatterjee, 1995; Jarvenpaa & Todd, 1997, Lohse & Spiller, 1998; Ridgon, 1996 Consumer perspective considers the benefits recognized by the customers, i.e search efforts, price sensitivity, overall satisfaction etc.

Develops a new metrics which reflects the economic benefits such as lower trasnaction and opportunity costs and gained by the e-channel.

Lee 2001

Examine the effect of online elements of consumers‟ searching efforts

Johnson, Moe, Fader, Bellman and Lohse 2004

Study how the online environment influenceon consumers„ price sensitivity Chu, Chintagunta and Cebollada 2008; Donthu and Garcia 1999; Wu, Ray, Geng and Whinston 2004 Prove that user-features

influence positively on customer behavior and satisfaction

Venkatesh and Agarwal 2006

(15)

15

2.3 Generalized Frameworks

All of the discussed researches provide particular insights into the e-commerce theory and practice. However, they focus only on particular parameters which highlight the specific aspect. Thus, they fail to deliver a 360 degree view of the online business practice. On the other side, the main subject of this paper – e-performance measurement model – requires a complete overview of the forces and effects on the e-marketplace. Indeed, few authors try to enhance a more comprehensive view of e-business processes. For instance, by Lages, Lages and Rita (2004) present a broad model of online activities and factors, and their effect on performance. They build a conceptual framework that connects five major factors – internal forces (centralization, bureaucratization, innovation etc), external forces (market structure, competition etc.), past web and firm performance, and e-marketing strategy in terms of the “4Ws” - Web-Design, Web-Promotion, Web-Price and Web-CRM, (Lages, Lages and Rita 2004). They include the latest because of they affect positively current business practices and profitability and improve customer purchasing experience (Strauss and Frost 2001 as cited by Lages, Lages and Rita 2004). The model enhances the essential business forces and provides a general overview of the theory but does not probe into details. Lages, Lages and Rita (2004) suggest further researches on the link between web parameters, current and past and performance.

The idea of the Balanced Scorecard (BSC) model (Kaplan and Noman 1992; Marr and Shiuma 2003) significantly resembles the model of Lages, Lages and Rita (2004). It is a diverse set of measurements including operational measures of customer satisfaction, improvement and innovation activities as well as financial measures.1

BSC is described as an influential and dominant tool for measurement and controlling functions (Barnes and Hinton 2007). The BSC could be used as a ground for development of a new metrics which enhances diverse parameters representing financial, strategic and organizational policies (Barnes and Hinton 2007). This metrics need to consider the major forces shaping e-business performance: website performance (Zeithaml et al., 2000; Barnes and Vidgen 2001 as cited by Barnes and Hinton 2007), business process performance

1

(16)

16

(Hinton et al., 2003; Wu et al., 2003 as cited by Barnes and Hinton 2007), customers‟ performance (Hinton et al., 2003; Wu et al., 2003; Minocha et al., 2004; Voss, 2003 as cited by Barnes and Hinton 2007) and the linkage between e-business performance to business strategy (Chang et al., 2003; Porter, 2001 as cited by Barnes and Hinton 2007).

The BCS and the 4Ws Model by Lages, Lages and Rita (2004) refer to the web strategy as a powerful force in shaping business performance. Both models suggest a deep and thoughtful analysis of the website features and functions in pursuing a better performance measurement.

Table 3 presents the literature on various measurement models.

Table 3: Literature on generalized e-measurements. Source: own illustration

Measurment Models

Model Paper Specifics Authors

4Ws” - Design, Promotion, Price and Web-CRM

Develop a conceptual framework that connects five major factors – internal forces), external forces, past web and firm performance, and e-marketing strategy. Enhance the complete business process but do not go into details

Lages, Lages and Rita (2004)

Balanced

Scorecard (BSC) model

Develop a diverse set of measurements including operational measures of customer satisfaction, improvement and innovation activities as well as financial measures.

Kaplan and Noman 1992, Marr and Shiuma 2003

e-performance measure

Examine the usage of an e-performance measure and the different options for a sustanable model. Discuss BSC and its application in the e-business. Prove an urgent the urgent need of relevant measure

(17)

17

2.4 Website Metrics

The above models strongly highlight the role of the website as a major tool in company‟s online operations. Indeed, the importance of the website is recognized by many researchers. Recent studies proved that the website is the public face of the organization and bears considerable responsibility for business and customers‟ performance (Palmer 2002; Hwang, McMillan and Lee 2003; Haubl and Trifts 2000; Lohse and Spiller 1999). It reveals the company‟s online strategy and involvement (Hwang, McMillan and Lee 2003). The web features delineate the corporate planning and objectives and facilitate the communication between the company and its stakeholders (Palmer 2002).

The website is perceived as a tool for communication and promotion (Hwang, McMillan and Lee 2003). It proved to be an influential instrument for shaping consumer experience by exploitation of interactive web tools and rich media (Haubl and Trifts 2000). There exists a relationship between “web-based shopping environments” and financial performance (Lohse and Spiller 1999).

Being such a powerful driver of the company‟s performance, the corporate website has been a subject of many studies. There exist various theories about website functionality and performance. For instance Palmer (2002) offers several major dimensions of the website – “usability, design and performance metrics, including download delay, navigability, site content,interactivity, and responsiveness”. These dimensions represent the consumer perspective and prove to be tightly related to the overall business performance (Palmer 2002). Another categorization of website features is introduced by Agarwal and Venkatesh (2002) who develop five groups regarding the website usability – content, ease of use, promotion, made-for-the-medium, and emotion. A different perspective is undertaken by Eighmey (1997), Ghose and Dou (1998), Keeney (1999), and Nielson (1993), who suggest informativeness, interactivity, entertainment and order as the main benefits which the online user seeks for.

These categorizations reveal particular similarities; however, the two approaches still show some major differences. This proves that such methodologies could be highly subjective or incomplete. Although, the literature does not present a sustainable model for website metrics, it proves the website as a central element in virtual markets. This paper

(18)

18

follows this theory and exploits the website as a representation of companies‟ online strategy. Table 4 summarizes the theory review on website metrics.

Table 4: Literature on website metrics. Source: own illustration

Website Metrics

Website Metrics Papers Specifics Authors

The website gains considerable scientific attention. Researchers examine its

characteristics,

functions and effects of customers„ experience and on business performance. Many studies have

Website, defined as the public face of the organization which bears considerable responsibility for business and customers‟ performance

Palmer 2002; Hwang, McMillan and Lee 2003; Haubl and Trifts 2000; Lohse and Spiller 1999

Website as a reflection of the company‟s online strategy and involvement

Hwang, McMillan and Lee 2003

Website as a tool for

communication and promotion, for shaping customers„ experience

Haubl and Trifts 2000

Differentiates several major dimensions of the website with respect to the usability, design and performance

Palmer 2002

Categorize website features is introduced by Agarwal and

Venkatesh (2002) who develop five groups regarding the website usability – content, ease of use, promotion, made-for-the-medium, and emotion.

Agarwal and Venkatesh (2002

Webite features related to informativeness, interactivity, entertainment and order are the main benefits which the online user seeks for.

Eighmey (1997), Ghose and Dou 1998, Keeney 1999, and Nielson 1993

There exists a relationship between “web-based shopping

environments” and financial performance

Lohse and Spiller 1999

(19)

19

2.5 Consumer Behaviour and Business Performance

After the revision of the general approaches to the e-metrics and identification of the website as a main measurement unit, this section proceeds with reviewing the theories which investigate the concept of business performance.

The ultimate goal of each business relates to profits. In order to understand what drives profits and how online strategy fits in this process it is important to enhance the customer view and to realize the main urges behind the actual purchase.

The purchasing process is presented in the literature as a combination of psychological, behavioral and marketing stimuli (Kotler 2006). This idea not new to the e-commerce literature. Some authors already tried to establish the Theory of Planned Behavior (Pavlou and Chai 2002).

To explain in details its reasoning it is necessary to mention the classical studies on conscious human behavior. The starting point for these studies is the theory of reinforcement behavior (Watson 1913-1930; Pavlov 1920; Tolman 1932). The traditional studies suggest that human actions are responses to conditional and unconditional stimuli (Watsonian Behavioralism, Watson 1913-1930; Classical Conditioning, Pavlov 1920). Later, this theory is augmented by Tolman (1932), Additional intervening variables such as emotions, attitudes, perceptions and preferences are introduced by Tolman (1932), (Psychological neobehaviorism, 1930-1960). These theories stipulate the basic factors in human behavior and therefore are applicable in explaining the purchasing process.

2.5.1 Belief Reinforcement Model (BRM)

The modern theories develop different models which present the purchasing process in the setting of the online commerce. For instance, the Belief Reinforcement Model (BRM) synthesizes the theory of planned behavior with some aspects of social psychology, consumer behavior and endeavors to relate them to categorized Web-design elements in order to create a theoretical model of “Web shoppers‟behavior” (Song and Zanedi 2005). This model measures the impact of the website elements on the online purchasing behavior.2 It highlights the significant dependence of business performance on human

2

(20)

20

psychology. However, this model is not completely reliable due to the limited sample size comprised of “young and skilled segments of the Web-customer population”, (Song and Zanedi 2005). Such a limitation suggest that a broader research could produce other results since the personal factors such as gender, age, skills, culture etc, have a decisive role of such measurement, (Venkatesh 2003). Another weakness of the BRM model is that it analyses the consumer behavior in general without a particular emphasis on the purchase. However, for precision and accountability reasons, it is vital to focus exclusively on the purchasing process and its phases.

2.5.2 Kotler’s Five-Step Purchasing Model

Opposite to BRM, the Kotler‟s Five-Step Purchasing model examines the buying behavior in details, differentiating several consecutive phases. This model depicts the five major steps of decision making process of turning a prospect into an engaged customer. The general steps are: problem recognition (the prospect becomes aware of his/her problem), information search (he/she start seeking for a solution –particular product but no brand preferences), evaluation of alternatives (he/she has already developed a choice set of definite brands, and start to evaluate and compare the options), purchase decision (he/she decides on particular option and makes the purchase) and post-purchase behavior (possible recalls, maintenance requests, further purchases, positive/negative recommendations).

At each of the stages the consumer is influenced by various internal and external factors.

These factors are decisive whether the consumer will reach the next stage. The BRM model could be used to identify these factors. According to BRM the problem recognition is usually initiated by some external and internal stimuli. The external forces could be economical or social, while the internal are rooted in the human nature and could be best represented by the Maslow‟s hierarchy of human needs.

The next steps of information search, evaluation and purchase intention are affected mainly by subjective factors such as attitudes, beliefs, preferences and perceptions as well as by other opinions and recommendations. Some external, unanticipated agents may Figure 1: Five-step purchasing model. Source: adapted from Kotler (2006).

(21)

21

impact the decision. The post-purchase phase is the moment of reconsideration of the complete process and evaluating the overall experience. If the experience was positive it might reinforce another purchase or recommendation. Otherwise, the purchase will not be repeated. Thus, the Kotler‟s model explains the decision making process of a prospective buyer and provides valuable insights to businesses.

2.5.3 iPACE Model

A similar model representing the web-purchasing process is the more recent iPACE, developed by Hanson and Kalyanam (2006). It is a framework for online performance focused mainly on the companies‟ website. The name, iPACE, stands for the “basic shopping needs” of the online customer – Information, Price, Assortment, Convenience and Entertainment. These are the key points of online business models which should be considered by the online retailer in order to identify the “right approach for the right channel” (Hanson and Kalyanam 2006). The authors claim that organizations have to identify and understand these “shopping needs of the target consumer for a particular product category” and to provide “online features and functionality that support” them (Hanson and Kalyanam 2006).

The phases of the purchasing process delineated by iPACE are slightly different from those in Kotler‟s model. iPACE discerns the following phases - product search, information search, alternatives evaluation, order placement and product returns. The distinctive feature of this model is that it discusses website features and the online business activities which affect each of the phases.

For instance, at the first step - finding the right product, the e-retailers should facilitate the search and to help the shoppers to narrow their selection. At the second step of acquiring information, the companies should focus on overcoming the physical limitations of the online shopping via extensive product descriptions and a variety of options. Rich media, including images, zoom, ratings and reviews, virtual models and diagrams is especially important. The next step of evaluation could be propagated via different shopping tools, such as a comparison matrix or assistant. At the moment of order the businesses should enable complete transparency and safety, so the consumer would feel comfortable and secure. The post-purchase step includes notifications such as order confirmation, total price, shipping date and expected delivery date. It should also facilitate the returns, technical support training, etc.

(22)

22

iPACE is valuable in addressing the main challenges faced by retailers working across online and traditional channels. It considers the needs of the hybrid customer who is willing to shop online but buy at a traditional store, and vice versa. This model helps in decision such as online/offline product mix, communications and logistics, pricing discrepancies, planning consistent promotional offerings etc (Hanson and Kalyanam 2006). Figure 2 depicts the main stages of the online purchasing process combined with the online factors of influence.3

3

For the original iPACE framework see Appendix 4.

(23)

23

Table 5, summarizes the literature on consumer behavior and e-commerce

Table 5: Literature on consumer behavior in online and offline environment. Source: own illustration.

Consumer Behavioral & E-commerce Studies

Model Model description and contribution Authors

Watsonian Behavioralism Classical Conditioning

Human actions are responses to conditional and unconditional stimuli

Watson (1913-1930) Pavlov (1920) Phsychological

neobehaviorism

Additional stimuli such as emotions, attitudes, perceptions are added to the previous models

Tolman (1932) Belief Reinforcement

Model (BRM)

Highlights the significant dependence of business performance on human phsychology

Kaplan and Noman 1992; Marr and Shiuma 2003 Five-Stage Purchasing

model

Describes the purchasing process in five major steps - problem recognition, information search, evaluation of alternatives, purchase decision and post-purchase behavior

Kotler 2006

iPACE Define purchasing process and its drivers for the e-envorinment. The name, iPACE, stands for the basic shopping needs of the online customer – Information, Price, Assortment, Convenience and Entertainment

Hanson and Kalyanam 2006

(24)

24

3 CONCEPTUAL MODEL

Based on the prior literature and the modern e-commerce implication, this section presents a new model which reflects the specifics of the classical consumer behavior in the online environment. The new model is a combination of BRM, iPACE and Kotler‟s Five-Step Purchasing Model. It unifies their main characteristics such as behavior theory approach (BRM), detailed buying process (Kotler 2006) and website features categorization (iPACE). The new model unifies in an extended and clarified framework for online measurement. The new model consists of three main steps – Pre-purchase, Purchase and Post-purchase. Kotler‟s terminology would be used to describe the sub steps.

Figure 3 summarizes the entire conceptual framework of the Three-Step Purchasing Model built upon the classical theory of human behavior, enhanced by BRM, combined with the traditional marketing view of Kotler and the modern e-commerce perspective brought by iPACE. This model focuses on the consumers‟ buying experience from a business perspective, i.e. examines how companies could better satisfy customers‟ needs and thus boost their online and offline performance. This model is appropriate for this purpose because it highlights the general needs of the online shoppers – information, price, assortment, convenience and entertainment and examines the alternative options for companies to satisfy them.

(25)

25

Figure 3: Purchasing model derived from consumer behavioral studies. Source: own illustration.

3.1 Pre-Purchase Phase

The Pre-purchase phase describes initial steps of the consumer toward the purchase. In this phase, the consumer becoming aware of his own problems and the solutions, offered on the market. At this phase it is important to present nicely the solution to the potential customers and to stand out among the competitors. The Pre-purchase phase enhances several sub steps (see Figure 3).

(26)

26

3.1.1 Problem Recognition

This step starts with problem recognition which is induced by certain stimulus - external influences (social, economical), and internal needs - physiological, safety, social, esteem, self-development (Maslow 1954, see Appendix 5). The external influence might be provoked by some general economic factors, market changes, income fluctuations etc. There could also be a social influence from different public entities, organizations, media, social groups etc. These forces are considerably volatile and uncontrollable by external entities (i.e. companies). The internal forces are rooted deeply in the human nature. They are described by Maslow (1954) in his theory about the human needs.

Figure 4: Pre-purchase phase - Problem Recognition. Source: own illustration.

The basic needs could hardly be affected by companies. However, most of the business actions are directing the higher level of self-esteem and self-actualization needs. At this stage the companies‟ main aim is to create awareness. This could be done via banner ads, pay search marketing, email campaigns, as well as via the offline channel. At this stage the company‟s website does not play an essential role.

(27)

27

3.1.2 Information Search and Evaluation of Alternatives

Problem recognition arouses wants and then the conscious information search begins. During this phase consumers actively look for information, compare and evaluate different offers. Kotler (2006) differentiates the phases of information search and evaluation of alternatives (see Figure 5). However, in this model they are analyzed together since the online environment stimulates both phases simultaneously.

The website provides highly efficient instruments to support this phase. Some websites provide extensive information about business propositions accompanied by rich media, presentations, advices from specialists, and credentials. The information content of the website is highly important (Song and Zanedi 2005). The product information could be presented various forms and styles, with different degree of involvement. Images, videos, graphical and tabular representations increase information visualization and the ease of use (Hu et.al. 1999, Benbasat and Dexter 1985, Spool et al. 1999, as cited by Song and Zanedi 2005).

An essential tool at this stage is the exploitation of search and comparison options on the website (Hanson and Kalyanam 2006). Haubl and Trifts (2000) refer to such interface features as “interactive decision aids” and confirm their importance for the searching process. Enhancement of the social media is another method to facilitate the pre-purchase phase. Public forums, ratings, blogs are widely used to provide information and establish trust (Hsu and Chiu 2004; La Tour and Manrai 1989).

These instruments not only facilitate consumers to acquire information, but also they influence their decision (Bearden and Rose 1990; Calder and Burnkrant 1977). This is the phase in which they form their attitudes and perceptions toward the brand (Haubl and Trifts 2000). Communication with other consumers and with the company has a significant impact on their purchasing intentions.

(28)

28

The website is an efficient medium between the company and the prospect during the information search. The website features used in this step could be grouped in the following categories - image, information, navigation (including search options) and recommendation (including company‟ advices, social media, etc). The above discussion suggests that all these categories have positive effect on the customer and facilitate the conversion process, i.e. provoke more purchases. Therefore, the Hypothesis 1 states that the above categories are positively related to company‟s online and offline performance.

H1: Categories in the Pre-purchase phase – Image, Information, Navigation and Recommendation - positively affect business performance.

3.2 Purchase Phase

The pre-purchase phase is usually followed by forming a purchase intention and consequently a purchase decision (see Figure 6). The most efficient actions, typical for the online commerce are various promotional campaigns, discounts, affiliate programs etc. The web environment stimulates such activities, due to their virility and effectiveness. Hence, promotion is one of the main categories of website features which affect consumers in the purchase phase.

Another important category is facilitation. The purchase decision is highly dependent on convenience and trust established by the e-channel. The customers pursue different things by buying online. Some find it efficient because of the time/money savings, other strive for experience (e.g. online auctions) etc., (Darian 1987). However, all users insist on a fast, Figure 5: Pre-purchase phase – Information Search & Evaluation of Alternatives.Source: own illustration

(29)

29

easy and secure process. Consequently, the website features related to payment options form a significantly category.

Many users do not feel comfortable with buying online (Hanson and Kalyanam 2006). They prefer to search for information on the Web but to make the purchase at the offline store (Choudhury and Karahanna 2008; Verhoef, Neslin and Vroomen 2007). They might convert easier if the website contains information on store location, inventory check etc. For them the online-offline integration is a key factor to convert.

Consequently, it could be assumed that all the categories – promotion, usability, integration and payment are positively related to performance. However, the facilitation and payment feature have higher impact on online sales and online conversion rate because they ensure easier and safer online purchasing experience while the integration features would have higher effect on offline performance since they provide more information about offline outlets and induce customers to shop offline.

H2: All categories in the Purchase phase – Promotion, Facilitation, Integration and Transaction positively affect business performance (online and offline).

H2 a: Facilitation and Payment exercise a high positive effect on online performance (online sales).

H2 b: Integration exercises a high positive effect on offline performance. H2 c: Promotion has a similar effect on online and offline performance.

(30)

30

3.3 Post-Purchase Phase

After the purchase, consumers, based on their experience with the website and the product impression develop their post purchase attitude. Pleasant experience would induce further purchases and would create a loyal customer, while the negative feelings could result in bad word-of-mouth. Therefore, this phase emphasizes mainly on customers‟ experience created by the purchasing process (see Figure 7).

The website could offer different post-purchase functions such as account history, order confirmation and status, delivery tracking, return and repair options, etc. It could also provide an easy call back function and could involve some CRM (Customer Relationship Management) activities which could induce other purchases.

Hence, the post-purchase services promote long-term customer relations and induce future profitability. Thus, they are key drivers of the long-term business success.

H3: Post-purchase services are positively related performance.

(31)

31

3.4 Summary

In summary, the Pre-purchase phase which enhances information search and evaluation is represented by features categorized as Image, Information, Navigation and Recommendation. The Purchase phase, consisting of purchase intention and decision and implementation is described by Promotions, Facilitation, Integration, and Payment. The Post-purchase phase is represented by one main category – Post-purchase services (Status Check, Return and Repair services and CRM).

Figure 8: Three-Step Purchasing Model Phases. Source: own illustration.

At the Purchase phase consumers face the choice between online and offline purchase. Website feature affect differently this decision. The facilitation and payment features are closer related to the online, while the integration is closer to the offline performance. An additional assumption states that the online and offline performance are positively related. Indeed, many authors point out the business benefits of the multichanneling and exploitation of hybrid models (Steinfield, Adelaar and Bowman 2001; Buklin, Ramaswamy and Majumdar 1995). This assumption is the basis for the last hypothesis:

H4: The online and offline performance are positively related.

The following illustration (Figure 9) represents the complete framework. It includes the website features extracted from one of the models used in this paper as a benchmark metric system. The analogous methodology will be applied to the other model to achieve a compatible framework.

Pre-Purchase Phase

Image Information Navigation Recomendation

Purchase Phase

Promotion Facilitation Integration Transaction

Post-Purchase Phase

Status Check Return and Repair

(32)

32

Figure 9: Conceptual model and hypotheses. Source: own illustration.

The hypotheses stated in the model are summarized in Table 6.

Table 6: Hypotheses – Purchasing process. Source: own illustration.

H1 Categories in the Pre-purchase phase – Image, Information, Navigation and Recommendation - positively affect business performance (online and offline)

H2 All categories in the Purchase phase – Promotion, Facilitation, Integration and Transaction positively affect business performance (online and offline)

H2 a Facilitation and Payment exercise a high positive effect on online performance (online sales) H2 b Integration exercises a high positive effect on offline performance

H3 Promotion has a similar effect on online and offline performance.

H2 c Post-purchase services affect positively business performance (online and offline) H4 Online and offline performance are positively related

(33)

33

This framework enlightens the consumer behavior throughout the purchasing process. The model captures the specifications of the website as a shopping venue. However, several other considerations need to be introduced in order to enhance the entire online market environment.

An important business characteristic is the level of online integration (Liang, Lin and Chen 2009). The pure Web companies concentrate entirely on their online activities and performance. They construct their website in a way which provides the complete customer support and guidance. In contrast, the mixed (hybrid models) might offer only a few functions on the website in order to stimulate other channels‟ performance. The hybrid models need to consider certain synergies and losses of the cross channeling (Lin and Chen 2009; Steinfield, Adelaar and Bowman 2001; Buklin, Ramaswamy and Majumdar 1995; Verhoef, Neslin and Vromen 2007).

Prior studies have proved a relationship between the type of business model and companies‟ performance (Liang, Lin and Chen 2009). Therefore, in this research two main types are discerned – pure Web vs. hybrid models. These two types represent the extremes, since all companies in the sample have at least a simplified website. Usually the hybrid models collect revenues from a variety of business channels, while for the pure Web models rely entirely on the e-channel. Therefore, it is logical to assume that the pure web model is a predicator for higher online sales in comparison to the mixed businesses. This is presented in Hypothesis 5:

H 5: Pure web businesses have better online performance than mixed (online – offline) models.

Furthermore, the type of industry has also revealed a significant influence on the performance (Liang, Lin and Chen 2009; Swaminathan, Lepkowska-White and Rao 1999). Different product characteristics affect the online business performance. Online companies from the automobile industry experience different problems from the food e-retailers. The companies take into consideration the specifications of the industry in planning and monitoring their online activities. The reflection of this differentiation is obvious on the website. Therefore, the Hypothesis 6 states:

(34)

34

In this research the industries are defined by the generic product characteristics. Five main industries are defined – Apparel, Computers and Electronics, Food and Drugs, Specialties and ass merchants. Apparel is combined with Sporting Goods since both types represent personal items which usually have to be tried on before the purchase. Moreover, they are both seasonal goods which could follow certain fashion. Presumably, the image and promotion features are the most influential drivers of performance for these industries. Computers, electronics and other hardware have many technical characteristics which are important for the buyers. It is important to clarify them to the online customers and to facilitate them in making their choice. Hence, comparison and recommendation features as well as technical information are supposed to have positive influence on performance.

Mass Merchants require a broad array of features and function due to the diversity of their products and target groups.

Specialty goods include flowers, gifts, jewelry, books other kinds of small luxurious items bought to satisfy higher level needs (Maslow 1954). Online customers who buy such items insist on easy and secure payment, efficient (on time delivery) and promo offers.

Food and Drugs are everyday goods. In most cases they are generic and the online consumers value the reliability and easiness of the service. The important aspect in this case is the immediate delivery, easy payment and many facilitation features. Based on the specifics of different industries, additional hypotheses could be stated:

Table 7: Hypotheses – Merchant type and industry. Source: own illustration.

H5 Pure web businesses have better online performance than mixed (online – offline) models. H6 The industry type has an effect on performance.

H6 a _{Image features have high positive impact on performance in Apparel industry.}

H6 b Information and recommendation features have high positive effect on performance in Computers&Electronics industry.

H6 c _{Payment, delivery tracking and promotion have a high impact on performance in selling Specialty goods.} H6 d Payment and facilitation have a high effect on Food&Drugs industry.

(35)

35

Based on these hypotheses the final aim of the conceptual model is to illustrate the interrelation between the main elements – the website, online and offline performance, and other external factors such as offline strategy.

External factors Industry Merchant type Brand Equity Company Size Technology Geography …

(36)

36

4 METHODOLOGY

The literature review showed that many authors recognize the effect of website features are on business online performance. This research approaches the problem form a different perspective. It applies the conceptual model, developed in the previous section and distributes the website features into categories defined by the purchasing phases.

The conceptual model visualizes how the online strategy, embodied into the company‟s website stimulates consumers on their way to finalize a purchase. The model presents the purchasing process as a sequence of steps, each of which is susceptible to different influences. Each step has a significant effect on performance. In addition, several firms‟ characteristics are considered as possible triggers of performance. Thus, the complete picture on the online-offline dynamics is captured by the analysis.

One of the goals of this paper is to analyze the effect of each phase. Another goal is to reveal which website characteristics influence each of the purchasing steps. These results would be sufficient to build a framework which relates website characteristic to companies‟ profits. Such a model could be used as a metrics for online business activities.

The actual analysis was performed in two steps. First, two independent website metric models were analyzed and compared. The aim was to measure the impact of each phase and to identify the web features which most influence on performance. The second step of the analysis was to analyze the influence of different industries and merchant types on performance, using only one of the models (Model 2, since it contains large number of observations, coded with concise headings).

4.1 Data

Two datasets were used to test the hypotheses and to validate the model. These datasets are based on two independent website metric models. The models list various website features and classify them into several broad groups. In practice, a website contains a large number of features and functions which forced the authors to select only those which are considered to be meaningful for the companies‟ performance.

The preliminary review showed the authors used different sets of website features. Both models contain comparable data which covers similar websites over the same period of

(37)

37

time. The models are compared in order to identify which is more efficient at predicting performance, i.e. which contains more relevant set of website features.

4.1.1 Model 1

Model 1 is designed by Adelaar (2007). It contains website features from online and offline retailers which have a commercial web presence. This model is considerably profound and explanatory. For most of the features an extensive narrative descriptionis provided in order to minimize confusion. It contains 73 website features clustered in 10 functional groups (see Appendix 6).

4.1.2 Model 2

Model 2 is developed by the Internet Retailer Top 500 Guide, 2008. It has a similar structure – it lists website features in several categories. This model is less specific than Model 1 – it clusters features into three broad groups - features & functions, payments and customer services (see Appendix 7). However, the coding of separate variables is more definite and clear. Model 2 contains 500 observations – the Top 500 Retailers in terms of revenue for 2008.

Model 2 concentrates mainly on the pre-purchase phase by listing features which facilitate promotion activities and information search. It also lists many possible payment methods. The company information and channel integration elements are relatively few. Probably, the developers of the second model did not consider them as meaningful for the business performance.

4.1.3 Initial Settings

First part of the analysis consists in comparing and discussing the two datasets. In order to compare the two models the datasets were adjusted. The data scrutiny revealed that the both models contain only 96 common results which could be used in this analysis. The original idea was to derive a set of features which are listed by both models. This set would have been treated as a primary set of features which affect companies‟ performance. After estimating the effect of this set, the other variables would have been added to the analysis in order to identify their incremental influence on performance. This would have given a clear overview of the importance of each feature. The basic assumption behind this plan was that the two models contain compatible datasets. However, a detailed analysis revealed

(38)

38

significant differences in the two datasets. The models presented different results for the same websites. Approximately 50% of the results are not compatible. The possible explanation might be the high volatility of the online venue. The website is not a static environment. Most of the companies maintain and develop their websites on daily basis. The timeline of both researches comprehends a period of one year. Another reason might be the high degree of subjectivism in coding and data-mining. Since websites are comparatively new phenomena most of the definitions of their features are constructed by the authors of the models. Hence, there could be many cases in which the match between two models is simply impossible due to their inherent characteristics.

Although unsuccessful, this first attempt to compare the models provided a deep insight into the nature of the e-market environment. Being elusive and subjective, it requires a careful and comprehensive approach which is not based only on statistical results but rather on intangible considerations. This is another proof of the urgent need for a decent e-commerce metrics.

After the initial examination of the data the direction of the research was readjusted. The features listed in the models were grouped in identical functional groups. These groups were clustered into higher-level categories. These categories were used as independent variables in the regression. Both models were examined via a hierarchical regression analysis and the results were compared.

The further analysis aimed to distinguish the effect of the website features in different industries. For this purpose, the observations were grouped according to the industry and each group was subjected to a separate regression analysis. Thus, the differences between the industries were observed. Since the models have only 96 common observations, additional clustering would have lead to considerably small samples and unstable regression results. Therefore, only Model 2 was used in this step. Its observations were clustered in 5 major groups – Apparel (Apparel and Sporting Goods), Computer & Electronics (Computer & Electronics and Hardware), Mass Merchant, Specialties (Books, Jewelry, Flowers&Presents, Heatlth&Beauty) and Food&Drugs. This grouping was made based on the characteristics of the products and their online presence.

(39)

39

4.2 Independent Variables

This grouping of was performed in order to reduce the large number of website features, serving the same or similar functions. Although, some of these features are correlated, there is no reason to delete them from the analysis (as a standard statistical procedure would suggest). The point of the research is to identify how the separate features perform a particular function which promotes performance. The grouping of features would imply that not the single features but the number of similar features on a website affects online success, i.e. a website with many image-related features is supposed to have a higher sales rate than one with scarce imaging options.

It is important to state, that there are many options for grouping variables. One is to use hierarchical cluster analysis, or to apply a stepwise regression analysis (Lohse and Spiller 1999). The problem with these methods is that they do not consider the logic behind the conceptual model. They group variables based on common variance explained, on their correlation etc. However, in such grouping would totally undermine the entire framework built on the prior theory. The idea of this research is to present a new model and to test it.

This involves grouping variables in the predetermined categories and examining the influence of these particular categories on performance. Using statistical clusters would be a backward approach, since would let the computer program to create clusters and then the researcher has to step back and to find the right interpretation. This is not a scientific approach; therefore in this research the variables are grouped based on assumptions stemmed by the theory review and the conceptual model.

The initial groups provided common coding for the website features but were too many for a regression analysis. Therefore, these groups were clustered around more general functions. These higher-level functional clusters were used as independent variables in the analysis. A further step attributed each of these variables to a particular purchasing phase. Thus, the analysis would enhance the effect of the single variable and the entire purchasing phase.

Website Features and Functions as predictors of business performance