Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 ANDREA RANGONE AND RAFFAELLO BALOCCO INTRODUCTION Business-to-consumer (BtoC) e-commerce has increased during the last years as the Internet has grown and spread almost world-wide. The literature on decision-making variables at a company’s disposal to improve its ecommerce strategy is contrasting, probably because of the evolutionary nature of the matter. On the one hand, there is increasing interest in in-depth study of specific topics, such as online marketing, website design and implementation, and transactions security over the Internet, while, on the other, some very important subjects, e.g. logistic process management, the means to determine product portfolio choices and pricing strategies with regards to e-commerce have received little attention. Besides these varying levels of interest in different otpics, little attempt has been made to define a systemic framework integrating decision-making variables and performance indicators that can be used by management in planning and controlling e-commerce strategy. The present paper aims to provide a systemic framework of the decisionmaking variables available to a company running a BtoC e-commerce business and of the performance indicators for an ex ante evaluation and an ex post control of actions on these variables. This paper deals with the following topics: 1. Contributions in the literature; summary of contributions from academics, practitioners and consultants concerning decision-making variables on which an ecommerce company can act, and performance indicators to control the actions on those variables; 2. Decision-making variables; the ecommerce mix framework provides a systemic view of the decisionmaking variables that management can use in the planning and decision-making process; 3. Performance indicators; the framework defines a set of indicators to evaluate and monitor the effects of action on e-commerce mix variables. A b s t r a c t The literature on decision-making variables at a company’s disposal to improve its e-commerce strategey is contrasting, probably because of the evolutionary nature of the matter. Besides the varying levels of interest in different topics, little attempt has been made to define a systemic framework integrating decision-making variables and performance indicators that can be used by management in planning and controlling a BtoC e-commerce strategy. The present paper aims to provide a systemic and overall framework both of the decision-making variables available to a company running a BtoC e-commerce business and of the indicators for an ex ante evaluation and ex post control of actions on these variables. CONTRIBUTIONS IN THE LITERATURE The analysis of the literature dealing with decision-making variables relevant to an e-commerce business covers three main areas, website design, security over the Internet and payment methods, and online promotion, and a number of residual matters, such a logistic process, pricing strategy and pre- and after-sales services. As regards website design, there is general agreement on the different issues such as Web design, Web navigaton and Web usability (e.g. Bucha- A u t h o r s Andrea Rangone (andrea.rangone@polimi.it) is Professor of e-commerce at Politecnico di Milano. His research focuses on performance measurement systems for e-commerce and strategic analysis and economic justification of e-business projects. Raffaello Balocco (r.balocco@libero.it) is a research fellow at Department of Economics and Production at Politecnico di Milano. Copyright © 2000 Electronic Markets Volume 10 (2): 130–143. www.electronicmarkets.org A Performance Measurement System for Planning and Controlling a BtoC E-Commerce Strategy Keywords: e-commerce, decision making, performace evaluation RESEARCH Electronic Markets Vol. 10 No 2 Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 nam and Lukaszewski 1997; Instone 1997; Nemzow 1997; Ruh et al. 1997; Fleming 1998). Neverthless, one problem is still open: the application of principles and general guidelines to e-commerce websites. Although a number of authors have studied and described in depth the principles for designing and implementing generic websites (e.g. Emery 1997; Graham 1997; Heba 1997; Helinski 1997; Nemzow 1997; Parker 1997; Radosevich 1997; Scholtz 1997; Siegel 1997; Sullivan 1997; Whinston et al. 1997; Fleming 1998; Haine 1998; Morville and Rosenfeld 1998; Lynch and Horton 1999; Wilson 1998, Tilson et al. 1998) some gaps remain in the application of these principles to e-commerce websites and, therefore, to the process of ordering products and commercial content. Internet security is addressed in a good number of contributions focusing attention on cryptography systems (public and private key cryptography) that ensure secure data transmission, and on-line payment methods based on the Secure Socket Layer (SSL), Secure Electronic transaction (SET) or virtual cash systems, such as Mondex or Digicash (e.g. Bhimani 1996; Panurach 1996; Spar and Bussgang 1996; Kalakota and Whinston 1997; Nemzow 1997; Borenstein 1998; Honeycutt et al. 1998; Treese and Stewart 1998). In the studies on online promotion, the emphasis given by many authors (e.g. Hoffman and Novak 1996; Alba et al. 1997; Hamil 1997; Kannan et al. 1998; McCullough 1998; Nadherny 1998) to the new communication channels made available over the Internet is not reflected in equal attention to the ‘operational’ tools at a company’s disposal. While various authors have developed theories regarding new communication models, few have sought to define a set of guidelines to assist management in promoting and advertising the e-commerce website. The literature also reveals that some subjects concerning e-commerce decision-making variables have still received little attention. These include: 131 · the logistic process, which is critical in BtoC e-commerce because of the element of home-delivery; · the portfolio choices, which are of critical importance, since each product type has a different degree of saleability over the Internet; · pricing strategy, which might have specific features in ecommerce; · pre- and after-sales services, such as the virtual shop assistant or the order tracking and interactive services, which aim to increase the value of the commercial activity and may represent an effective variable for online differentiation. The work on the performance indicators available to companies to control and monitor e-commerce performance fails to provide an overall view of all the decisionmaking variables. In particular, the authors focus on monitoring promotion and the level of usability of the site. The indicators for the former are essentially based on the analysis of log-files (e.g. Buchanam and Lukaszewski 1997; Emery 1997; Treese and Stewart 1998), while the techniques to test site usability are termed Web Usability Engineering (Buchanam and Lukaszewski 1997; Instone 1997; Nielsen 1998) and can be divided into two categories: · user testing: site users are asked to complete questionnaires from which the usability level can be calculated; · heuristic evaluation: tests undertaken by a small group of experts to verify violations of fundamental principles for navigation, design, speed and compatibility. Overall, the analysis of the literature highlights the lack of a framework that integrates all decision-making variables relevant to a BtoC e-commerce project, and of a performance measurement system able to measure the effects of management’s action on those decision-making variables. The e-commerce mix framework described in the following chapters provides a systemic view of these decision-making variables and performance indicators. Specifically, the framework has two objecties: 1. to support management in the planning and decisionmaking process; 2. to control the effects of action on decision-making variables, so initiating a learning process that might be critical in future decisions. THE DECISION-MAKING VARIABLES The variables on which the company can act to increase Net Cash Flow are divided into three categories on the analogy of the concept of ‘marketing mix’ (Kotler 1997),1 since these variables strongly affect the attractiveness of online business and, therefore, the expected revenue (see Figure 1): 1. website variables; 2. products and services variables; 3. promotion variables. These macro-categories are then broken down hierarchically to give a group of elementary variables. Website This macro-variable is broken down into three first-level variables: web interface, transaction management and content. Web interface. The interface includes all the web pages and their links in the virtual store. As shown in Figure 1, the web interface can be taken as the result of three variables: 1. web design, for which the elementary variables: (a) site architecture, i.e. the conceptual design of the links Transactions. Online transactions involve two phases: 1. order process, i.e. how the customer chooses the products to buy and sends the order together with the relevant personal data. This variable is divided into two basic variables: (a) the order system, i.e. shopping cart, online form, e-mail; (b) the order process, i.e. the navigation path followed by the user from the initial product selection to completion of the order placement, that may merely consist of a single click (see, for instance, Amazon one-click-ordering service); 2. payment, which may be online by credit card or virtual cash, or offline by bank transfer, cheque, money order or cash-on-delivery. Content. The informational content within a website can be classified in two categories: 1. commercial information, including a full description of the product range (images and text), the means of delivery (options, times, costs) and the method of payment (description of the security system used for data transfer); 2. background content, such as ‘about us’ information, details on the economic sector or other information not strictly related to the business activity (see, for instance, Peck.it ‘about us’ web page, where customers can find the story of the store and other related information). Products and Services This macro-variable is broken down into five first-level variables: products, pre- and after-sales services, pricing strategy, delivery services and virtual community services. Planning and Controlling a BtoC E-Commerce Strategy between web pages and, therefore, the organization of content; (b) graphics, i.e. the use of images, text and frames, and the layout of these elements in the web pages; 2. navigation system, divided into: (a) navigation through pages, i.e. the use of tools (links navigation bar etc.) to move from one page to another; it is possible to find many examples of the use of such tools browsing the best known web stores: Amazon.com, Cdnow.com, BarnesandNoble.com, Landsend.com etc.; (b) navigation in the same page, i.e. a design which allows easy navigation within a single page; (c) information search, i.e. the use of tools (search engine, site map, table of contents) to find information or products; those tools are useful to improve web store navigability as the depth and the width of the products range increases: Peck.it, a web store of Italian gourmet, that offers thousand of wines from serveral Italian producers, allows customers to search by name, producer, year, region, type and price; 3. personalization, i.e. the possibility for users to personalize web pages (e.g. the home page elements) according to their needs and tastes; Dell.com, for instance, allows customers to personalize the support of section of the web store (http://support.dell.com) according to their technical skills and computer features. Rangone Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 Figure 1. E-commerce mix decision-making variables 132 Products. The choice of products to offer via the website can be based on the following variables: 1. portfolio, i.e. the type of the products and the breadth, depth and complementarity of the range; 2. product differentials, i.e. the products competitive advantages with respect to online and offline competitors (e.g. quality, brands, etc.). Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 Pre- and after-sales services. The Internet is an extremely effective and efficient means of offering customers pre- and after-sales services. It is effective because it supports a range of high value added services to the customer, and efficient because the company can reduce the costs traditionally associated with these services. This lever can therefore be used both to differentiate with respect to competitors and to create cost differentials in particular with offline suppliers. Over the Internet it’s easy to find pre- and after-sales services. Here there are some examples: Landsend.com helps the customers build a 3-D model using exact measurement for virtually trying clothes on before buying (see ‘Your personal Model’ web page); Garden.com allows web navigators to design a virtual garden using the ‘Garden Planner’ service and to order all the selected plants and flowers by a single-click; CdNow.com allows customer to listen to some tracks from the CDs sold over the web store; several computer web store (e.g. Dell.com, IBM.com, Hp.com, etc.) offer an online trouble-shooting and customer support services. Electronic Markets Vol. 10 No 2 Pricing strategy. The elements to be considered in the definition of a pricing strategy are: 133 1. price level, in comparison to online and offline competitors. In electronic commerce, companies can use price discrimination and through market segmentation establish different price levels for different groups of customers. Using online tools such as active agents (e.g. MySimon.com), it is also possible to continually monitor prices set by online competitors in order to avoid different price levels with respect to them; 2. discount policies, allow the company to increase the number of purchases via the Web. The company can provide different type of discounts (e.g loyalty discounts, cross selling discount, etc.). Delivery services. This is a particularly critical variable for BtoC e-commerce, because a sales channel via the World Wide Web is accessible from everywhere (unless the company introduces its own restrictions), meaning that there is no spatial relationship between the customer and the supplier: an order can arrive from any Internet user. The company or the supplier must be able to deliver quickly for a reasonable cost. The decision-making variables affecting delivery on which the company can act are: 1. front-office, i.e. the delivery methods offered and the tracking system, which determine how the customer perceives the service; 2. back-office, i.e. distribution channel (internal or external, existing or new), inventory management (which products are held in stock and where) and the logistic process. Virtual community services. The company can stimulate the development of a virtual community on its website for a number of reasons: to collect information on the members of the community, to establish long-term relationships with users to create loyalty, to focus efforts on a group of consumers with similar interest. The typical services of a virtual community can be divided into two categories: 1. interactive services, divided into among users and userwebmaster services. The former can be direct when information is transmitted without any involvement of the company (e.g. non-moderated chat and discussion forums), or indirect when the company verifies content before publication (FAQ, user feedback). The userwebmaster services services primarily facilitate contact between potential customers and the company (chat with company employees, newsletter). Garden.com, for instance, organises virtual meetings where customers can chat online with expert gardeners and doctors; 2. informational content strictly connected to the social dimension; i.e. information services which develop a sense of community within the website. For example, Studiomoda.com, a web store selling Italian apparel and accessories, contains an electronic magazine called ‘YesPlease’ about Italian style. Promotion This macro-variable concerns all the decision-making variables on which it is possible to act to increase Web store visibility. Two first-level variables can be defined: online promotional channels and offline promotional channels based on traditional methods of advertizing. Online promotiuonal channels. These can be divided into two categories: 1. based on the WWW, which includes search engines, web directories, banners, insterstitials, the exchange of links and affiliate programmes; 2. below the WWW, i.e. advertizing postings, sponsorships, direct e-mailing lists, online public relations. THE PERFORMANCE INDICATORS This chapter proposes a system of indicators, which company management, can use to: · make the most effective choices when planning a BtoC e-commerce strategy; · monitor action on decision-making variables constantly and, in the benchmarking phase, compare the performance of its own Web store with that of competitors; · assess the performance of its own strategy to ensure improvement and learning processes. The starting point in the definition of the indicators is the objective that a company involved in e-commerce sets. Strategic literature states that a long-term objective for a company must be the creation of economic value, defined as the sum of NCFs over a given time horizon which goes from 0 to infinity: First level; the incoming cash flows are given by the product of the number of buyers (n. buyers) and the average value of a purchase (Ap). The two values are calculated separately for first-time (1st time) and returning (returning) customers. Second level; the number of new customers (n. buyers 1st time) depends no the number of visitors accessing the site for the first time, the percentage of visitors exiting the site after having viewed the access page and thus not entering the site, (exit rate), and the percentage of visitors making a purchase (conversion rate). For the customers who have already made a purchase, the relationship is similar, except that the exit rate is not significant: a customer returning to the site is obviously interested in its content. Third level; The number of first-time visitors is calculated as the sum of the visitors learning of the site through the company’s promotional activities, and those discovering the site through word-of-mouth. The number of returning customers can be divided into those returning as a result of promotional activities, those returning spontaneously, i.e. without being induced by a promotional activity, and customers encouraged by the company to return through e-mailings, i.e. mailings or newsletters publicising promotions, new products, new sections in the site, etc. Figure 2 also shows the relationships between the basic components of the cash flows and the three macro variables of the e-commerce mix framework: promotion, website and products and services. 1 X NCF(t) (1 ‡ r) t If the opportunity cost of capital (r), which depends on the level of risk involved in the commercial activity, is taken as fixed, economic value is given by Net Cash Flow, that can be seen as the difference between cash inflows and cash outflows: NCF ˆ FFIN - FFOUT Below, two groups of indicators (financial and nonfinancial) for incoming and outgoing cash flows are proposed, highlighting how they are measured and their limitations. Specifically, promotion influences: Indicators for Cash Inflows Figure 2 shows the breakdown of cash inflows into the basic components which are affected at different levels by the three macro variables identified in the e-commerce mix model that we call value driver products and services, web site and promotion). The following sections describe the breakdown of incoming cash flows illustrated in Figure 2. · number of visitors (for both first and return visits); exit rate, as the promotional message becomes more focussed on the right company customers percentage of users leaving the site after viewing the access page diminishes they are more interested in the products offered; · conversion rate; there is an indirect correlation between promotion and conversion rate: as the level of focus on · Planning and Controlling a BtoC E-Commerce Strategy t ˆ0 Rangone Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 V (0) ˆ Figure 2. Breakdown of cash inflows 134 the right company customers increases, the probability that users make a purchase on the site is greater; · number of visitors returning by e-mailing; through promotions based on e-mailing customers and registered users (newsletter, direct mailing, mailing list), the company can encourage customers to return to the site. The website influences: · Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 number of visitors returning spontaneously or through word-of-mouth. The overall effectiveness of a web site (interface, order and payment system, content) can encourage customers to return spontaneously to the site and may favour word-of-mouth; · conversion rate; the usability of the web interface (clear navigation, good design, compatibility and speed), the effectiveness of the informational content and the transaction system (order and payment) increase the probability that users make purchases; · exit rate; the effectiveness of the home or site access page has a major impact on the percentage of users leaving the site after seeing this page. Products and services influence: · number of visitors returning spontaneously or thorugh word-of-mouth. The features of the products and services portfolio can encourage customers to return spontaneously to the site and may favour word-of-mouth publicity; · conversion rate, as for the web site, a good products and services portfolio influences the number of users placing an order. For each of the three categories (website, products and services and promotion), a number of indicators have been defined, underlining their limitations and the difficulties in measurement when evaluating and monitoring the effects of acting on e-commerce mix decision-making variables. Performance Indicators for the Website Electronic Markets Vol. 10 No 2 To assess the effectiveness of the website, three methods are proposed which can be classified in two dimensions: 135 · output of the assessment (analytical or synthetic): the analytical output allows individual features of the site to be monitored, while the synthetic output gives an overall evaluation of the site effectiveness; · the subject performing the assessment (internal to the company or visitor/user). Figure 3 classifies the the three methods proposed in two dimensions: As shown in Figure 3, a synthetic method applied by an external subject (lower right cell) is not very significant. The output is an overall judgement of the effectiveness of the site which, as evidenced below, can be deduced by reprocessing an analytical survey (upper right cell). Figure 3. Methods to assess web site effectiveness Fuzzy Model In order to obtain an overall, synthetic assessment of the effectiveness of an e-commerce site a model based on Fuzzy Set Theory (Zadeh 1965; Lee 1990; Zimmermann 1991) is proposed. Figure 4 shows a group of 35 indicators whose values can be used to calculate overall effectiveness. These indicators can be divided into two categories: 1. boolean indicators which can be measured by On-Off type values. This category includes all the indicators in round brackets in Figure 4 (e.g. presence of a navigation bar or no ambiguous links or dead-end documents as navigability); 2. discrete indicators, which can be measured qualitatively (e.g. High-Medium-Low). Examples of such indicators (in square brackets in Figure 4) are attractiveness and descriptive power for the assessment of the effectiveness of the images. The following paragraphs describe an assessment model based on Fuzzy Set Theory which is able to convert all types of boolean indicator intod iscrete indicators and subsequently integrate all these indicators to obain an assessment of the overall effectiveness of a virtual store or one of its components. Specifically, two models based on Fuzzy Set Theory 2 are used: 1. the fuzzy heuristic model to transform the values of the boolean indicators into values of a linguistic variable (Dubois and Preade, 1984; Lee 1990; Kosko 1993); 2. the fuzzy linguistic model to provide an overall assessment of the effectiveness of the website based on the effectiveness of its components (Zimmermann 1991; Rangone 1997; Perego and Rangone 1998). Implementaton of the fuzzy model requires the following steps: 1. Determine the importance weightings of the various discrete indicators identified in the analytical model. 2. Define the decision-making rules of the fuzzy heuristic model for the boolean indicators and assess the performance of the virtual store for each indicator; 3. Calculate the overall effectiveness of the virtual store by weighting the assessments with the importance weightings. Step 1. Determine the importance weighting of the discrete indicators. In a fuzzy model, the relative impor- S X ˆ (Very High, High, Medium, Low, Very Low), where (Lee 1990; Liang and Wang 1993; Perego and Rangone 1998): Very Low ˆ (0;0;0,3), Low ˆ (0;0,3; 0,5), Medium ˆ (0,2;0,5;0,8), High ˆ (0,5;0,7;1), Very High ˆ (0,7;1;1). Step 2. Define the decision-making rules to apply to the fuzzy heuristic model for the boolean criteria and assess the performance of the virtual store for each assessment criterion. Definition of the decision-making rules allows all the possible combinations of values for the boolean criteria to be converted into the fuzzy linguistic values used to assess the virtual store.3 For example, the following decision-making rules may be used to asses graphical coherence. Graphical coherence [GC] Criterion Subcriteria GC GC1 GC2 GC3 GC4 Description Overall coherence graphical Consistent background for each section of the site Consistent navigation bar Consistent fonts Consistent layout of the various pages Values On/Off On/Off On/Off On/Off 1. If [(GCi ˆ On)], then [GC ˆ Very good], 8 i ˆ 1, . . . 4. If all criteria are On, the graphical coherence is Very good. 2. If [(GCi ˆ Off )] then [GC ˆ Good], i ˆ 2, 3, 4. If only one criterion is Off, except background (i ˆ 1), graphical coherence is Good. 3. If [(GC1 ˆ On) And (GCi ˆ GCj ˆ Off )] then [GC ˆ Fair], i 6ˆ j ˆ 2, 3, 4. If the background is uniform, but two of the other three criteria are Off, the coherence is Fair. 4. If [(GC1 ˆ Off ) then [GC ˆ Poor]. If the background is not consistent in each section, graphical coherence is Poor. 5. If [(GCi ˆ Off ), then [GC ˆ Very poor], 8 i ˆ 1, . . . 4. If all the criteria are Off, graphical coherence is Very poor. Now it is necessary to measure the performance of the virtual store on each indicator: 1. For the discrete indicators, the linguistic variable W is used, as in the definition of the decision-making rules in Step 2. These can be measured on the following linguistic scale (Liang and Wang 1993; Perego and Rangone 1998): S W ˆ VP, P, F, G, VG, where VP ˆ Very Poor (0, 0, 0.2), P ˆ Poor (0, 0.2, 0.4), F ˆ Fair (0.3, 0.5, 0.7), G ˆ Good (0.6, 0.8, 1), VG ˆ Very Good (0:8, 1, 1); 2. For the boolean indicators, the decision-making rules devised in Step 2 must be applied, in order to obtain a linguistic value belonging to the linguistic variable W . Planning and Controlling a BtoC E-Commerce Strategy tance of the different assessment criteria is measured by means of a linguistic variable, using the following scale: Rangone Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 Figure 4. Framework to evaluate website effectiveness 136 Step 3. Calculate the overall effectiveness of the virtual store. To calculate the overall effectiveness, the formula proposed by Zadeh can be used (Zadeh 1965): K ˆ (1=m) [(A 1 (A 3 W3) W 1) K (A 2 (A m Score 1. Graphical coherence in the various sections of the site 2. Legibility of the web pages 3. Quality of the images 4. Navigation between pages 5. Locating information/products within the site 6. Download speed of the web pages 7. Ordering (system and process) 8. Payment and security system 9. Textual description of the products 10. Product images 11. Background content (about us, information about the sector, other information) From 1 to 5 W2) Please assess: W m )] where: is a fuzzy number representing the overall effectiveness of the virtual store; W J is the weighting of the jth assessment criterion; A j is the linguistic value for the jth criterion and m the number of assessment criteria. K From From From From 1 1 1 1 to to to to 5 5 5 5 From From From From From From 1 1 1 1 1 1 to to to to to to 5 5 5 5 5 5 Set of Indicators Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 An easier way to mmonitor site performance that can be used if there’s no need to have a synthetic output but it is sufficient to monitor some critical performance, is to define a set of indicators starting from those given in the fuzzy model (see Figure 3). It is prefereable to consider only the quantitative indicators, which can be calculated objectively, while the qualitative values should be obtained using the analytical survey detailed in the next section. Example indicators might be: · products; · pre- and after-sales services; · pricing strategy; · delivery; · virtual community services; · download speed of the web pages; · number of broken links; · number of dead-end pages; · average percentage of page errors during a session. The company must decide the most appropriate set of indicators, basing this choice strategy to be adopted, on its sector, on the external context in which it operates, etc. Products Some possible indicators that can be used to monitor the effectiveness of an e-commerce product portfolio are: 4 these indicators can be measured with special software and allow the company to monitor the quantitative variables which affect the site usability. Electronic Markets Vol. 10 No 2 Analytical Survey 137 This involves asking users to assess the effectiveness of different features of the site. The starting point for the questionnaire can be the indicator tree used in the fuzzy model (see Figure 4). The company can decide how far to go down the tree and which parts to emphasise. The visitors are asked to respond to a questionnaire, which is sent by e-mail or available as an online form. Responses can be evaluated using, for example, a points score method. An example questionnaire is given below. The score for each response can be summed (e.g. as a weighted average) to obtain o aggregated assessments or even a full overall evaluation. · range, i.e. the number of different product types (range width) and variants of each type (range depth) in comparison with online and offline competitors. This can be measured directly within competitors’ sites for online competitors or by an analysis of each traditional sales channel for offline competitors, as the same product type can be offered in different ranges in function of the sales channel; · product competitive differntials, which measures the presence of attractiveness differentials linked to product features in comparison with online and offline competitors (e.g. quality, brands, etc.). As for the breadth and depth of the product range, it can be measured directly within competitors’ sites for online competitors and by an analysis of each traditional sales channel for offline competitors. Pre- and After-sales Services Performance Indicators for Product and Services For the products and services variable, various indicators are proposed which refer to: There are many pre- and after-sales services and, as stated previously, they often depend on the type of product traded (see above). The indicators which can be used to monitor the effectiveness of these services can be divided Delivery into two categories on the basis of the means of measurement: Three possible indicators to measure delivery effectiveness are proposed: · indicators measured by the company itself using log files; · indicators measured by survey. · Some possible indicators which the company can measure are: Number of options offered. This measures the number of delivery options offered by competitors on their sites in order to conduct comparative analyses. The indicator can be obtained by an online analysis of competitors’ sites in the same strategic group (for comparable products). · Delivery time. It measures the delivery time applied by competitors in order to conduct comparative analyses and can be obtained by: The indicators which can be measured by survey are heavily dependent on the type of service provided by the company. Generaly speaking, the company can measure: 1. an online analysis of competitors’ sites in order to verify the declared data; 2. an empirical survey by means of an order placed with a competitor; 3. a customer surveys to obtain own delivery time. · · the level of use of pre-sales services by visitors and customers; · the level of use of after-sales services by customers; · the effectiveness of each service from the user’s point of view; · possible improvements to existing services; · the need for new services. Distribution of preferences between different delivery options. It is useful to identify possible shortcomings in the options offered and it can be obtained by analysis of orders received over a given period of time. Virtual Community Services · Indicators measured by the company itself using log files; · Indicators measured by survey. Pricing · The indicators used to monitor pricing strategy measure the differentials with respect to online and offline competitors, as well as the handling and shipping charges applied by other suppliers. Price differentials with respect to offline competitors. The indicator can be obtained by monitoring the prices of competitors operating in traditional channels (for comparable products). It measures the customer advantage in purchasing online rather than through traditional channels; · Price differentials with respect to online competitors. The indicator can be obtained by an online analysis of competitors’ sites in the same strategic group (for comparable products). It measures the customer advantage in purchasing online from one company rather than from another; · Delivery charge. The indicator measures the delivery charge applied by competitors in order to conduct coparative analyses. The indicator can be obtained by an online analysis of competitors’ sites. A number of indicators which can be measured internally are described below: · particpants in virtual community, which is useful to evaluate the level of interest in the virtual community; · participants in an interactive service/website visitors, which evaluates the level of interest in the interactive service. This can be measured with log files only if the interactive services (e.g. chat, discussion forum, mailing list) require a specific user registration; · buyers registered in the community/total buyers, indicates the community impact on web store sales. A survey can measure indicators similar to those described above for pre- and post-sales services (level of use, effectiveness, possible improvements, need for new services), as well others linked to the social dimension. These latter include, for example, relationships between users, effectiveness of interaction with the webmaster, usefulness of contributions received from users. Planning and Controlling a BtoC E-Commerce Strategy As for the pre- and post-sales support, the interactive services of the virtual community can also be monitored by two types of indicator: Rangone Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 · number of users accessing a given service/overall number of users; this gives the level of interest in each service; · percentage of customers which use a pre-sales service prior to purchase; this measures the effectiveness of each presales service; · percentage of customers which use an after-sales service; this measures the usefulness of the service for the customer, but can only be measured if registration is requested to access the service. 138 Online Promotion Indicators tors, some of which are measured differently in function of the channel in question: This section proposed a set of indicators to assess the effectiveness of online promotion in increasing website visibility. The effectiveness of the ith promotional channel can be calculated by the following equation: i Nbuyers BDG i · (1) CPM is useful to evaluate the cost of promotional activity. The measurement depends on the promotional channel in question; · click-through rate evaluates the efficiency of a promotional channel and of the message content. The measurement depends on the promotional channel in question; · exit rate evaluates the efficiency of the message content. It can be obtained by log file analysis; · conversion rate also assesses the correct targeting of promotional activity, and can be obtained by log file analysis. where: i N buyers is the number of buyers who entered the site via the ith promotional channel Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 BDG i is the budget allocated for the ith promotional channel. Obviously, the assessment of the effectiveness of a given promotional channel by means of the ratio the number of buyers and the budget presupposes that the company’s objective is the increase site sales. Other effects, which a promotional campaign might provoke, e.g. increased brand awareness, are not considered. The equation (1) can be further broken down into some basic components: i N buyers BDG i ˆ i N impression BDG i where: i N impression is the number of times that the promotional message is seen by potential customers through the ith channel; CTR i (click through rate) is the percentage of visitors entering the site after having seen the promotional message through the ith channel; ER i (exit rate) is the number of visitors arriving through the ith channel who leave the site after seeing the access page; CR i (conversion rate) is the percentage of visitors arriving through the ith channel who make a purchase; CPM i (cost per million) is the cost of 1000 impressions through the ith channel. The effectiveness of each promotional channel can therefore be monitored using the following four indicaElectronic Markets Vol. 10 No 2 · search engines and Web directories; · banners; · other tools based on e-mail (newsgroups, and mailing lists, etc.); In addition to the indicators already described, various specific criteria for individual channels are also given. . CTR i . ER i . CR i CTR i . 1:000 . ˆ ER i . CR i CPM i 139 In the next section, the means of measuring CTR and CPM is given for the following promotional channels: Search engines and Web directories. Possible specific indicators for search engines and web directories are: · frequency of key words used by visitors to find the site. These key words can be identified by analysing the log files, as the URL from which the visitor accesses the search engine contains the key word used; · positioning ij , i.e. the position5 of the site in the results of the ith search engine for a query with the jth key word. This is an indicator of the efficiency of the subscription by the company. The main problem is the time horizon for which the indicator is valid, as the position can vary frequently, meaning that the monitoring must be almost continuous. The indicator can be measured by querying the search engine using the key words in question, or with special software which verifies automatically the position in various engines. Table 1. CPM and CTR measurement for search engines and web directories CPM ˆ (BDG=N impression ) 1:000 The numerator (BDG) can be calculated as the purchasing cost of the key words, as the cost of the man-hours necessary to subscribe to the search engines, or as the purchasing cost of the software for automatic subscription. The demoninator (Nimpression ) is difficult ot measure since it is not known how often the search engine user views the site in the query results. It can be calculated indirectly by the number of search engine users querying with a specific key and the position of the site when using this key word. CTR ˆ Nvisitors =N impression The numerator (Nvisitors ) can be calculated by analysis of the log files Table 2. CPM and CTR measurement for banners CPM ˆ (BDG=N impression ) 1:000 Defined when the banner campaign is purchased CTR ˆ N visitors =Nimpression Analysis of log files The following section describes relationships between the basic components of the cash outflows and the threemacro variables of the e-commerce mix: website, products and services and promotion. The website influences: · Indicators for Cash Outflows Products and services influence: · fixed costs of customer care, as interactive services and pre- and after-sales support vary in function of the type of product range. Costs for personnel to manage these services fall into this category if software for automatic management is not used; · variable product costs, i.e. production costs of costs of purchases from suppliers; · delivery costs for packaging and dispatch, which depend on the type of product and the stock management, itself determined by the breadth and depth of the product range. Table 3. CPM and CTR measurement for e-mail based tools CPM ˆ (BDG=N impression ) 1:000 The numerator can be calculated as the cost to acquire the e-mail addresses (for direct mailing), or as the cost to sponsor the newsletter or mailing list, or as the cost of the man-hours expended to set up and maintain a customer database. The denominator can be calculated as the number of e-mails sent (for direct mailing), or as the number of subscriptions to the newsletter or mailing list (for sponsorships). Newsgroups are difficult to measure, as the number of users is not known. CTR ˆ N visitors =Nimpression The numerator is calculated as the number of accesses to the access page, as shown in the log files. Planning and Controlling a BtoC E-Commerce Strategy The indicators for cash outflows are the variable and fixed cost sustained by the company for the start-up and management of the virtual store. In this case, too, the three-macro variables of the e-commerce mix (website, products and services and promotion) can be used, establishing for each the cost items. These costs must be defined ex ante in the budget phase (standard costs), and subsequently monitored in order to identify any changes. Figure 5 shows the breakdown of cash outflow and the relationships between the basic components of the cash outflow and the three-macro variables of the e-commerce mix. Rangone Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 Banners. Other tools based on e-mail. A possible specific indicator for tools based on e-mail is the percentage of protest mail. This can be calculated as the ratio between the number of protest e-mails and the total number of emails sent through a given promotional channel, so assessing the company’s respect of netiquette. The only problem with the measurement is identification of the protest e-mails regarding a specific advertising campaign, so that the numerator and denominator remain coherent. initial investment for graphical design and production of the web pages, integration between the web pages and the e-commerce software, purchase of software licences and hardware, etc.; · fixed costs of site management, i.e. residence and maintenance costs, personnel costs for ordinary updates, and for order processing; · fixed costs of customer care, determined by the use of ad hoc software for the automatic management of the interactive services offered on the site; · variable costs in function of the number of financial transactions, i.e. costs for credig card payment management, order processing costs, etc. Figure 5. Breakdown of cash outflows 140 Finally, promotions determine the following cost items: · fixed promotional costs, i.e. costs for marketing personnel and for advertising campaigns in various channels, etc.; · variable promotional costs with regard, for example, to affiliations, promotional campaigns paid on the basis of sales generated (e.g. banner campaigns, etc.). The indicators to monitor outgoing cash flows involves various cost items. As stated above, these must be constantly monitored to identify any changes with respect to budgeted figures. CONCLUSIONS Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 Analysis of the literature reveals an absence of a framework to integrate decision-making variables and indicators that can be used by a company in running a BtoC e-commerce business. The e-commerce mix framework described in the previous sections provides a systemic view of the decisionmaking variables and the performance indicators available to management. Specifically, the framework has two objectives: 1. to support management in the planning and decisionmaking process; 2. to control the effects of action on decision-making variables, so initiating a learning porocess that might be critical in future decisions. Electronic Markets Vol. 10 No 2 Notes 141 1. Note that ‘pricing’ variable is considered within the ‘products and services’ variable. 2. See Appendix 1. 3. These are linguistic values belonging to the linguistic variable W (cf. Step 3). 4. For example, Site Mapper, Astra site manager, InfoWeb, Riadalinx, Xenu’s link sleuth. 5. The term refers to the number of pages the user has to turn in the search engine results before reaching the company’s link. 6. In the present case, the alternatives refer to different virtual stores analysed in the benchmarking phase. References Alba, J., Lynch, J., Weitz, B., Janiszewski, C., Lutz, R., Sawyer, A. and Wood, S. (1997) ‘Interactive home shopping: consumer, retailer, and manufacturer incentives to participate in electronic marketplace’, Journal of Marketing 61, 38–53. Azzone, G. and Rangone, A. (1996) ‘Measuring manufacturing competence: a fuzzy approach’, International Journal of Production Research 34(9), 2517 –32. Bhimani, A. (1996) ‘Securing The Commercial Internet’, Communication of the ACM 6, 29–35. Borenstein, S.N. (1998) ‘Perils and Pitfalls of the Practical Cybercommerce’, Communication of the ACM 6, 36–44. Buchanam, W.R. and Lukaszewski, C. (1997) Measuring the Impact of Your Web Site, New York: Wiley & Sons Inc. Dubois, D. and Preade, H. (1984) ‘Fuzzy logic and the generalised modus ponens revisited’, Cybernetic System 15, 3– 4. Emery, V. (1997) How to Grow Your Business on the Internet, Scottsdale, AZ: Coriolis Group Books. Fleming, J. (1998) Web navigation, designing the user experience, Cambridge, MA: O’Reilly. Graham, P. (1997) ‘Four secrets to internet retailing’ [www.eretail.net] accessed March 1999. Haine, P. (1998) ‘Five most serious web design errors’ [www.hp.com] accessed March 1999. Hamil, J. (1997) ‘The Internet and International Marketing’, International Marketing Review 5, 300 –23. Heba, G. (1997) ‘Digital Architectures: a Rhetoric for Electronic Document Structures’, IEEE Transaction on Professional Communication 4, 276 –83. Helinski, P. (1997) ‘Building a quality Web site’, Web Techniques magazine 2 [http://www.webtechniques.com] accessed March 1999. Hoffman, D.L. and Novak, T.P. (1996) ‘Marketing in Hypermedia Computer-Mediated Environments: Conceptual Foundations’, Journal of Marketing 60, 50– 68. Honeycutt, E.D., Flaherty, T.B. and Benassi, K. (1998) ‘Marketing Industrial Products on the Internet’, Industrial Marketing Management 27, 63–72. Instone, K. (1997) ‘Site usability evaluation’ [www.webreview.com] accessed March 1999. Kalakota, R. and Whinston, A.B. (1996) Frontiers of Electronic Commerce, Reading, MA: Addison-Wesley. Kalakota, R. and Whinston, A.B. (1997) Electronic commerce: a manager’s guide, Reading, MA: Addison-Wesley. Kannan , P.K., Chang, Ai-Mei, and Whinston, A.B. (1998) ‘Marketing Information on the I-Way: Data Junkyard or Information Gold Mine?’, Communication of the ACM 41(3), 36–43. Karwowski, W. and Evans, G.W. (1986) Potential Applications of Fuzzy Sets in Industrial Safety Engineering, Fuzzy Sets and Systems 19(5), 105–20. Kosko, B. (1993) Fuzzy Thinking: The New Science of Fuzzy Logic, New York: Hyperion. Kotler, P. (1997) Marketing Management: Analysis, Planning, Implementation, and Control, Upper Gaddle River, NJ: Prentice-Hall. Lambin, J.J. (1990). Marketing, McGraw-Hill. Lee, C. (1990) ‘Fuzzy Logic in Control System: Fuzzy Logic Controller-Part 1’, IEEE Transaction on Systems, Man, and Cybernetics 20(2), 404 –18. Liang, G.S., and Wang, M.J. (1993) ‘A Fuzzy Multi-Criteria Decision-Making Approach for Robot Selection’, Robotics & Computer-Integrated Manufacturing 10, 267 –74. Zadeh, L.A. (1965) ‘Fuzzy Sets’, Information and Control 8, 338 –53. Zimmer, A.C. (1983) Verbal versus numerical processing in individual Decision Making Under Uncertainty, North Holland. Zimmermann, H.J. (1991) ‘The Concept of a Linguistic Variable and its Application to Aproximate Reasoning-II’, Information Sciences 8, 301 –57. APPENDIX 1 Notes on Fuzzy Set Theory Fuzzy Set Theory has been applied to numerous company problems with non-probabilistic uncertainty essentially due to the vague and ambiguous definition of the relevant variables (Karwowski and Evans 1986; Zimmermann 1991; Azzone and Rangone 1996; Rangone 1997). Zimmer, for instance, has asserted that humans tare unsuccessful in making quantiative predictions, whereas they are more efficient in qualitiative predictions (Zimmer 1983). For present purposes, attention is focused on two typical assessment models based on Fuzzy Set Theory and used to evaluate the effectiveness of a virtual store. These are: Fuzzy Linguistic Models Fuzzy linguistic models are based on the concept of linguistic variables. A linguistic variable is a variable for which the values are not numbers, but words or phrases of a natural language (Zimmermann 1991). A linguistic variable represents the group of its linguistic values, as each value is a fuzzy set. There are numerous fuzzy linguistic models (cf. for example, Liang and Wang 1993; Rangone 1997). The one used to assess the effectiveness of a virtual store is built in the following steps: 1. definition of the importance weightings of the assessment criteria; the importance of the criteria can be measured by the linguistic variable X ; 2. attrition of the values for each assessment criterion in the alternatives analysed;6 attribution is by the linguistic variable W ; 3. calculation of the overall assessment of each individual alternative, weighting the values attributed to the respective importance criteria and ranking the alternatives. The overall value of the assessment can be calculated with the following weighted average operator (Zadeh 1965): Planning and Controlling a BtoC E-Commerce Strategy · fuzzy linguistic models; · heuristic models based on fuzzy logic. Rangone Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 Lynch, J.P. and Horton, S. (1999) Web Style Guide, Basic Design Principles for Creating Web Site, New Haven, CY: Yale University Press. McCullough, D. (1998) ‘Web-Based Research: The Dawning of a New Age’, Direct Marketing 8, 36–8. Morville, P. and Rosenfeld, L. (1998) Information Architecture for the World Wide Web, Cambridge, MA: O’Reilly. Nadherny, C.C. (1998) ‘Technology and Direct Marketing Leadership’, Direct Marketing 7, 42–5. Nemzow, M. (1997) Building CyberStores, installation, transaction process & management, New York: McGrawHill. Nielsen, (1998) ‘That mess on your Web site’, Technology review 5, 72–5. Panurach, P. (1996) ‘Money in Electronic Commerce: Digital Cash, Electronic Funds Transfer, and Ecash’, Communication of the ACM 6, 45–58. Parker, R.C. (1997) ‘Guide to web content and design, Eight Steps to Web Site Success’, Parsons, A.J., Zeisser, M. and Waitman, R. (1996) ‘Current Research: Organising for digital marketing’, The McKinsey Quarterly 4, 185 –92. Perego, A. and Rangone, A. (1998) ‘A Reference Framework for the Application of Madm Fuzzy Techniques to Selecting Amts’, International Journal of Production Research 2, 437 –58. Radosevich, L. (1997) ‘Fixing web site usability’ [www.infoword.com] accessed March 1999. Rangone, A. (1997) ‘Linking Organisational Effectiveness, Key Success Factors and Performance Measures: an Analytical Framework’, Management Accounting Research 8, 207 –19. Ruh, J.H., Romano, C. and Ratner, J. (1997) ‘Roles, Organisation, and Support: Building Usability Into The Design Process’ [www.uswest.com] accessed March 1999. Scholtz, J. (1997) ‘Web Usability: The Search for a Yardstick’ [www.uswest.com]. Siegel, D. (1997) Creating Killer Web Sites, Hayden Books. Spar, D. and Bussgang, J.J. (1996) ‘The Net Ruling, the Internet Promises to be the Site of a Commercial Revolution, Harvard Business Review 3, 125 –33. Sullivan, T. (1996) ‘User Testing Techniques-A Reader Friendliness Checklist’ [www.pantos.org]. Sullivan, T. (1997) ‘The Vision Thing’ [www.pantos.org]. Tilson, R., Dong, J., Martin, S. and Kieke, E. (1998) ‘Factors And Principle Affecting The Usability Of Four ECommerce Sites’ [www.research.att.com] accessed March 1999. Treese, G.W. and Stewart, L.C. (1998) Designing systems for Internet commerce, Addison Wesley. Whinston, A.B., Stahl, D.O. and Soon-Yong, C. (1997) The Economics of Electronic Commerce, Indianapolis, IN: MacMillan Technical Publishing. Wilson, R.F. (1998) ‘Seven Debilitating Diseases of Business Web Sites (and Their Cures)’ [www.wilsonweb.com] accessed March 1999. 142 K i ˆ (1=m) K [(Ai1 (A im W 1) (Ai2 W 2) (Ai3 W 3) W m )] where: K i is a fuzzy number representing the overall value obtained in the assessment of the ith alternative; W J is the importance of the jth criterion; A ij is the linguistic value associated to the ith alternative for the jth criterion, and m is the number of assessment criteria; ‘ , ’ are the fuzzy algebraic addition and multiplication operators. By the Zadeh’s extension principle, the extended algebraic operations on triangular fuzzy numbers used in the framework are as follows: Downloaded By: [German National Licence 2007] At: 14:52 11 March 2010 (a 2 , b 2 , c 2 ) ˆ (a1 ‡ a 2 , b 1 ‡ b 2 , c 1 ‡ c 2 ) Multiplication (a1 , b 1 , c 1 ) (a 2 , b 2 , c 2 ) (a 1 a 2 , b 1 b 2 , c 1 c 2 ) Division (a1 , b 1 , c 1 ) Æ (a2 , b 2 , c 2 ) (a 1 =c 2 , b 1 =b 2 , c 1 =a 2 ) (1, 1, 1) Æ Electronic Markets Vol. 10 No 2 U i2 ˆ Ni X Yi ˆ Ni X Qi ˆ Ni X Zi ˆ Ni X ( pij - oij )(b ij - a ij )=N i iˆ1 iˆ1 ( pij (aij - b ij ) ‡ b ij (oij - pij ))=N i q ij c ij =N i j ˆ1 oij a ij =N i j ˆ1 pij b ij =N i j ˆ1 H i1 ˆ T i2 =2T i1 The formula gives a fuzzy number with a non-triangular membership function that can be approximated to a fuzzy triangle whose vertices coincide with the weighted average of the assessments of the various alternatives: K i ˆ (Y i , Q i , Z i ), (Zadeh 1965). On the basis of K i , the alternatives analysed can be ranked. The literature proposes different methods, including that based on the centre of gravity of the fuzzy number K i , calculated as: … xdS i Si … X Gi ˆ dS i Si Reciprocal 143 Ni X H i2 ˆ - U i2 =2U i1 Addition (a 1 , b 1 , c 1 ) U i1 ˆ (a, b, c) (1=c, 1=b, 1=a) The membership function of K can be calculated by the formula proposed by Zadeh. By the extension principal, K i is a fuzzy number with the following membership function: 8 > - H i1 ‡ [ H 2i1 ‡ (x - Y i )=T i1 ]1=2 > > > > > > > Yi < x < Q i > > < K i (x) ˆ H i - [ H 2i2 ‡ (x - Z i )=U i1 ]1=2 > > > > > > Q i < x < Z i i ˆ 1, 2, KM > > > > : 0 otherwise T i1 ˆ Ni X (aij - q ij )(a ij - c ij )=N i T i2 ˆ Ni X (q ij (a ij - c ij ) ‡ c ij (aij - q ij ))=N i iˆ1 iˆ1 Heuristic Models Based on Fuzzy Logic The models in this category are based on the concept of linguistic decision-making rules ( fuzzy) formulated in the following way (Dubois and Preade 1984; Lee 1990; Kosko 1993): IF (a series of conditions is satisfied) THEN (a consequence can be deduced): A fuzzy rule can be considered as a fuzzy conditional expression (Lee 1990), in which the condition (IF) expresses a combination of values assumed by the indicators, while the consequence (THEN) represents the corresponding assessment. This model has been used to transform the values of the On/Off indicators into linguistic values belonging to the linguistic variable W .