Preliminary Comments welcome
Does Online Customer Experience Affect the Performance of E-commerce Firms? Shivaram Rajgopal Assistant Professor Department of Accounting University of Washington Box 353200 Seattle, WA 98195 Tel: (206) 543 7525, Fax: (206) 685 9392 E-mail:
[email protected] Mohan Venkatachalam Assistant Professor-Accounting Graduate School of Business Stanford University Stanford, CA 94305 Tel: (650) 725-9461, Fax: (650) 725-0468 E-mail:
[email protected] Suresh Kotha Associate Professor-Strategy and e-commerce University of Washington Box 353200 Seattle, WA 98195 Tel: (206) 543 7525, Fax: (206) 685 9392 E-mail:
[email protected] September 2000
Rajgopal and Kotha appreciate funding from the Herbert Jones Foundation administered by the Program for Entrepreneurship and Innovation (PEI) at the University of Washington. Mohan Venkatachalam appreciates funding from the Center for Electronic Business and Commerce, Stanford University. We thank Gary Hansen and Micheal R. Bauer for their encouragement and support, Media Metrix for providing us with Web traffic data and Qian Wang for excellent research assistance. We also thank Mary Barth, Sandra Chamberlain, George Foster and workshop participants at the 2000 Stanford Summer Camp for their comments.
Does Online Customer Experience Affect the Performance of E-commerce Firms? Abstract: Claims have often been made, most notably by Jeff Bezos, CEO of Amazon.com, that the quality of the online customer experience in terms of Website ease of use, selection of goods offered, quality of customer service, the effectiveness of virtual community building, and site personalization are crucial to the success of e-commerce firms. Using scorecards of online customer experience provided by Gomez Advisors, we find that a 1% increase in the composite score of various dimensions of online customer experience is associated with a 1.66% increase in traffic reach and a 0.84% increase in revenues for the average e-commerce firm. However, investing resources in online customer experience does not appear to be cost-beneficial in the short-run. This is because a 1% increase in the composite score of online experience is associated with a 0.91% increase in operating expenses and a 2.11% decrease in net income for the average firm. Moreover, investors do not appear to believe that the quality of online experiences provides a sustainable advantage in the long run as evidenced by the absence of significant associations between online experience scores and market to book ratios. We also investigate whether there are statistically detectable differences in the customer experience scores of pure Internet firms when compared to those of their clicks and mortar counterparts. Although the average pure Internet firm provides a better online customer experience than the average clicks and mortar firm, this dominance is restricted to 6 of the 19 industries examined. Our study suggests that superior online experience can give pure Internet firms only a transient competitive edge in the battle for the domination of e-commerce.
1
Does Online Customer Experience Affect the Performance of E-commerce Firms? 1. Introduction This study examines whether the quality of online customer experience affects the performance of Internet firms focused on Business to Consumer e-commerce (hereafter “ecommerce” firms). Forrester Research, an Internet research firm, estimates that revenues in the Business to Consumer segment will grow from $20 billion in 1999 to $184 billion by 2004 (Economist, 2/26/00: 6). Such explosive growth is due, in most part, to superior shopping experiences offered by the new e-commerce firms. For example, Amazon made it easier and convenient to buy books, E*Trade made trading stocks cheaper and faster than calling a traditional broker, and eBay revolutionized trading by providing an online auction forum for anyone to reach a vast community of buyers and sellers. When asked why people come to his firm’s site, Jeff Bezos, CEO of Amazon.com, responded (Fast Company, 1996): Bill Gates laid it out in a magazine interview. He said, “I buy all my books at Amazon.com because I’m busy and it’s convenient. They have a big selection, and they’ve been reliable.” Those are three of our four core value propositions: convenience, selection, [and] service. The only one he left out is price: we are the broadest discounters in the world in any product category…These value propositions are interrelated, and they all relate to the Web.
However, skeptics point to low barriers to entry and intense competition on the Web and question whether superior online customer experience offered through greater convenience, selection, service and attractive pricing actually improves an e-commerce firm’s financial performance (New York Times 3/29/00, 6/7/00; Economist 7/1/00; Wall Street Journal 6/27/00). More importantly, they argue that the quality of online experience can be easily imitated by established businesses with deep pockets. Steven Riggio, the CEO of Barnes and Noble, observes (Fortune, 1997: 248): Anything Amazon.com can do on the Internet, so, too can Barnes & Noble. There was a mystique about how difficult it was to get started on the Web, but it is quickly fading. Hiring hot designers from Silicon Valley, Barnes & Noble now offers a Web shopfront that’s just as inviting and useful as Amazon’s, with easy-to-use subject indexes, online author events every day, book forums, book reviews, and other features. 2
Thus, it is unclear whether investments in the quality of online experience can create a sustainable long-term competitive advantage for pure Internet firms. Using scorecards of online customer experience compiled by Gomez Advisors, a respected Internet rating firm, this study attempts to empirically determine whether the quality of online customer experience is associated with performance measures such as Web traffic, revenues, gross margin, net income and market to book ratios for a sample of pure Internet firms. Gomez Advisors, as explained in greater detail in section 2, provides a scorecard that attempts to systematically capture the quality of the online customer experience offered by firms along five dimensions. These five dimensions are the Website ease of use, customer confidence in the Web business, selection of goods and services offered on the site, the effectiveness of relationship services such as virtual community building and site personalization, and the extent of price leadership practiced by the firm. Gomez Advisors also aggregates these five dimensions into a composite score. We find that most dimensions are positively associated with Web traffic and revenues. A 1% increase in composite score of the five dimensions is associated with a 1.66% increase in the percentage of Internet population reached and a 0.84% increase in revenues for the average firm. However, costs of providing superior online experiences seem to offset revenues because we observe no systematic relation between gross margins and online customer experience scores. Moreover, experience scores exhibit strong positive associations with operating expenses and significant negative associations with net income. A 1% increase in the composite score of online experience is associated with a 0.91% increase in operating expenses and a 2.11% decrease in net income for the average firm. Thus, offering a high quality online experience does not appear to be cost-beneficial, at least in the short run. Considering that most e-commerce firms are start-ups attempting to build market share at the expense of current profits, it is perhaps not surprising to find that the costs of providing better 3
online customer experience exceed immediate benefits. To assess whether the investments in the quality of online experience are viewed (by the stock market) as a source of sustainable long run advantage, we examine the association between the online scores and the market to book ratios. In general, we do not find any systematic relation between online experience scores and market to book ratios. Thus, there is no evidence to suggest that capital market participants view better online customer experiences as a persistent competitive advantage in the long run. Our second research objective relates to the understanding of the existence, duration and sources of competitive advantage currently enjoyed by pure e-commerce firms over their predominantly offline “clicks and mortar” competitors. Amazon so revolutionized the book industry that bricks and mortar companies were afraid of being “Amazoned” (that is, upstaged by an Internet start up). Responding to the entry of pure play e-commerce firms, incumbent offline firms in many industries have set up their own e-commerce ventures. When we compare the scorecards of all online and offline firms, we do not find overwhelming evidence to indicate that click and mortar firms are being “Amazoned.” Composite online scores of pure Internet firms exceed those of click and mortar firms in only 6 of the 19 industries: banks, computer sales, furniture, gifts, mortgage and toys. The score gap is attributable to aggressive pricing, more effective community building and store personalization strategies followed by Internet firms when compared to their click and mortar competitors. In sum, our study suggests that superior online experience can give pure e-commerce firms only a transient competitive edge, at best, in the battle for the domination of online commerce. This study contributes to the literature on the value-based management perspective advocated by Ittner and Larcker (2000) in their recent review of empirical research in management accounting. The value-based management perspective encourages researchers to identify the financial and operational drivers leading to increased shareholder value.
By
identifying the specific factors that cause costs to arise or revenues to change, value driver 4
analysis is expected to improve resource allocation and performance measurement. Consistent with this perspective, we examine the impact of online customer experience -- a key competitive tool in e-commerce -- on performance measures such as Web traffic, revenues, costs and marketto-book ratios. Moreover, most previous studies have examined the performance implications of customer satisfaction in the traditional bricks and mortar world (see Ittner and Larcker (2000) for a review). Web based commerce enables rating agencies like Gomez Advisors to gather data on multiple attributes of customer satisfaction that are either not applicable (e.g., Website ease of use, site personalization, virtual community building) or not easily available (e.g., data on dynamic selling prices charged retailers) in the bricks and mortar world. Evidence related to the revenue-relevance or value-relevance of online customer experience measures also contributes to the emerging literature on the value-drivers of Internet stocks (Demers and Lev 2000, Hand 2000a and b, Rajgopal et al 2000 and Trueman et al 2000a and b). The remainder of the paper is organized as follows. Section 2 discusses the data gathered by Gomez Advisors to capture the five dimensions of online experience and the hypothesized links between these five dimensions and firm performance. Section 3 presents the empirical specifications tested. Section 4 describes the data, descriptive statistics and results of the empirical specifications. Section 5 examines the differences in online customer experiences between pure online firms and their clicks and mortar counterparts while section 6 provides some concluding remarks. 2. Gomez Online Customer Experience Scores and their Links to Performance We begin by discussing the data set of customer experience scorecards compiled by Gomez Advisors, a leading rating agency.1 Gomez Advisors covers a firm’s Website if the firm
1
Bizrate.com and Forrester.com also provide scorecards on online customer experience e-commerce firms. Bizrate provides consumer’s perceptions whereas Forrester and Gomez Advisors provide expert ratings of online customer experience. Forrester and Gomez both rank firms on various dimensions; however Gomez Advisors covers significantly more public firms (51 e-commerce firms compared to 20 by Forrester). Bizrate does not break out its 5
operates in the national market in an industry that meets certain (undisclosed) minimum standards of service in terms of the breadth and depth of product sold. Data collection methods employed by Gomez Advisors include a direct examination of the Website, monitoring the performance of the ranked firm’s secure and non-secure Web pages every five minutes, mock transactions and customer service interaction over the telephone and the Internet. The ranked firms are also required to fill out a supplemental questionnaire. Data thus collected feeds into 150 to 250 criteria for every ranked firm and is condensed into a score for each of five dimensions of online customer experience: ease of Website use (EASE), customer confidence (TRUST), onsite resources (SELECTION), relationship services (REL) and price leadership (COST). Gomez Advisors publishes a score on a scale of 1 to 10 every quarter on these five aspects for the firms they cover. These five scores are aggregated by Gomez Advisors in an unspecified way into a composite score (CSCORE) for each firm. In the following paragraphs we discuss the five dimensions in greater detail. Each paragraph starts with a description of the methodology adopted by Gomez Advisors to measure a particular dimension. This description is followed by an analysis of the links between that dimension and firm performance. Ease of use (EASE): Top ranked firms in this category have an intuitive layout with tightly integrated content, useful demonstrations and extensive online help. Gomez Advisors examines the functionality of the site, availability of online help, a glossary of terms, a list of FAQs (frequently asked questions), the degree of simplicity of account opening and the transaction process, consistency in Website design and navigation, and tight integration of data providing efficient access to information commonly accessed by consumers.
customer ratings by dimension. Moreover its coverage seems limited (e.g., the firm did not have a rating for Amazon.com in the books category although Amazon.com is the leader in this category). 6
We posit that low barriers to entry, the lack of location-based advantages (i.e., competition is a click away) and low switching costs (Shapiro & Varian, 1999) on the Internet may make e-commerce firms more dependent on the willingness of customers to remain on their Website and undertake a commercial transaction. In a recent customer survey conducted by Cognitiative, an Internet consulting firm, 54% of respondents ranked navigational ease as the most important reason why they patronize an online business regularly (The Wall Street Journal, 11/22/1999). If improving ease of Website use reduces the cost of acquiring new customers or creates more loyal repeat customers, such improvement would boost firm performance via higher traffic and greater profitability (Anderson et al, 1994; Reichheld and Sasser 1990). Customer confidence (TRUST): The leading firms in this category operate highly reliable Websites, maintain knowledgeable and accessible customer service organizations and provide quality and security guarantees. Gomez Advisors investigates the posted availability of customer service via phone, e-mail, and branch locations, privacy policies, service guarantees, fees and explanations thereof. Moreover, test phone calls are made and e-mails are sent to customer service units covering simple technical and industry specific questions. The responses are measured in terms of the quality, speed, and accuracy. Each Website is monitored every five minutes for speed and reliability of both public and secure areas of the site. Other factors such as technological abilities, technological independence, years in business, years online and membership in trade organizations also contribute to a higher rank on the customer confidence dimension. Several characteristics of the Internet intensify the importance of customer confidence or trust in Internet-based exchange relationships. The novelty of the Internet creates pervasive uncertainty among buyers and sellers, which in turn, increases the importance of trust in mediating commercial transactions among them (Weigelt & Camerer, 1988). To illustrate, consumers rely on numerous cues in making purchasing decisions. The marketing literature has 7
shown that although consumers use both intrinsic cues (e.g., ingredients, taste, and texture) and extrinsic cues (e.g., price, packaging, and labeling) as indicators of product or service quality (Richardson, Dick & Jain, 1994), they rely more on intrinsic cues in their purchase decisions. Since the Internet deprives consumers of intrinsic cues, it increases their transaction risk. Hence, a firm’s ability to signal trust, and thus engender customer confidence, becomes more important as a guarantor of quality (Shapiro 1983, Yoon, Guffey & Kijewski 1993).2 Furthermore, e-commerce firms employ innovative software tools such as email alerts, chat rooms, and collaborative filtering to deliver value to consumers in ways that have not been economically viable in traditional physical settings (Hamel & Sampler, 1998; Kotha, 1998).3 Hence, cognitive schemas used by customers in the offline world may not smoothly translate to Internet-based transactions. Therefore, providing explicit statements of privacy policies, highly reliable Websites, information on quality and security guarantees aimed to engender customer confidence may be of greater importance in inducing online purchasing decisions than offline buying (Ward, Bitner & Barnes, 1992). Thus, efforts to enhance customer confidence would attract new customers and retain older customers and thereby reduce customer acquisition costs and hence, improve firm performance. Onsite Resources (SELECTION): This dimension of the ranking process measures the range of products and services that the ranked firm carries. Firms are also ranked on whether the Website provides detailed information on the product through electronic forms and information look-up facilities. Onsite resources signal the breadth and depth of product and service offerings available on a firm’s Website. This dimension measures the richness of product and service information (e.g., product reviews) that a firm is able to assemble on its Website. 2
A case in point is Ebay’s decision to suspend an art-seller from their site after he artificially inflated the sale price (The New York Times, 5/11/2000). 3 With collaborative filtering, a site collects data a visitor provides about her taste in say, books or movies. The filtering software compares her preferences with those of thousands of other people and suggests fresh books or movies that the visitor might like. 8
Given that the Internet provides “infinite shelf space,” many pure e-commerce firms explicitly tout selection as one of their most important value propositions vis-à-vis their traditional brick-and-mortar counterparts. For example, the largest superstore of Barnes & Noble can only accommodate about 175,000 books, whereas Amazon.com claims that its virtual store carries over 3 million books (Kotha, 1999). Based on such claims by leading online retailers (e.g., Amazon.com and CDNow), many customers now expect that they should be able to enter a site and see the whole catalog (The Wall Street Journal, 11/22/1999). Hence authoritative selection is one strategy that an online firm can employ to attract customers to its Website (Evans and Wurster 1999). A few exemplar online retailers are then able to induce or entice such customer traffic into creating content (onsite resources) such as customer book reviews or product reviews. Unlike online retailers, firms such as drkoop.com use content (in this case health related content) to signal the breadth and depth of their onsite resources. Such content is generated internally or aggregated from multiple sources. Although expensive to generate, it is the breadth and depth of such onsite resources that attracts customers to the site. The online firm then aims to monetize this traffic via advertising revenues from various sponsors. In sum, the greater the onsite resources a firm has been able to assemble, the greater the firm’s potential to attract new and retain old customers leading to greater traffic, greater sales and improved financial performance. Relationship Services (REL): The relationship dimension measures the ability of the firm to build electronic relationships through personalization by enabling customers to make service requests and inquiries online and through programs that build customer loyalty and a sense of community. Gomez Advisors examines the availability of advice, tutorials, ability to customize a site, whether the customer data is re-used to facilitate future transactions, support of repeatbuying including offer of frequent buyer incentives. 9
E-commerce firms invest in relationship services to motivate repeat purchases from their customer base and decrease the customer’s propensity to switch to a competitor. Online firms build customer relationships primarily through virtual communities and site personalization.4 Activities aimed at building a virtual community heighten consumer involvement with the firm, create a sense of collective belonging (Hagel & Armstrong, 1997) and hence generate a critical mass of customers and make it difficult for new entrants to draw customers away to other communities.5 Personalization is another important strategy undertaken by many online firms to generate repeat buying (e.g., Amazon.com, CDNow, and Yahoo). The Internet offers means for “memorizing” detailed personal knowledge about exchange partners in transactions. Many ecommerce firms have taken advantage of personalization and interaction technologies to develop relationships with customers that were both personal and communal and thereby increase customer loyalty. The “Eyes” program run by Amazon is a personal notification service in which customers can register their interests in a particular author or topic. Once customers register, they are notified each time a book by their favorite author, or topic of interest is published. Personalization increases costs of switching to the nearest competitor and thus creates loyal customers for the firm leading to higher traffic, lower customer acquisition costs and better firm performance.6 Cost (COST): The cost dimension rates the cost competitiveness of purchasing a typical basket of goods or services sold on a site. Costs include a basket of typical services and 4
Virtual communities are online forums that include contributions from, and encourage discourse among, specific sets of like-minded netizens (Kotha 2000). 5 For instance, Amazon offers space for readers to post their “own” reviews. It then steps out of the way and lets its customers sell to each other. Thus, customers themselves (along with the firm’s editors) create much of the editorial content on the firm’s site. As the content grows, it attracts others to add to the richness of the mix, thus creating a virtuous cycle (Economist, 1997; Kotha 1998). 6 Amazon.com has placed personalization at the core of its relationship with consumers. To this end the company developed proprietary technologies dedicated to creating personalized e-mails, compiling customer purchase histories, and recommending additional books based on customers’ manifest preferences. Explains Jeff Bezos, CEO 10
purchases and added fees due to shipping and handling. Firms with higher cost scores offer services and costs at a lower cost. A low-cost strategy, where the total cost of ownership for a typical basket of services sold on the site is less than those available via a competitor’s site, represents an attempt by an online firm to generate a competitive advantage by becoming a price leader. In pursuing this approach, the emphasis is on the efficient and rigorous pursuit of cost reduction from all possible sources (Porter, 1980). To some firms (e.g., buy.com) that compete on low price, the allure of the Internet is simple: Online technologies provide a low-cost, extremely efficient way to display merchandise, attract customers and handle purchase orders (Wall Street Journal, 11/22/1999). To price-sensitive customers, such firms offer the appropriate value proposition and thus generate a competitive advantage vis-à-vis their competitors who emphasize other dimensions of the online experience. Hence, ceteris paribus, price leadership would generate greater traffic and sales. If the firm makes up these lower costs by greater volume, price leadership would be positively associated with long run profitability. 3. Tests of association between online customer experience and firm performance In this section we describe the tests that associate various dimensions of the online experience and firm performance. We consider three sets of performance measures in our empirical analysis: (1) a non-financial measure namely Web traffic; (2) financial measures such as sales, gross margin and net income, and; (3) a market based measure – the market to book ratio. Although section 2 broadly argues for a positive association between superior online experiences and long run profitability, our empirical measures of profitability constitute net income and gross margin assessed over two quarters. This is because our sample consists predominantly of start up firms that have been public for just about a year or so. Because firms
of Amazon.com: “We also want to ‘redecorate the store’ for every customer. We can let people describe their preferences, analyze their past buying patterns, and create a home page specifically for them.” (Kotha, 1999). 11
have to incur significant costs to generate superior online experiences, the costs of providing such experiences may exceed short run benefits in the form of current revenues. Hence, it is important to note that the ex-ante impact of superior online experiences on short run profitability is unclear. We investigate the relation between long run profitability and superior online experiences using the market-to-book ratio as a proxy for expected long run profitability. We discuss the empirical specifications along with the predictions in the following paragraphs. Web Traffic and Income Statement Performance Measures E-commerce firms need to generate traffic to survive. The primary objective behind investments in the discussed strategies is to attract new customers to the site and to retain old customers. Both the acquisition and retention of customers would create greater Website traffic. Hence, we posit a positive association between the online customer experience and Web traffic. Web site traffic is measured as the extent of reach (REACH) of the Internet population, i.e., unique visitors to a Web site as a percentage of total estimated Web population. We estimate the following regression model: REACHjt = γ0 + γ1 DIMENSIONjt +γ2 INDUMjt + γ3 QTRDUMjt + ϕjt
(1)
where j and t are firm and time subscripts respectively. DIMENSION refers in turn to EASE, TRUST, REL, SEL, COST and CSCORE. Industry identifiers labeled as INDUM accounts for unobservable inter-industry differences that might affect traffic.7 A quarter dummy, QTRDUM, is added to the specification to account for unobservable time-specific differences in REACH across various quarters. As discussed in section 2, we expect the coefficient on each DIMENSION to be positive. Revenue (REV) is a closely followed measure for pure online firms. In general, we would expect a firm with easy to use Websites, greater levels of customer confidence, better
12
onsite resources and greater relationship services to be able to attract more Web traffic and hence, potentially sell more goods and services than the average firm. Hence, firm-quarters with greater EASE, TRUST, SELECTION, REL and COST scores would be associated with higher revenues. Thus, in the following regression model, we expect the coefficient on DIMENSION to be positive. REVjt = γ0 + γ1 DIMENSIONjt +γ2 TASSjt + γ3 INDUMjt + γ4 QTRDUMjt + ϕjt
(2)
Total assets (TASS) is added as scale control. We also introduce REACH as an additional control variable in a different version of equation (2) to assess whether online experience scores explain cross-sectional variation in revenues after controlling for the variation explained by REACH. This sensitivity check is potentially interesting because previous research has documented that Web traffic accounts for a significant proportion of the cross-sectional variation in revenues (see Rajgopal et al. 2000; Trueman et al. 2000b). As before, industry and time dummies are added to account to unobservable industry- and time-specific differences in the dependent variable. Next, we examine the relation between gross margin (GM), measured as sales minus cost of goods sold, and the five aspects of the customer experience. We do not predict the direction of this association because an assessment of whether the benefits of investing in online experience, i.e., revenues, exceed costs of such dimensions embedded in cost of goods sold is an empirical issue. Hence, we do not predict the sign on γ1 in model (3): GMjt = γ0 + γ1 DIMENSIONjt +γ2 TASSjt +γ3 INDUMjt + γ4 QTRDUMjt + ϕjt
(3)
Contrary to the common perception that anybody can set up a Web store, building navigable and personalized Web storefronts and generating customer confidence is an expensive endeavor. Moreover, many firms have sought to increase market share by spending lavishly on 7
Gomez classifies e-commerce firms under six major industry groups: personal finance, shopping, health care, computers and office equipment, automobile, travel, home and garden and dynamic pricers which cover auctions 13
customer acquisition (see Demers and Lev, 2000). Hence, we hypothesize that various dimensions of the online customer experience are positively associated with the firms’ total operating expenses (OPEX). Operating expenses typically include marketing and sales expenses, product development, general and administrative expenses, compensation and amortization of goodwill and other purchased intangibles during mergers and acquisitions made by the firm. Regression model (4) is run to test the discussed relations: OPEXjt = γ0 + γ1 DIMENSIONjt +γ2 TASSjt + γ3 INDUMjt + γ4 QTRDUMjt + ϕjt
(4)
We expect γ1 in (4) to be positive. As before, we include total assets as a control variable to control for scale effects. For completeness, we also examine the effects of online experience scores on net income (NI) (see equation (5) below). NIjt = γ0 + γ1 DIMENSIONjt +γ2 TASSjt + γ3 INDUMjt + γ4 QTRDUMjt + ϕjt
(5)
As with predictions for model (3) above, we do not predict the direction of the association between online experience and net income. It is ex ante unclear whether the costs of providing better online experience exceed the benefits thereof via greater margins.
Market to Book Ratio Examining the associations between online scores and income statement items cannot tell us whether investments in online experience create long-term value. The marketing literature suggests that exemplary online shopping experiences can lead to greater customer loyalty and hence secure future revenues (Fornell 1992, Rust et al 1994, 1995). Moreover, better online experiences could potentially reduce the cost of future customer transactions, new customer acquisition, decrease price elasticity of demand and minimize the likelihood that customers will defect if quality falters (see Anderson et al 1994). The resource-based view in the strategy literature is also consistent with this point of view (Barney 1991, Wernerfelt 1994, Mazvancheryl
and buying services. Our industry dummies reflect the membership of sample firms in these industry groups. 14
et al 1999). From this perspective, superior business performance results from access to superior, inimitable resources and/or skills to use these resources better than others.8 Based on these arguments, we would expect to observe positive associations between online experience scores and the market-to-book ratio of the firm. However, skeptics argue that superior online shopping experience can be imitated by online and offline competitors. If the barriers to imitating superior online experiences were low, we would expect that any superior returns to provision of better online experiences would be competed away (Varian 1992, Tirole 1988). Hence, there can be no excess rents accruing to a particular firm on account of superior online experiences. Therefore, we would expect zero or even negative relations between online scores and the market-to-book ratio. We estimate the following equation to assess whether superior online experiences create sustainable long-term competitive advantages: MV/BVjt = γ0 + γ1 DIMENSIONjt +γ2 NI/BVjt + γ3 INDUMjt + γ4 QTRDUMjt + ϕjt (6) The intercept and net income scaled by book value of equity (NI/BV) can be viewed as controls for financial information motivated by Ohlson (1995). We expect γ1 to be positive if superior online experiences create long-term value. If investors do not expect excess rents to accrue to a particular firm on account of superior online experiences, we would expect γ1 to be zero or negative. 5. Data, Descriptive Statistics, and Results Data We begin with the universe of firms for which Gomez Advisors provides quarterly scorecards across the five dimensions of the online customer experience. We obtain scorecards for Winter 1999 and Spring 2000 quarters from Gomez Advisors’s Website (www.gomez.com). 8
For example, Amazon has developed a vast and unique database of customers’ preferences and buying patterns that will enable the firm to offer “mass customized” services at no incremental cost (Economist 1997). 15
Because Gomez Advisors releases scores for each industry on various dates throughout a quarter, we match these scores with accounting data for the quarter in which the scores are released.9 For example, Spring 2000 scores for mortgage brokers were released on February 11, 2000. Hence, for a calendar year mortgage broker, accounting data for sales, expenses, gross margin from the first quarter 10-Q are lined up with the Spring 2000 experience scores. Market value of equity is measured on the last day of the fiscal quarter in which the scores are released. Hence, most market values corresponding to the Spring 2000 quarter are as of March 31, 2000.10 Our initial sample of 927 observations comprises of 550 online and offline firms. Of these, we find only 51 firms that operate predominantly online and trade in public markets. Following Hand (2000), we classified a publicly traded firm as a pure online firm if the firm was a part of the Internet stock list compiled by www.internet.com. We eliminated 4 firms that derived less than 50% of revenues from online operations. The final list of the 47 pure Internet firms used in the analysis is reproduced in Panel A of Table 1. We hand collect all financial information variables from SEC 10-K and 10-Q filings from the SEC’s EDGAR database at the www.sec.gov Website. Stock prices are obtained from www.yahoo.com. Data on Web traffic are obtained from Media Metrix. Media Metrix provides information on unique visitors to Websites and reach of Websites on a monthly basis. We use the average quarterly Website reach as our measure of Web traffic.11 Due to the unavailability of financial data or stock price data, we were left with 75 usable firm-quarter observations. Panel
9
We came across 6 firms where Gomez provides scores in multiple product categories (Amazon, buy.com, Netbank, Value America, Yahoo, barnes and noble.com) For example, Gomez scores for Amazon are available for books, music, videos, toys, electronics and auctions. In such cases, we use the equally weighted average of these scores as the independent variable in our empirical specifications. Because segment disclosures of product-wise sales are patchy or non-existent we cannot use weighted average scores. 10 We are in the process of collecting Gomez scores for Summer 2000. This would allow us to incorporate the impact of the April 2000 stock market “crash” into our analysis. 11 For a detailed description of the methodology used by Media Metrix in determining Web traffic measures see www.mediametrix.com (see also Trueman, et al. 2000a). 16
B of Table 1 provides descriptive statistics for the relevant dependent and independent variables used in the study. Descriptive Statistics The first part of panel B provides the descriptive statistics on the scores for various dimensions of e-commerce customer experience. Note that we have only 39 firm-quarter observations for the COST dimension. This is because Gomez Advisors does not rate the COST dimension in some industries (apparel, furniture, gifts, health advice, home buying, insurance, sporting goods). It is evident from panel B that the scores display modest variation. The standard deviation scaled by the mean score ranges from about 23% for the ease of use (EASE) dimension to 30% for the relationship (REL) dimension. The overall score (CSCORE), which is a weighted average of the scores on the individual dimensions, displays the least cross-sectional variation (standard deviation/mean is 18%). Limited variation on the scores might dampen the power of the empirical tests in detecting significant relations between the scores and the performance measures. Such limited variation is probably on account of restricting the sample to pure Internet firms. Moreover, Gomez probably chooses to cover well-followed Websites that meet certain minimum thresholds of customer confidence and reliability. Hence, self-selection in firm coverage possibly precludes more cross-sectional variation in reported scores. Panel C of Table 1 presents the correlation matrix of the overall score (CSCORE) and its component dimensions. Not surprisingly, CSCORE is highly correlated with component dimensions. Also note that some of the individual dimensions display high correlation with one another (for example, correlation between SELECTION and EASE is 0.57 and the correlation between REL and EASE is 0.45). Such correlation is not entirely unexpected. For example, firms that invest heavily in community building and site personalization are also likely to have navigable Web sites.
17
The descriptive statistics related to financial measures of the sample firms reveal some interesting regularities. The median firm has a high market to book ratio 4.17 suggesting substantial future growth opportunities. The market value of the median firm is $317.54 million. The median firm sells $15.1 million of goods and services and reports a gross margin of $3.7 million (approximately 25% margin on sales). It is interesting to note that the median firm incurs $31.38 million in operating expenses, approximately half of which ($14.42 million) is spent on marketing and advertising. Results We begin by exploring the implications of various dimensions of online customer experience scores for generating Web traffic. That is, we estimate equation (1) that relates Gomez scores to REACH and present the results in Table 2.12 The results in Table 2 indicate a significant positive relationship between the overall CSCORE and Web traffic. All dimensions with the exception of TRUST are positively related to Web traffic. Thus, offering superior online customer experience by way of navigable Web sites, broad selection of goods and services, effective relationship services and good prices attracts more traffic to Websites. To provide an intuitive feel for the magnitude of the coefficients, we compute the sensitivity of scores to traffic. Given that the mean reach is 4.46% and the mean composite score is 6.36 (see panel b of table 1), the coefficient of 1.18 on CSCORE implies that a 1% increase in the composite score increases the REACH of the mean firm by 1.66%.13 Similar sensitivities for a 1% increase in EASE, SELECTION, REL and COST are 1.47%, 1.16%, 1% and 1.03% respectively. Whether such experiences induce Web surfers to buy goods and services is investigated in model (2), results related to which are presented in Table 3. 12
We delete outlier observations representing observations with R-student greater than the absolute value of 3 in all our regression estimations. 13 The coefficient of 1.18 represents δREACH/δCSCORE. The sensitivity of REACH to a 1% change in CSCORE is computed as the (δREACH/REACH)/(δCSCORE/CSCORE) or (δREACH/δCSCORE) x (CSCORE/REACH). 18
Results presented in Table 3 suggest that the overall CSCORE is strongly associated with revenues. In particular, easy to use Websites and greater selection of goods appear to drive sales. These relations are robust to the introduction of Web traffic as an additional explanatory variable. Consistent with prior research (Rajgopal et al., 2000, Trueman et al., 2000b), Web traffic is significantly related to revenues across all estimations. While price leadership is associated with sales in the model estimated without traffic, that relation disappears when traffic is introduced into the model. A 1% increase in the composite scores increases the mean firm’s revenues by 0.84%. Similar sensitivities for a 1% increase in EASE and SELECTION are 0.36% and 0.65% respectively. In sum, superior customer experience is associated with greater revenues. Whether such experiences generate gross margins for e-commerce firms is investigated in specification (3). Results related to specification (3) are presented in Table 4. There is no systematic relation between online experience scores and gross margins. This suggests that costs of providing superior online experiences embedded in the cost of goods sold number perhaps offset the revenues generated by such experiences. Examples of such costs include the provision of gift certificates to stimulate greater buying (one form of price leadership) and the costs of generating content to attract buyers. Thus, improving the quality of online customer experience seems to have no impact on gross margins. However, improving online experience appears to increase operating expenses and hence depress net income of e-commerce firms. Results of specifications (4) and (5) presented in Table 5 support this observation. All dimensions of online experience, except SELECTION, exhibit strong positive associations with total operating expenses and strong negative associations with net income. A 1% increase in the composite score increases operating expenses by 0.91% and depresses net income by 2.11%. This result is consistent with these firms investing resources in
When the mean CSCORE of 6.3 and mean REACH of 4.46% are substituted in the derived relation, the resultant sensitivity is 1.66 (1.18 x 6.3/4.46). 19
improving customer experiences in the short-run to acquire and retain customers so as to maximize the long run lifetime value of earnings that can be derived from the customer base. If investments in better customer experience were likely to create a sustainable long run advantage, we would expect to observe positive associations between Gomez scores and the firms’ marketto-book ratios. An analysis of such association is presented in Table 6. In general, we observe no systematic relations between Gomez scores and firms’ marketto-book ratios (see part A of Table 6). It is interesting to note that the coefficient on earnings scaled by book value is negative and strongly significant in all specifications. Consistent with Hand (2000a), this result probably indicates that market values of Internet firms increase with investments in marketing and research and development. Hence, we investigate whether a decomposition of net income into gross margin (GM), operating expenses (OPEX) (such as marketing and R&D expenditure) and other income or expenses (OTH) affects the results. Further, we also investigate whether traffic, a variable that have previously been shown to be value relevant (Trueman et al., 2000, Hand 2000b and Rajgopal et al. 2000), is a potential omitted variable that might affect the significance of online experience scores. Results presented in part B of Table 6 suggest that the earlier inferences are unchanged when net income is decomposed and traffic is introduced as an additional variable. Thus, we find no evidence to indicate that the quality of the online customer experience is perceived by the stock market as a long run competitive advantage for pure e-commerce firms. 6. Clicks versus Mortar: A comparison of online and offline firms The empirical analysis in sections 4 and 5 was restricted to e-commerce firms that operate predominantly via the Internet. We are unable to conduct a similar investigation for clicks and mortar firms (firms that transact business offline but have some Internet based operations) because such firms do not disclose income statements and balance sheets for their Web operations. However, the Gomez Advisors dataset provides a unique opportunity to 20
examine whether there are statistically detectable differences in the online customer experience scores of pure online firms when compared to those of their clicks-and-mortar counterparts. Responding to the entry of pure play e-commerce firms, incumbent offline firms in many industries have set up their own e-commerce ventures. One of the most debated issues in ebusiness is whether pure Internet companies can offer a better e-commerce experience than their offline counterparts who have recently began setting up e-commerce enabled Web storefronts (see Economist, 2/26/00, Wall Street Journal, 4/17/00 and 7/17/00). It has often been alleged that clicks and mortar stores face several large obstacles in responding rapidly to the entry of pure online e-commerce companies in their product space. Examples of such obstacles include a fear of cannibalization (the Internet might displace existing land based sales at lower selling prices) and channel conflict (existing sales forces and intermediaries resist distribution of goods and services through a new channel-the Web). Moreover, the stock market has given lofty valuations to pure e-commerce firms facilitating stock-based acquisitions of competitors and the ability to offer attractive stock options to hire skilled talent. On the other hand, clicks and mortar companies have well-established brand names, supplier relationships and deeper pockets than pure e-commerce companies. Such firms can also leverage their network of land-based stores to offer convenient merchandise returns. If clicks and mortar companies have indeed been “Amazoned,” we should observe statistically detectable differences in the quality of online experience offered by pure Internet firms when compared to their clicks and mortar counterparts. To examine the differences between online and clicks and mortar firms we conduct t-tests of differences in mean Gomez scores and Wilcoxon non-parametric tests. Table 6 presents the results of these tests categorized by industry type and over time. Unlike sections 4 and 5, we can use Gomez scores of both public and private firms for this analysis. As mentioned before, a publicly-traded firm is classified as an online firm if the firm features on the Internet stock list compiled by Internet.com and if more than 50% of the firm’s revenue is earned through the 21
Internet (as reported in its 10-K report). To assess whether private firms are to be classified as online or not, we examine the “About us” tab on the firm’s Website and its’ press releases. A firm that does not have land-based stores is classified as an online firm. For the overall sample, we find that pure online firms appear to dominate clicks and mortar firms on the CSCORE, EASE, SELECTION and REL dimensions irrespective of whether the mean or median scores are considered. The median COST score for pure online firms is statistically higher than the median score for the offline counterpart, suggesting that pure online firms undercut their clicks and mortar counterparts. It is interesting to note that there is no statistical difference in TRUST scores between the two categories. Considering that the average clicks and mortar firm has existed longer than the average online firm, we would have expected higher TRUST scores for clicks and mortar firms. An industry by industry examination of the difference in scores reveals interesting regularities. The CSCORE for pure online firms exceeds that of clicks-and-mortar firms in only 6 of the 19 industries examined: banks, computers, furniture, gifts, mortgage and toys. In the grocery industry, mean overall scores for the clicks-and-mortar firms exceed the corresponding score for pure online firms. Thus, the quality of online customer experience offered by pure online firms does not, on average, appear to completely dominate the experience provided by clicks-and-mortar firms. Differences in scores across various dimensions would identify particular areas in which pure online firms enjoy a comparative advantage. Online firms dominate clicks-and-mortar in two areas: customer relationships REL (8 of 19 industries) and price leadership COST (7 of 12 industries). Thus, pure online firms appear to compete via aggressive pricing strategies and personalization of the e-commerce experience. Finally, we also investigate whether differences in scores between online and clicks-andmortar firms has changed over the two Gomez Advisors reports for winter 1999 and spring 2000. 22
It is interesting to note that the gap in scores between the two firm types has remained relatively constant over the time periods examined. 7. Conclusions This study provides evidence on three interesting research questions: (1) Does the quality of online customer experience affect the performance of pure Internet firms? (2) Does superior online customer experience offer a sustainable long run competitive advantage to pure Internet firms? and (3) Does the quality of online experience provided by pure Internet firms dominate the experience offered by their clicks-and-mortar counterparts? Using Internet scorecards provided by Gomez Advisors, we find that providing easy to use Websites, broader selection of products and services, building customer relationships, and price leadership are positively associated with Web traffic. We also find that improving ease of Website use and providing a broad selection of products and services appears to positively influence sales. However, we find no relation between gross margin and online experience suggesting that the costs of providing a superior customer experience offset the revenues that such experiences might generate. Furthermore, strong positive relations between the quality of the customer experience and operating expenses imply that the costs of providing better online customer experiences do not exceed benefits of such investment, at least in the short-run. However, when the relation between such scores and market to book ratios are examined, we find that investors do not appear to believe that the quality of online experience provides a sustainable advantage even in the long run. Although there is some evidence to suggest that, on average, the quality of online experience offered by pure Internet firms dominates the experience provided by clicks and mortar firms, the dominance is restricted to 6 of the 19 industries examined. There are several limitations to our analyses. First, we treat the quality of online experience as an exogenous variable although such experiences are very likely endogenous to 23
firm performance. Second, we have access to a limited time-series of Gomez scores. This data restriction may affect the power of our statistical analyses. Hence, it is somewhat premature to conclusively infer that superior online experiences do not provide a long run competitive advantage for pure Internet firms. Third, because we rely on experience scores provided by an external agency, these scores are likely to be influenced by measurement error and potential selfselection issues created by the methodology they employ. Analogous to caveats involved in using equity analysts’ forecasts, the experience scores used may be influenced by undisclosed economic relationships, if any, between Gomez Advisors and the rated firms. Finally, we assume that the relation between e-commerce experiences and performance measures is linear but the true functional form that links such experiences to firm performance is unknown. Future research could address some or all of these issues.
24
References Anders, G. 1999. Better, Faster, Prettier: Companies are spending loads of money in hopes of mastering the Internet’s promise. The Wall Street Journal, November 22, p. R6. Anderson, E.W, C. Fornell and D.R.Lehmann. 1994. Customer satisfaction, market share, and profitability. Journal of Marketing, 56, July, 53-66. Barney, J. 1991. Firm resources and sustained competitive advantage. Journal of Management, 17 (1), 99-120. Barrett, A. 2000. Amazon to launch site in France; ready to compete with the locals. The Wall Street Journal, August 30. Demers, E. and B. Lev. 2000. A Rude Awakening: Internet Value Drivers in 2000, Working paper, University of Rochester and New York University. Dreazen, Y. 1999. Firms shaping the new age. The Wall Street Journal, Dec 31, R, 20:4 The Economist. 1997. A survey of electronic commerce. May 10, 1-18. The Economist. 1999. Business and the Internet: The Net imperative, June 26, 1-54. The Economist. 2000. Shopping around the Web: A survey of e-commerce, February 26, 1-54. The Economist. 2000. Too few pennies from heaven, July 1, 2000. Evans, P. and T.S. Wurster. 1999. Getting real about virtual commerce. Harvard Business Review, November-December 1999. Fast Company. 1996. Who is writing the book on Web business? October-November, 132-133. Fornell, C. 1992. A national customer satisfaction barometer: The Swedish experience. Journal of Marketing, January, 1-21. Fiske, S. & Taylor, S. 1991. Social Cognition. New York: McGraw-Hill. Frankel, K.A. 2000. Portals struggle to convert browsers to shoppers. E-Commmerce-A special section. The New York Times June 7. Hagel, J. and A.G. Armstrong. 1997. Net Gain: Expanding Markets through Virtual Communities. Harvard Business School Press, MA. Hamel, G & Sampler, J.1998. The E-Corporation. Fortune, 138 11, pp. 80-93. Hand, J. 2000a. Profits, losses and the non-linear pricing of Internet stocks, Working paper, University of North Carolina, Chapel Hill.
25
Hand, J. 2000b. The role of accounting fundamentals, Web traffic and supply and demand in the pricing of US Internet stocks, Working paper, University of North Carolina, Chapel Hill. Hansell, S. and J.H. Dobrzynski. 2000. Ebay cancels sale in auction of abstract painting. The New York Times, May 11. Ittner, C. and D. Larcker. 2000. Assessing empirical research in managerial accounting: A value based management perspective, Journal of Accounting and Economics (forthcoming). Kotha, S. 1998. Competing on the Internet: How Amazon.com is rewriting the rules of competition. Advances in Strategic Management, 15, 239-265. Kotha, 1999. Amazon.com: Expanding beyond books. University of Washington Business School case. Kotha. 2000. A Note on Virtual Communities. University of Washington Business School case. Kaufman, L. 2000. New species, old struggle. E-commerce-A special section. The New York Times, March 29. Mazvancheryl, S.K., E.W. Anderson and C. Fornell. 1999. Customer satisfaction and shareholder value: The association between ACSI and Tobin’s q, Working paper, SUNY at Stony Brook and University of Michigan. Murphy, I.P. 1997. Web “Communities” a target marketer’s dream. Marketing News, 31, Issue 14, July. Ohlson, J. A. 1995. Earnings, equity book values, and dividends in equity valuation, Contemporary Accounting Research, 661-687. Porter, M. E. (1980). Competitive Strategy. New York: Free Press. Pulliam, S. and M. Mangalindan. 2000. Amazon stock performance put focus on Web's value. The Wall Street Journal, June 27. Reichheld, F.F. and W.E.Sasser. 1990. Zero defections: Quality comes to services, Harvard Business Review, 68, Sept-Oct, 105-111. Rajgopal, S., S. Kotha and M. Venkatachalam. 2000. The relevance of Web traffic for Internet stock prices, Working paper, University of Washington and Stanford University. Richardson, P., Dick, A., and Jain, A. 1994. Extrinsic versus intrinsic cue effects on perception of store brand quality, Journal of Marketing, 58, 4, p. 28. Rust, R.T., A.J. Zahorik, and T.L.Keiningham. 1994. Return on quality: Measuring the financial impact of your company’s quest for quality. Probus, Chicago.
26
Rust, R.T., A.J. Zahorik, and T.L.Keiningham. 1995. Return on quality (ROQ): Making service quality financially accountable. Journal of Marketing, 59, April, 58-70. Shapiro, C. 1983. Premiums for high-quality products as returns to reputations. Quarterly Journal of Economics, 98: 659-681. Shapiro, C. & Varian, H. 1999. Information Rules. Boston, MA: Harvard Business School Press. Stross, R. 1997. Why Barnes and Noble may crush Amazon. Fortune. September 29. Tirole, J. 1988. The Theory of Industrial Organization. The MIT press, Massachusetts Institute of Technology, Cambridge, MA 02412. Trueman, B., M. H. Wong, and X-J. Zhang, 2000a, The eyeballs have it: Searching for the value in Internet stocks, Working paper, University of California, Berkeley. Trueman, B., M.H. Wong, and X-J. Zhang, 2000b, Back to basics: Forecasting the revenues of Internet firms, Working paper, University of California, Berkeley. Varian, H.R. 1992. Microeconomic Analysis. W.W. Norton and Company, New York, NY 10220. Wall Street Journal. 2000. The Bricks fight back: Internet upstarts were going to own the new economy. Not so fast. July 17. R1-R56. Ward, J., M. Bitner and J. Barnes. 1992. Measuring the prototypicality and meaning of retail environments. Journal of Retailing, 68, 2, pp.194 -124. Weigelt, K. and C. Camerer. 1988. Reputation and corporate strategy: A review of recent theory and applications. Strategic Management Journal, 9: 443-454. Wernerfelt, B. 1994. The resource-based view of the firm:Ten years after. Strategic Management Journal, 16(3), 171-175. Yoon, E., H. Guffey and V. Kijewski. 1993. The effects of information and company reputation on intentions to buy a business service. Journal of Business Research, 27: 215-228.
27
Table 1 Sample and descriptive statistics Panel A: Sample of firms Name 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37
Name
1-800-flowers.com Adam.com Amazon.com Ameritrade Autoweb.com Autobytel.com Barnesandnoble.com Bigstar.com Blue Fly BUY.com CDNOW drKoop.com Drugstore.com E*Trade Ebay Egghead E-Loan eToys.com Expedia fatbrain.com Fogdog FTD HealthCentralRX.com HomeGrocer.com HomeSeekers.com Insweb Mortgage.com Net.B@nk Onhealth Outpost Pets.com PlanetRx.com Quotesmith.com Realtor.com SmarterKids.com Sportsline.com Streamline
38 39 40 41 42 43 44 45 46 47
28
Theglobe.com Travelocity Uniglobe.com Value America varsitybooks.com Utrade Web Street WebMD Webvan Yahoo!
Table 1 (continued) Panel B: Descriptive statistics Variable
N
Mean
Std.dev.
Median
1st quartile
3rd quartile
Gomez Scorecard Overall Score (CSCORE)
75
6.30
1.11
6.19
5.56
7.28
Ease of use (EASE)
75
7.16
1.68
7.42
5.62
8.61
Customer confidence (TRUST)
75
6.42
1.53
6.75
5.44
7.48
On-site resources (SELECTION)
75
6.12
1.55
6.20
5.01
7.19
Relationship (REL)
75
5.48
1.65
5.52
4.30
6.75
Cost (COST)
39
6.96
1.92
7.18
5.98
8.46
Web Traffic (REACH%)
75
4.46
10.86
1.30
0.40
Book value of equity (BV) $ mil.
75
286.20
566.86
100.1
39.40
266.28
Market value of equity (MV) $ mil.
75
4506.30
17191.42
317.54
149.87
1283.25
Net Income (NI) $ mil.
75
-28.82
58.70
-13.03
-32.85
-6.86
Market-to-book ratio (MB)
75
20.56
105.23
4.17
2.31
7.99
Sales (REV) $
75
58.56
117.93
15.10
5.55
47.16
Total Assets (TASS) $ mil. Gross Margin (GM) $ mil.
75
796.58
2099.45
129.13
50.91
519.84
75
22.76
57.10
3.70
0.84
14.13
Operating Expenses (OPEX) $ mil.
75
54.36
69.90
31.38
15.01
61.65
Marketing Expenses (MKTEX) $ mil.
72
27.20
36.30
14.42
7.65
29.32
R&D Expenditures (R&D) $ mil.
72
6.35
11.60
2.676
1.03
5.44
Web Traffic and Financials
mil.
3.91
Panel C: Correlation matrix CSCORE Ease of use (EASE)
0.71
Customer confidence (TRUST) On-site resources (SELECTION) Relationship (REL) Cost (COST)
0.69 0.78 0.66 0.49
EASE
0.33 0.57 0.45 0.16
TRUST
0.44 0.23 0.29
SELECTION
0.38 0.37
Notes: Statistically significant correlation coefficients at the 5% level of significance (two-tailed) are indicated in bold.
29
Table 2 Summary regression statistics of regressing Web traffic on online experience scores REACHjt = γ0 + γ1 DIMENSIONjt +γ2 INDUMjt + γ3 QTRDUMjt + ϕjt Variable
(1)
Pred. Sign
(1)
(2)
(3)
(4)
(5)
(6)
?
-4.94 (-1.56)
-3.65 (-1.20)
-0.38 (0.14)
-3.38 (-1.23)
-2.03 (0.84)
-1.99 (-1.23)
CSCORE
+
1.18 (2.59)
EASE
+
TRUST
+
SELECTION
+
REL
+
COST
+
Intercept
Scores
N Adj. R square
0.92 (2.55) 0.32 (0.33) 0.85 (2.54) 0.81 (2.58) 0.66 (3.07) 71 8%
71 8%
71 0%
71 11%
71 7%
37 20%
Notes: 1. For a description of variables see Table 1. 2. Coefficients on quarter and industry dummies have been suppressed. 3. t-statistics are presented in parenthesis. t-values are White (1980) adjusted wherever applicable. Statistically significant coefficients at the 5% level of significance (one-tailed when the sign is predicted, two-tailed otherwise) are indicated in bold. 4. Outliers representing observations with R-student greater than the absolute value of 3 were deleted.
30
Table 3: Summary regression statistics of regressing revenues on online experience scores after controlling for size and traffic REVjt = γ0 + γ1 DIMENSIONjt +γ2 TASSjt + γ3 INDUMjt + γ4 QTRDUMjt + ϕjt Variable
(2)
Pred. Sign
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
?
-29.97 (-0.94)
-47.64 (1.80)
-4.32 (-0.88)
-27.81 (-1.15)
-8.19 (-0.30)
-15.81 (0.68)
-23.09 (-0.85)
-24.68 (-1.07)
19.11 (0.78)
3.91 (0.81)
-23.73 (0.85)
64.49 (1.67)
CSCORE
+
7.88 (1.69)
9.87 (2.55)
EASE
+
2.96 (1.88)
5.92 (1.90)
TRUST
+
3.46 (1.08)
3.60 (1.33)
SELECTION
+
6.28 (1.87)
5.38 (1.88)
REL
+
-1.68 (-0.51)
0.29 (0.11)
COST
+
10.89 (2.89)
-3.39 (0.62)
Intercept
Scores
Controls TASS
+
REACH
+
N Adj. R square
0.02 (10.87)
73 72%
0.02 (12.22) 3.18 (5.46)
0.02 (11.02)
0.02 (12.07) 3.25 (5.38)
0.02 (11.30)
0.02 (12.61) 3.07 (5.07)
0.03 (6.74)
0.02 (11.76) 2.97 (5.00)
0.02 (11.12)
0.02 (12.13) 3.05 (4.95)
0.03 (3.40)
0.01 (5.39) 24.77 (12.33)
73 80%
73 69%
73 79%
73 70%
73 78%
73 71%
73 79%
73 69%
73 78%
36 44%
36 89%
Notes: 1. For a description of variables see Table 1. 2. Coefficients on quarter and industry dummies have been suppressed. 3. t-statistics are presented in parenthesis. t-values are White (1980) adjusted wherever applicable. Statistically significant coefficients at the 5% level of significance (one-tailed when the sign is predicted, two-tailed otherwise) are indicated in bold. 4. Outliers representing observations with R-student greater than the absolute value of 3 were deleted.
31
Table 4: Summary regression statistics of regressing gross margin on online experience scores after controlling for size GMjt = γ0 + γ1 DIMENSIONjt +γ2 TASSjt + γ3 INDUMjt +γ4 QTRDUMjt + ϕjt Variable
(3)
Pred. Sign
(1)
(2)
(3)
(4)
(5)
(6)
?
4.98 (0.27)
11.87 (0.73)
6.31 (0.41)
-3.90 (0.25)
12.68 (0.91)
-1.45 (0.14)
CSCORE
+/-
1.09 (0.41)
EASE
+/-
TRUST
+/-
SELECTION
+/-
REL
+/-
COST
+/-
Intercept
Scores
-0.19 (-0.09) 0.77 (0.42) 2.63 (1.37) -0.43 (-0.24) 1.75 (1.27)
Controls TASS
N Adj. R square
+
0.01 (12.90) 74 78%
0.01 (13.87)
0.01 (13.35)
0.01 (12.45)
0.01 (13.84)
0.01 (19.30)
74 78%
74 78%
74 79%
74 78%
38 92%
Notes 1. For a description of variables see Table 1. 2. Coefficients on quarter and industry dummies have been suppressed. 3. t-statistics are presented in parenthesis. t-values are White (1980) adjusted wherever applicable. Statistically significant coefficients at the 5% level of significance (one-tailed when the sign is predicted, two-tailed otherwise) are indicated in bold. 4. Outliers representing observations with R-student greater than the absolute value of 3 were deleted.
32
Table 5: Summary regression statistics of regressing operating expenses and net income on online experience scores after controlling for size OPEXjt = γ0 + γ1 DIMENSIONjt +γ2 TASSjt + γ3 INDUMjt + γ4 QTRDUMjt + ϕjt
(4)
NIjt = γ0 + γ1 DIMENSIONjt +γ2 TASSjt + γ3 INDUMjt + γ4 QTRDUMjt + ϕjt
(5)
Dependent variable: OPEX Variable
Dependent variable: NI
Pred. Sign
(1)
(2)
(3)
(4)
(5)
(6)
Pred. Sign
(7)
(8)
(9)
(10)
(11)
(12)
?
-11.85 (-0.71)
13.50 (0.85)
-4.95 (0.36)
14.75 (0.97)
14.57 (1.11)
17.31 (1.07)
?
40.29 (2.68)
20.72 (1.43)
16.79 (1.2)
0.54 (0.04)
9.03 (0.73)
-22.71 (-1.6)
CSCORE
+
7.91 (3.23)
+/-
-9.67 (-4.27)
EASE
+
TRUST
+
SELECTION
+
REL
+
COST
+
Intercept
Scores
+/-
3.04 (1.78)
-5.80 (-3.09)
+/-
6.05 (3.77) 2.78 (1.48)
-4.78 (2.93)
+/-
-2.08 (-1.12)
+/-
3.71 (2.12) 4.02 (1.75)
+/-
+
-4.73 (-2.85) -4.10 (-2.16)
Controls TASS
N Adj. R square Notes:
+
0.03 (14.79) 71 80%
0.03 (14.37)
0.03 (15.26)
0.03 (13.26)
0.03 (14.31)
0.03 (6.07)
71 77%
73 81%
73 78%
73 78%
35 56%
0.00 (0.47)
-0.00 (-0.11)
-0.00 (0.12)
-0.00 (-0.11)
0.00 (0.12)
-0.00 (0.12)
72 37%
71 30%
72 29%
72 21%
72 28%
37 25%
1. For a description of variables see Table 1. 2. Coefficients on quarter and industry dummies have been suppressed. 3. t-statistics are presented in parenthesis. t-values are White (1980) adjusted wherever applicable. Statistically significant coefficients at the 5% level of significance (one-tailed when the sign is predicted, two-tailed otherwise) are indicated in bold. 4. Outliers representing observations with R-student greater than the absolute value of 3 were deleted.
33
Table 6: Summary regression statistics of regressing market to book ratio on online experience scores MVBVjt = γ0 + γ1 DIMENSIONjt +γ2 NI/BVjt + γ3 INDUMjt + γ4 QTRDUMjt + ϕjt (6) Part A Variable Financials Intercept NI/BV
Sign
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
+
5.18 (0.44) -22.05 (-3.47)
10.50 (1.02) -22.17 (-3.51)
2.27 (0.23) -22.16 (-3.51)
0.84 (0.09) -22.25 (-3.53)
4.62 (0.51) -22.41 (-3.48)
-32.46 (-2.91) -45.40 (-4.80)
3.74 (0.84)
8.09 (2.06)
+
Scores CSCORE
+/-
EASE
+/-
TRUST
+/-
SELECTION
+/-
REL
+/-
COST
+/-
Decomposed NI/BV GM/BV
+
OPEX/BV
-
Others/BV
+
Omitted variable REACH
+
N Adj. R square
Part B
0.25 (0.88)
(9)
(10)
(11)
(12)
1.79 (0.49)
5.76 (1.58)
4.15 (1.21)
-8.69 (-1.25)
0.17 (0.27) -0.71 (0.55)
-0.58 (-1.18) 0.72 (0.77)
0.48 (1.17) 1.02 (0.84)
-0.18 (-0.43) 0.43 (0.27)
0.11 (0.26) 2.27 (1.70)
73 27%
73 28%
73 27%
73 28%
73 28%
37 38%
-0.25 (-0.32) 10.67 (1.49) -6.67 (-0.59) -8.35 (-1.20)
9.92 (1.40) -6.78 (0.71) -8.18 (-1.29)
10.59 (1.41) -6.37 (0.57) -8.27 (-1.30)
10.79 (1.51) -6.33 (-0.56) -8.04 (-1.26)
10.87 (1.50) -6.77 (-0.49) -8.49 (-1.29)
39.39 (3.63) -0.25 (-0.37) -30.90 (-3.28)
1.17 (12.5) 71 84%
1.15 (12.3) 71 84%
1.17 (12.17) 71 84%
1.17 (12.5) 71 83%
1.17 (12.4) 71 84%
1.39 (2.45) 37 84%
Notes: (1) For a description of variables see Table 1; (2) Coefficients on quarter and industry dummies have been suppressed; (3). t-statistics are presented in parenthesis. t-values are White (1980) adjusted wherever applicable. Statistically significant coefficients at the 5% level of significance (one-tailed when the sign is predicted, two-tailed otherwise) are indicated in bold; (4) Outliers representing observations with R-student greater than the absolute value of 3 were deleted..
34
Table 7 Panel A: Differences in mean and median online experience scores for online and offline firms by industry Co. Type
CSCORE
TRUST
EASE
SELECTION
Mean Median (N=449,478) 5.95 6.08 6.00 6.15 (-0.49) (-0.76)
Mean Median (N=449,478) 5.81 5.90 6.41 6.60 (-5.38) (-5.30)
Mean Median (N=449,478) 4.35 4.47 4.93 5.13 (-5.22) (-5.20)
REL
COST
1
All 0 1 (0-1)
Mean Median (N=449,478) 5.17 5.19 5.52 5.56 (-4.64) (-4.35)
Mean Median (N=449,478) 3.84 4.00 4.11 4.15 (-2.32) (-2.02)
Mean Median (N=226,226) 6.56 6.55 7.01 7.50 (-0.98) (-4.28)
Industry: Personal Finance Co. Type Banks 0 1 (0-1) Broker 0 1 (0-1) Mortgage 0 1 (0-1) Insurance 0 1 (0-1)
CSCORE Mean Median (N=93,25) 5.50 5.41 6.36 6.40 (-6.16) (-4.86) (N=42,57) 5.15 5.12 5.36 5.37 (-0.97) (-1.16) (N=15,26) 3.92 3.83 5.34 5.32 (-5.17) (-3.94) (N=29,28) 4.88 4.90 5.18 5.25 (-1.12) (-0.77)
TRUST
EASE
SELECTION
REL
COST
Mean Median (N=93,25) 5.86 6.04 5.87 5.76 (-0.04) (-0.15) (N=42,57) 6.01 6.06 5.84 6.15 (0.80) (0.48) (N=15,26) 4.12 3.91 5.07 4.94 (-1.76) (-2.14) (N=29,28) 6.22 6.39 6.71 6.89 (-1.50) (1.63)
Mean Median (N=93,25) 6.12 5.92 6.58 6.67 (-1.76) (-2.13) (N=42,57) 4.70 4.60 5.06 5.00 (-1.25) (-1.31) (N=15,26) 3.70 3.92 5.27 4.92 (-3.97) (-2.98) (N=29,28) 5.98 6.40 5.58 5.59 (0.85) (0.89)
Mean Median (N=93,25) 4.47 4.34 5.17 5.23 (-2.57) (-2.35) (N=42,57) 5.17 5.37 5.55 5.79 (-0.99) (-0.96) (N=15,26) 4.38 4.30 4.75 5.06 (-0.75) (-1.04) (N=29,28) 3.29 3.35 4.18 4.47 (-2.41) (-2.39)
Mean Median (N=93,25) 4.02 3.97 4.83 4.60 (-2.91) (-2.45) (N=42,57) 4.16 3.99 3.96 3.72 (0.58) (0.26) (N=15,26) 2.07 1.50 5.31 5.15 (-6.09) (-4.16) (N=29,28) 4.11 4.51 3.93 3.31 (0.40) (0.47)
Mean Median (N=93,25) 6.88 6.94 8.91 9.14 (-8.81) (-6.14) (N=23,32) 6.55 7.00 8.11 8.19 (-3.65) (-3.33) (N=15,26) 5.81 5.51 6.57 6.55 (-1.40) (-1.33) N/A
Notes: (1) For company type 0 represents offline firms while type 1 represents online firms; (2) N/A indicates insufficient observations to conduct analysis; (3) Numbers in parenthesis under the mean column represent t-statistic of difference in means and the number under the median column represents the Wilcoxon z-statistic. Statistically significant t-statistics and z-statistics at the 5% level (two-tailed) are indicated in bold.
Table 7 (continued) Industry: Shopping Industry
Co. Type
CSCORE
TRUST
EASE
SELECTION
Mean Median (N=38,3) 6.19 6.50 6.12 5.67 (0.11) (0.13) (N=5,23) 6.18 6.44 5.27 5.41 (1.11) (0.96) (N=12,20) 5.68 5.96 6.54 6.84 (-1.75) (-2.02) (N=5,14) 5.78 5.41 6.26 6.41 (-0.58) (-0.60) (N=7,13) 5.66 5.30 6.04 6.58 (-0.39) (-0.48) (N=17,14) 6.14 6.52 7.06 7.15 (-2.72) (-2.23)
Mean Median (N=38,3) 5.70 5.76 6.61 6.92 (-1.82) (-1.05) (N=5,23) 8.16 7.90 7.61 7.82 (0.93) (0.54) (N=12,20) 6.88 6.89 7.51 7.62 (-1.03) (-0.68) (N=5,14) 6.93 6.35 7.32 7.10 (-0.51) (-0.60) (N=7,13) 5.38 6.06 7.44 7.79 (-2.11) (-2.14) (N=17,14) 6.93 7.16 8.10 8.12 (-3.06) (-2.63)
Mean Median (N=38,3) 4.52 4.36 4.20 4.19 (0.71) (0.28) (N=5,23) 6.39 5.99 5.05 4.74 (1.61) (1.17) (N=12,20) 4.33 4.18 5.54 5.64 (-2.08) (-1.97) (N=5,14) 5.12 5.27 5.45 5.54 (-0.44) (-0.60) (N=7,13) 3.51 3.35 5.36 5.09 (-1.82) (-2.00) (N=17,14) 4.39 4.49 5.89 7.23 (-2.20) (-2.12)
REL
COST
1
Apparel 0 1 (0-1) Books 0 1 (0-1) Gifts 0 1 (0-1) Music 0 1 (0-1) Sporting 0 1 (0-1) Toys 0 1 (0-1)
Mean Median (N=38,3) 5.42 5.22 5.37 5.26 (0.17) (0.03) (N=5,23) 6.42 6.55 5.98 5.83 (1.03) (0.69) (N=12,20) 4.85 4.98 5.75 5.65 (-2.23) (-2.08) (N=5,14) 5.26 4.92 5.86 5.93 (-1.47) (-1.06) (N=7,13) 4.01 4.00 5.23 5.02 (-1.75) (-1.58) (N=17,14) 5.68 5.64 6.67 7.08 (-2.26) (-2.00)
Mean Median (N=38,3) 4.97 4.72 3.91 3.31 (1.09) (0.88) (N=5,23) 4.85 5.08 4.07 3.55 (1.03) (1.08) (N=12,20) 2.80 2.83 3.66 3.39 (-1.19) (-1.19) (N=5,14) 3.65 4.27 4.96 4.54 (-1.61) (-1.06) (N=7,13) 1.75 1.20 4.40 3.94 (-3.48) (-2.70) (N=17,14) 4.32 4.48 6.26 6.46 (-2.74) (-2.40)
Mean
Median N/A
(N=5,23) 6.27 6.64 7.26 7.78 (-0.69) (-0.66) N/A
(N=5,14) 4.79 4.69 5.33 5.70 (-0.44) (-0.51) N/A
(N=9,6) 7.22 7.83 7.83 7.82 (-0.57) (-0.06)
Notes: (1) For company type 0 represents offline firms while type 1 represents online firms; (2) N/A indicates insufficient observations to conduct analysis; .(3) Numbers in parenthesis under the mean column represent t-statistic of difference in means and the number under the median column represents the Wilcoxon z-statistic. Statistically significant t-statistics and z-statistics at the 5% level (two-tailed) are indicated in bold.
36
Table 7 (continued) Industry: Shopping (continued) Industry
Co. Type
CSCORE
TRUST
EASE
SELECTION
Mean Median (N=6,15) 5.25 5.32 6.20 6.54 (-1.23) (-1.13) (N=10,18) 6.53 6.32 5.84 5.98 (1.16) (0.74)
Mean Median (N=6,15) 7.05 6.78 7.14 7.01 (-0.17) (-0.39) (N=10,18) 6.13 6.15 6.60 6.56 (-0.63) (-0.46)
Mean Median (N=6,15) 4.28 4.19 3.96 3.97 (0.38) (0.51) (N=10,18) 5.17 5.56 5.05 5.27 (0.15) (0.07)
REL
COST
1
Video 0 1 (0-1) Pets 0 1 (0-1)
Mean Median (N=6,15) 5.41 5.27 5.47 5.60 (-0.12) (0.04) (N=10,18) 4.97 4.66 5.12 5.17 (-0.26) (-0.36)
Mean Median (N=6,15) 3.70 3.96 4.29 3.63 (-0.81) (0.00) (N=10,18) 2.42 1.83 3.28 2.91 (-1.03) (-1.06)
Mean Median (N=6,15) 7.60 7.83 6.52 7.12 (1.10) (0.78) (N=10,18) 6.30 7.07 7.53 7.76 (-1.37) (-1.22)
Industry: Computers and Electronics Industry
Co. Type
CSCORE
TRUST
EASE
SELECTION
REL
COST
Mean Median (N=10,12) 4.97 4.96 5.69 5.78 (-2.43) (-2.47) (N=19,22) 4.72 4.79 5.28 5.43 (-1.67) (-1.46)
Mean Median (N=10,12) 7.18 7.53 5.72 5.53 (2.61) (2.18) (N=19,22) 5.99 6.14 6.06 6.12 (-0.18) (-0.16)
Mean Median (N=10,12) 7.15 6.88 7.84 8.27 (-1.28) (-1.29) (N=19,22) 6.60 6.83 6.31 6.40 (0.70) (0.41)
Mean Median (N=10,12) 3.56 3.79 4.43 4.43 (-1.35) (-1.02) (N=19,22) 3.95 3.31 4.35 4.35 (-0.73) (-0.80)
Mean Median (N=10,12) 4.62 4.73 5.57 5.73 (-1.33) (-1.02) (N=19,22) 3.17 3.20 3.63 3.06 (-0.87) (-0.33)
Mean Median (N=10,12) 2.41 1.95 4.72 4.25 (-2.25) (-1.98) (N=11,11) 3.99 3.99 6.43 6.12 (-2.92) (-2.63)
1
Computer 0 1 (0-1) Electronic 0 1 (0-1)
Notes: (1) For company type 0 represents offline firms while type 1 represents online firms (2) N/A indicates insufficient observations to conduct analysis; (3) Numbers in parenthesis under the mean column represent t-statistic of difference in means and the number under the median column represents the Wilcoxon z-statistic. Statistically significant t-statistics and z-statistics at the 5% level (two-tailed) are indicated in bold.
37
Table 7 (continued) Industry: Health Industry
Co. Type
CSCORE
TRUST
EASE
Mean Median (N=4,12) 4.27 4.78 4.97 5.07 (-1.22) (-1.15) (N=9,17) 5.57 5.46 5.58 5.41 (-0.02) (-0.05)
Mean Median (N=4,12) 6.00 6.01 6.23 6.37 (-0.46) (-0.55) (N=9,17) 6.23 6.33 5.70 5.49 (0.99) (1.05)
Mean Median (N=4,12) 5.93 6.27 6.37 6.41 (-0.85) (-0.61) (N=9,17) 5.13 4.90 5.60 5.65 (-0.58) (-0.65)
SELECTION
REL
COST
1
Health 0 1 (0-1) Drug 0 1 (0-1)
Mean Median (N=4,12) 3.79 4.53 4.87 4.50 (-0.90) (-0.30) (N=9,17) 5.11 5.54 5.47 5.89 (-0.47) (-0.38)
Mean Median (N=4,12) 2.13 2.18 3.38 3.32 (-1.76) (-2.11) (N=9,17) 3.93 4.51 4.22 4.15 (-0.48) (-0.38)
Mean
Median N/A
(N=9,17) 7.97 8.15 7.80 8.04 (0.33) (0.03)
Industry: Home and Garden Industry
Co. Type
CSCORE
TRUST
EASE
SELECTION
REL
Mean Median (N=12,25)
Mean Median (N=12,25)
Mean Median (N=12,25)
Mean Median (N=12,25)
Mean Median (N=12,25)
5.15 5.50 5.36 5.28 (-0.44) (-0.39) (N=6,28) 6.60 6.74 7.17 7.22 (-1.37) (-1.04) (N=2,21) 8.15 8.15 6.62 6.90 (3.62) (2.02)
5.33 5.41 5.38 5.35 (-0.11) (-0.03) (N=6,28) 6.53 6.42 7.28 7.39 (-1.58) (-1.49) (N=2,21) 6.82 6.82 5.82 6.53 (1.69) (0.49)
3.24 3.19 4.30 4.39 (-1.87) (-2.20) (N=6,28) 4.44 4.61 6.07 6.17 (-3.77) (-2.37) (N=2,21) 6.00 6.00 4.29 4.45 (1.58) (3.19)
3.71 3.31 3.88 4.00 (-0.36) (-0.52) (N=6,28) 3.62 3.34 4.86 5.02 (-1.58) (-2.04) (N=2,21) 6.67 6.67 4.72 4.66 (1.80) (3.20)
COST
1
Home Furnishing 0 1 (0-1) Furniture 0 1 (0-1) Grocery 0 1 (0-1)
4.15 4.19 4.61 4.56 (-1.63) (-1.27) (N=6,28) 5.50 5.58 6.48 6.40 (-2.79) (-1.97) (N=2,21) 7.31 7.31 5.58 5.69 (1.91) (3.92)
Mean
Median N/A
N/A
(N=2,21) 9.10 9.10 5.27 5.35 (4.08) (2.13)
Notes: (1) For company type 0 represents offline firms while type 1 represents online firms; (2) N/A indicates insufficient observations to conduct analysis; (3) Numbers in parenthesis under mean column represent t-statistic of difference in means and the number under the median column represents the Wilcoxon z-statistic. Statistically significant t-statistics and z-statistics at the 5% level (two-tailed) are indicated in bold.
38
Table 7 (continued) Panel B: Differences in mean and median online experience scores for online and offline firms over time Report Date Winter ‘99
Co. Type 0 1 (0-1)
Spring ‘00 0 1 (0-1)
CSCORE Mean Median (N=230,250) 5.11 5.10 5.56 5.61 (-4.34) (-4.22) (N=202,188) 5.29 5.36 5.54 5.47 (-1.68) (-2.01)
TRUST Mean Median (N=230,250) 5.89 6.04 5.99 6.15 (-0.76) (-1.15) (N=202,188) 6.11 6.24 6.11 6.19 (0.00) (-0.02)
EASE Mean Median (N=230,250) 5.82 5.85 6.58 6.87 (-4.71) (-4.76) (N=202,188) 5.84 5.95 6.36 6.56 (-3.14) (-3.11)
SELECTION Mean Median (N=230,250) 4.37 4.36 5.08 5.19 (-4.23) (-4.13) (N=202,188) 4.63 4.66 5.19 5.17 (-3.28) (-3.28)
REL Mean Median (N=230,250) 3.98 3.89 4.31 4.12 (-1.85) (-1.58) (N=202,188) 4.11 4.18 4.23 4.15 (-0.65) (-0.47)
COST Mean Median (N=136,121) 6.32 6.18 6.74 7.10 (-0.56) (-3.49) (N=81,76) 7.04 7.20 8.00 8.30 (-3.73) (-4.26)
Notes: (1) For company type 0 represents offline firms while type 1 represents online firms; (2) N/A indicates insufficient observations to conduct analysis; (3) Numbers in parenthesis under mean column represent t-statistic of difference in means and the number under the median column represents the Wilcoxon z-statistic. Statistically significant t-statistics and z-statistics at the 5% level (two-tailed) are indicated in bold.
39
