Is stickiness the Holy Grail for e-tailing? In general, stickiness refers to the amount of time a person spends on a Web site during a visiting session (such as, session stickiness) or over a specified time period (such as, site stickiness). Zauberman12 equates stickiness and "within-site lock-in" and uses it to approximate visitors' loyalty to a Web site. The conventional wisdom suggests that stickiness is crucial and can contribute to e-tailers' bottom lines considerably.5 However, the direct economic impacts of stickiness have not been duly examined empirically, particularly from the perspective of consumers' within-session visiting behaviors.
E-tailing offers an exciting global virtual channel for marketing and exchanges. According to the U.S. Department of Commerce, the e-retailing industry has grown at a 29% compounded annual rate between 2000 and 2004, amounting to $81 billion in sales in 2005 and projected to reach $144 billion by 2010.9 Accompanied by this impressive growth is the increasingly fierce market competition that results from greatly reduced search costs, diminished product/service differentiation, and rapidly eroding customer loyalty.2 To survive and excel in this highly competitive market, e-tailers must be effective in converting their Web site visitors into paying customers. Such conversions are challenging, as manifested by the dismal conversion rate (for example, estimated at 2.3% by e-tailing.com) that is likely to continue to decline.
Stickiness serves as a common indicator of customer loyalty to e-tailers. Accordingly, firms have focused on effective Web site design and business strategies to "lock in" visitors by making their Web sites increasingly sticky. Despite the salient beliefs about the business value of stickiness in e-tailing, empirical evidence of its direct economic impacts is surprisingly limited. Moe and Fader7 discuss the importance of stickiness for business profits and advocate the use of visiting behaviors to investigate the relationship. Straub et al.10 also highlight the importance of analyzing prominent visitors' behaviors and particularly suggest the use of click-stream data to examine the effects of essential visiting behaviors on business outcomes, such as online purchases. However, few (if any) studies have examined these effects empirically.
In this study, we use within-session visiting behaviors, recorded by a designated client-side monitoring program, to examine the relationship between stickiness and conversion, an outcome metric directly affecting e-tailers' bottom lines. We respond to the call by Moe and Fader7 by examining session stickiness through analyses of visitors' within-session behaviors.a We measure session stickiness using both visiting session duration and total number of pages accessed during a visiting session.b We empirically test whether stickiness significantly affects conversion, and further investigate whether product category moderates the focal stickinessconversion relationship.
A host of metrics have been developed to characterize or assess visiting behaviors exhibited on a Web site, including page views, number of visits, and number of unique visitors. However, these metrics are aggregates and thus offer little value for supporting business operations or decision making in a direct and tangible way. According to Trueman et al.,11 the page view metric provides some explanatory utility but does not seem to affect business revenues directly. Financial analysts also have become increasingly skeptical of such aggregate metrics partly because they carry few tangible business implications. A disaggregated approach anchored by an analysis of within-session visiting behaviors may advance our understanding of the relationship between stickiness and conversion, and thus can shed greater light on the business practices that foster online purchases.
For visitors, stickiness manifests the "interestingness" of a Web site and can reveal their satisfaction with the Web site to some extent. Hoffman and Novak4 suggest a positive relationship between stickiness and visitors' loyalty to a Web site; that is, the stickier a Web site, the more satisfied its visitors appear. Through the lens of exchange, Agrawal et al.1 associate stickiness with attraction, conversion and retention; for example, the longer a prospect interacts with a product, the more likely he or she is to purchase it. In e-tailing, this suggests that the longer a visiting session is, the more likely the visitor will purchase from the Web site. Competing conceptualizations also exist. For example, according to the power law of practice,5 a shorter visiting session manifests greater exchange efficiency that results from learning through repeated visits. Accordingly, visitors who spend less time on a Web site may be more likely to make purchases online than those who hang around the site for longer periods. Such competing conceptualizations and their unsettling dispositions warrant empirical testing of the relationship between session stickiness and conversion, particularly on the basis of important within-session visiting behaviors.
Consumer behaviors vary with products which can be broadly classified as search or experience goods.c We therefore further examine whether the focal stickinessconversion relationship is robust across different product categories. In addition, many consumers gather product/service information online but choose to execute transactions in familiar channels, such as conventional store outlets. Moe and Fader7 have identified a group of consumers who visit e-tailers' Web sites frequently and extensively but have never made any purchases online; for example, hard-core never-buyers. These never-buyers offer minimal direct financial contributions to e-tailers, which therefore should prioritize the attention and resource allocation towards serving those who are more likely to purchase from their Web sites. As a result, we also assess how well the stickiness conversion relationship holds for online purchasers.
Data Set: We use panel data from comScore Media Metrix (http://www.comscore.com) that consist of detailed click-stream data from 100,000 volunteer households between July and December 2002. Designated clientside monitoring software records detailed browsing and purchases by each household and thereby generates our sample. For each visit, the software records the Web site visited, together with the date/time of visit, session duration, and number of page accessed. In our sample, a transaction is depicted by each product bought, its unit price and the quantity purchased as well as the grand total price of the transaction (for example, a shopping basket). Our data set also has essential household characteristics, including demographic information. In Table 1, we summarize the descriptive statistics of some important household characteristics. Furthermore, we note the significance of e-tailers, which represent more than half of the Web sites that record online purchases and account for 50% of the total sales in our sample.d
Data Cleansing: Data quality represents a common challenge in clickstream data analyses. Our data have relatively high quality because they are collected by designated client-side monitoring software which records each actual visiting behavior in a detailed, page-by-page manner. The collected data are then pre-processed by comScore, which further increases the data accuracy by removing data inconsistent or anomaly data. In addition, we enhanced data quality by excluding from our analyses all visits with session duration of zero minute (for example, the time unit in our data) or without accessing any pages; for example, reflecting errors in data recording or visitors' entering incorrect Web sites.
Firm Selection: We focus on the Web sites with online purchases recorded and select firms with at least 1,000 completed transactions to ensure each studied e-tailer has a sufficient number of purchases for our analysis. We select firms on the basis of data quality, business model, offline presence, and product offering. For example, jcpenney.com in our sample has a great number of visits but extremely few purchases. This discrepancy is suspicious and difficult to reconcile; we therefore do not include jcpenney.com in our analysis. Moreover, some Web sites adopt business models that make them inappropriate for our investigation. For instance, quixstar.com requires customer registration, thereby connecting each customer with a proximate agent. This firm depends heavily on offline customer-agent interactions to facilitate customers' purchases from its Web site. Such offline activities have substantial impacts on customers' online purchases and could bias our analysis of the stickinessconversion relationship, for which we concentrate on the recorded visiting and purchase behaviors. Consequently, we exclude quixstar.com from our analyses. Some firms have extensive presences in non-Internet based channels. Examples include QVC, a popular electronic retailer, which has a predominant TV presence that could influence our testing of the relationship between within-session visiting behaviors and online purchase decisions. In addition, we assess the primary offerings of each e-tailer and classify them according to their defined focusnamely, experience goods versus search goods. This classification supports our intended product category perspective in our analyses. As summarized in Table 2, a total of eight e-tailers are included in our analysis. The conversion rate of these studied e-tailers averages 6.20%, which, in light of our firm selection criteria, is reasonably comparable with those reported by previous studies.7
We perform multivariate ordinary least square linear regression analysis, using conversion rate as the dependent variable.e We control for Internet connection speed, the number of children in a household, the highest age of members of the household, and household income. In this section, we highlight our important analysis results.
Does the length of a visiting session affect conversion? Our analysis shows a significant, positive relationship between session duration and conversion, thus supporting the disposition that the stickier a visiting session, the more likely the visitor is to purchase from it. This finding supports the conventional wisdom embracing the positive value of stickiness for e-tailers. Further analysis by product category shows that the observed relationship between visiting session duration and conversion holds for all experience goods centric e-tailers but not for all e-tailers that specialize in search goods. Overall, our analysis suggests that session stickiness measured by session duration can signify the likelihood of purchase on an e-tailer's Web site, particularly for those concentrating on experience goods.
The weakened relationship we observe for the search goods category might be partly explained by the differential information requirements for consumer decision making. In general, experience goods inherently entail greater quality uncertainty prior to purchase (or consumption) than do search goods; therefore, consumers may have stronger motivations to search for more product/service information. Along this reasoning, long visiting sessions on an e-tailer's Web site that specializes in experience goods may result from visitors' engaging in extensive product information search or alternative product comparison, which manifests their purchase interests or intentions. On the other hand, consumers need relatively less information to assess, compare, or select products pertaining to search goods; therefore, long visiting sessions on a search goods centric e-tailing Web site might not signify their purchase intentions effectively. Our findings caution against using the length of the visiting session to interpret the business value of stickiness and highlight the need to assess its impacts on online purchases through the lens of product offering.
Does the number of pages accessed in a visiting session affect conversion? We find a significant, positive relationship between the number of pages accessed in a visiting session and conversion. Session stickiness, measured by the number of pages accessed in a visiting session, seems to signify a visitor's intention to purchase from an e-tailer's Web site across different product categories. Although the number of pages accessed and session duration are both considered capable of revealing visitors' interest in a Web site, our analysis yields empirical evidence that the number of pages might be more indicative of their purchase likelihood across different product categories. One plausible explanation is that the amount of time a visitor spends on a Web site during a visiting session conceivably is affected by his or her browsing pace and habits, and can be influenced by unexpected interruptionsfactors that do not relate directly to the visitor's interest in the Web site. The actual number of pages accessed is relatively immune to such factors or disturbances. Intuitively, the extensiveness of browsing should correlate positively with the length of a visiting session and can be assessed by the sheer volume of contents viewed during a visiting session. Overall, our comparative analysis shows that session stickiness appears more robust for signifying conversion when it is measured by the number of pages accessed than by the length of a visiting session.
Do purchasers behave differently from non-purchasers? We follow the classification by Moe6 and further our analysis by concentrating on purchasers; for example, those who have made at least one online purchase from any of the studied Web sites. This allows us to scrutinize the stickinessconversion relationship evinced by those who understandably are more valuable to an e-tailer in terms of direct financial contributions through their past and probable future purchases. We find visiting session durations comparable between purchasers and non-purchasers who, however, exhibit noticeably different page browsing behaviors; for example, purchasers accessing more pages in a visiting session than non-purchasers. Taken together, our findings suggest that visitors likely to purchase online appear to access more pages in a more efficient manner than do visitors who lack purchase intentions.
Does product category matter in online purchases? According to our analysis, session stickiness measured by visiting session duration exhibits a significant, positive relationship with conversion for e-tailers that specialize in experience goods, whereas session stickiness measured by the number of pages accessed during a visiting session has a significant, positive relationship with conversion regardless of product category. Purchasers seem to access a comparable number of pages on e-tailers' Web sites specializing in search versus experience goods; but, those who hang around longer periods of time on an experience goods centric Web site do not necessarily request more pages. This finding is intriguing, in that it suggests a probable moderating role of product category in the relationship between stickiness and conversion.
Although it may be intuitive to explain why the number of pages accessed in a visiting session is more indicative of conversion than visiting session duration, it is challenging to reason why visiting session duration can better reveal visitors' direct purchases from Web sites that specialize in experience goods. According to Moe,6 knowledge-building visitors in general spend a long period of time on an e-tailer's Web site but often access fewer pages. All else being equal, these visitors have considerable interests in an e-tailer's offerings and are likely to purchase from its Web site in current or future visits. In the case of experience goods, this visit pattern would reflect visitors' greater interest in learning more about specific products; for example, they might access few selected pages and read them carefully and thoroughly. In contrast, visitors do not need such extensive, detailed reviews of information about products pertaining to search goods, thus not frequently exhibiting "deep reading" behaviors on an e-tailer's Web site.
Overall, our findings support the exchange-oriented disposition by Hoffman and Novak4 and Agrawal et al.1 that stickiness is positively related to online purchase likelihood. Our results appear not supporting the power law of practice which suggests the shorter a visiting session, the greater the exchange efficiency. Plausible reasoning can be gained by a finer-grained analysis of the behaviors exhibited during a visiting session. In general, the duration of a visiting session consists of navigation time (for example, time spent for navigating through the Web site structure, selecting appropriate hyperlinks/paths, and completing the checkout process), and content gathering/review time (for example, time spent for accessing and reading pages about particular products/services). The power law of practice is mostly about the navigation time which decreases as customers get more familiar with a site. On the other hand, the exchange-oriented disposition suggests the reviewing time to increase as customers become more interested in the site. Affected by navigation and page reviewing time, the relationship between visiting session time and conversion is determined by their respective magnitude of influence.
Our results show a positive relationship between stickiness and conversion, hereby demonstrating that navigation time is trivial compared with the page reviewing time. For purchasers, probable reductions in the navigation time are more than offset by an increase in their page reviewing time. Given the increasingly mature Web site design today and people's familiarity using the Internet and various e-commerce Web sites, the postulated decrease of the navigation time to a marginal proportion of the visiting session time is reasonable and justifiable.
Our findings have several important implications. First, session stickiness can signify the likelihood of visitors' making online purchases. In particular, session stickiness offers a better indication of purchase when it is measured by the number of pages accessed in a visiting session than by visiting session duration, though the latter measure may be effective for signaling conversion in scenarios that involve experience goods. Second, product category may moderate the relationship between session stickiness and conversion. Last but not least, online purchasers generally exhibit visiting behaviors that differ from those of hard-core never-buyers. For purchasers, the number of pages accessed during a visiting session is particularly indicative of their purchase likelihood on e-tailers' Web sites that focus search goods.
This study yields some empirical evidence that suggests positive economic impacts of stickiness to e-tailers' bottom lines. Our results show that the number of pages accessed during a visiting session may better signify conversion, and that the effectiveness of stickiness in indicating online purchase likelihood may vary with product category. The positive correlation we observe between the number of pages viewed and conversion suggests that when consumers access more pages in their visits, they become more incline to purchasing from the Web site. In light of these findings, business managers should strive for providing rich and relevant production/service information as well as for motivating and facilitating visitors' accessing these pages. Some plausible strategies include the creation of appropriate "springboard" pages for guiding visitors to popular clusters of informative pages relevant to their interested products/services. Amazon.com's recommender system offers a good exampleproviding multiple, relevant hyperlinks that allow visitors to dive in for "deep reading" of specific pages of interest. Web site developers should also contemplate how to improve page designs to make them more attractive and interactive to visitors. This is particularly important for Web sites that specialize in experience goods (such as clothing). The virtual storefront of Landsend.com exemplifies effective use of interactive and personalized pages to attract consumers' attention and facilitate their interacting with different products.
Our findings highlight some cautions about assessing the business value of stickiness. Business managers should be mindful in their use of stickiness to gauge e-tailing performance, particularly with respect to stickiness measurement or product category. Although it may reveal visitors' purchase likelihood to some extent, session stickiness may not be equally effective in explaining or predicting the probability of converting individuals' Web site visits into completed transactions on e-tailers' Web sites that specialize in search goods.
9. Perez, J.C. U.S. e-tailing sales to have brisk growth through 2010. www.infoworld.com, (2006); http://www.infoworld.com/article/06/02/07/75135_HNetailingsales_l.htrnl?B-TO-C.
a. While recognizing the importance of site stickiness, we focus on session stickiness because it directly relates to within-session visiting behaviors (such as, our intended approach) and is intrinsically more relevant to conversion (for example, our dependent variable) than site stickiness.
b. In addition to the common measure of visiting session duration, we also measure session stickiness using the number of pages accessed during a visiting session, which offers a direct indicator of session stickiness and is highly correlated with visiting session duration.3 Our expanded measure of stickiness is consistent with current practice, which assesses stickiness on the basis of visiting time duration and page requests made in a visiting session (www.whatis.com).
c. We follow the definitions by Nelson (see Information and Consumer Behavior, The Journal of Political Economy 78, 2, (1970), 311329). In general, search goods refer to products for which full or sufficient information about dominant product attributes can be reasonably accessed and learned about prior to purchases, whereas experience goods are products whose quality is difficult to assess appropriately by consumers before making a purchase. Example search goods include books, videos and CDs, while clothing and perfumes are usually considered experience goods.
d. In this study, we define an e-tailer as a provider whose Web site is classified as a "shopping" site in the source data by comScore. Although restrictive to some degree, this definition excludes Web sites that do not support e-tailing.
©2010 ACM 0001-0782/10/0300 $10.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2010 ACM, Inc.
No entries found