GIST: A Model for Design and Management of Content and Interactivity of Customer-Centric Web Sites

Abstract:

• Customer-centric Web-based systems, such ase commerce Web sites, or sites that support cus tomer relationship management (CRM) activities, are themselves information systems, but their design and maintenance need to follow vastly different approaches from the traditional systems lifecycle approach.
• Based on marketing frame works that are applicable to the online world, and following design science principles, we develop a model to guide the design and the continuous management of such sites.
• The model makes extensive use of current technologies for tracking the customers and their behaviors, and combines elements of data mining and statistical analyses.
• A case study based on a financial services Web site is used to provide a preliminary validation and design evaluation of our approach.
• The case study showed considerable measured improve ment in the effectiveness of the company's Web site.
• In addition, it also highlighted an important benefit of the our approach: the identification of previously unknown or unexpected segments of visitors.
• This finding can lead to promising new business opportunities.

Introduction:

• Understanding the end users and their require ments has been the cornerstone of information systems analysis and design.
  • Understanding the factors that lead to user acceptance has received abundant attention by the information systems research community (see for example, Barki and Hartwick 1994; Davis 1989).
  • Most of the time, end users are a well-defined category, either com prising a specific segment of internal constituents (sales force, human resource, etc.) or well-known segments of the extended enterprise of suppliers and customers.
  • User participation, before and during the implementation of information systems, and training are fundamental in facilitating accep tance and increasing the utilization of the systems (Green and Hughes 1986; Lee et al. 1995; McKeenetal. 1994).
• Another characteristic of the more traditional system development lifecycle of information sys tems is system maintenance.
  • Aside from correc tions and debugging that might take place after the system is released, major upgrades or new versions are released very infrequently.
  • In other words, the cycle for incorporating major changes is relatively long.
  • Gaps that are identified between the end users' objectives and motives toward the system and the functionalities that the system actually delivers can only be addressed at very infrequent intervals.
• Customer-centric Web-based systems, such as e commerce Web sites, or sites that support cus tomer relationship management (CRM) activities, are themselves complex, multicomponent, multi tier information systems, but their design and maintenance need to follow vastly different ap proaches from the traditional systems lifecycle.
  • While some of the end-users may be either cur rent customers or potential customers, many of them are non-transactional visitors, often anony mous, that are part of the Internet user population at large.
  • These users may display a wide variety of preferences and motives toward the site, which are very difficult to capture.
• The Internet environment is a marketing channel for which Hoffman and Novak (1996) articulated a need to understand both the transaction and non transaction activity.
  • Non-transaction activities en tail experiential components, whereas transaction (shopping, online banking, etc.) activities are more goal-oriented.
  • Hoffman et al. (1996) classified sites into online storefronts, Internet presence sites, content sites, malls, incentive sites, and search agents.
  • An early analysis of these cate gories indicated that only 18 percent of the sites were online storefronts while the remaining 82 percent were informational or Internet presence sites (Kaul 1995).
  • The prevalence of the experi ential, non-transactional activities conducted on non-transaction site types in the early commer cialization years of the Internet continues to be the pattern, as recently demonstrated by Greenspan (2002): while 59 percent of online visitors conduct transactions, their primary activities continue to be information and communication based.
• The design and administration of Web sites visited by the non-transactional user remain a challenge due to the difficulty in obtaining design require ments from these potential visitors.
  • Their direct involvement during requirements definition and their training as end-users are usually not viable options. One can only successfully listen to the "voice of the customer" when there are well-under stood site audiences, as demonstrated by an interesting study conducted by El Sawy et al (1999).
• The Web environment is implicitly highly dynamic, defying customary geographic and temporal infor mation systems design assumptions.
  • In addition, a Web site is, in essence, competing with various other web sites trying to attract and capture the visitors, which introduces new market dynamics.
  • It is critical to conduct continuous assessment of potential gaps between intentions of users and the delivered experience by the site to maintain its utility.
  • This implies continuous management and updates of content and interaction, as opposed to the sporadic version release paradigm of tradi tional software development environments.
• We consider Web site design and its continuous redesign as a complex product design and product update problem.
  • We see the product itself as a multidimensional set of possible visitor experiences that the firm wants to enable.
  • Each such experience might have been determined by very different motives expressed by very different individuals.
  • On top of all of this, we consider the requirement of real time assessment of gaps between the visitors' intentions and motives and the site offerings.
  • This may lead to constant site redesign and updates.
  • The framework and meth odology we propose are intended to address this very complex product design problem when the users of the product are not well understood in advance.
• In this paper, our goal is to propose an operational framework for continuous redesign, especially for non-transactional sites.
  • While much of what we propose can be used for transactional sites, our approach also works for non-transactional sites, an area of research that has not attracted much research. Our basic research premise is that the design and maintenance of customer-centric Web sites need to be based on a customer-focused approach.
  • Therefore, we propose a Marketing centric approach to design and maintenance of such Web sites using a framework that we refer to as GIST: Gather-lnfer-Segment-Track. In devising and evaluating the proposed framework, we follow the principles of design science and follow the seven guidelines of design science research as suggested by Hevner et al. (2004).
• The collaboration between the Information Sys tems and Marketing disciplines is not new; however, it has become more tightly coupled as role of the Internet is increasingly examined in applied and scholarly research.
  • As Hong et al. (2002) state, e-commerce systems are both an information system and a marketing channel.
  • Each discipline views the electronic environment through its own theoretical frameworks or lens.
  • Further, Stafford (2003) sees the marketing view as more product-oriented and the information systems or technology view as more of a develop ment concept. These multiple perspectives have led each discipline to claim different aspects of the e-commerce paradigm.
• The marketing discipline's contribution to the IS new concept development lens is its identification of "and meeting human and social needs... profitably" (Kotler 2003, p. 3).
  • More specifically, marketing requires understanding the customer's (either a consumer or business) expectations from the firm and delivering products or services that meet or exceed these expectations.
  • This, coupled with the IS lens for understanding the end users' requirements (the cornerstone of information systems analysis and design), creates a powerful approach to understanding and designing customer-centric Websites.
  • However, surprisingly very little research has focused on integration of the two lenses.
• To the best of our knowledge, GIST is the first practical methodology that combines marketing research findings for the online context that identifies determinants of customer satisfaction, loyalty, and retention with the development cycle principles from information systems.
  • We incor porate several appropriate marketing frameworks into the GIST design.
  • Our focus is on trans forming existing marketing models for the online environment by leveraging the real-time and offline behavior of visitors, identify gaps in their expectations, and redesign the site to minimize or eliminate these gaps.
• In the next section, we overview how the design science framework outlined in Hevner et al. was used to guide the development of our research artifacts that constitute the proposed framework.
  • The following sections introduce the specific constructs and conceptual approaches that led to the GIST methodology's development and the information architecture that is available to imple ment the methodology.
  • We follow with detailed descriptions of each of the stages of GIST: Gather-lnfer-Segment-Track. We conclude with a case study of a Fortune 50 company that provides proof-of-concept for GIST.

A Design Science Approach:

• In their recent article, Hevner et al. (2004) provide a general framework to guide IS researchers and practitioners on how to conduct, evaluate, and present design science research.
  • We believe the work presented in this paper exemplifies the approach and evaluation criteria presented in their work.
  • Below we present an outline of how various components of GIST relate to the guidelines presented in Hevner et al.
• Design as an Artifact. The methodology we develop is itself the artifact that is created to address the important problem of designing and redesigning customer-centric Web sites in the presence of non-transactional visitors. The methodology consists of multiple steps: Gather-lnfer-Segment-Track (GIST), for which specific artifacts were developed. The combined artifact of these steps is a dynamic Web site design for a non-transactional Web site, described in the case study.
• Problem Relevance: As stated in the Introduction, the problem of designing and redesigning customer-centric Web sites is extremely relevant. This is a problem that very significantly highlights the interplay among business strategy, IT strategy, organizational infrastructure, and IS infra structure. The case study we present to evaluate our methodology further illustrates all of these aspects.
• Research Rigor: We combine rigorous elements of multidisciplinary fields in constructing our solution methodology: marketing, data mining, and systems design. We draw from the knowledge base in each of these areas and extend the ideas in the area of consumer segmentation, its identification and validation, and the use of consumer segmentation in business strategy.
• Design as a Search Process: Our meth odology is iterative by nature. At each step we search for gaps between design objec tives for the Web site and observed metrics. GIST internalizes the feedback loop in the continuous evaluation process and redesign.
• Design Evaluation: As suggested in Hevner et al., we use an observational evaluation method to evaluate the design. Specifically, we observe whether the redesign of the Web site in the business organization resulted in identification of business leads and in identification and mapping of new customer segments.
• Research Contributions: Our main con tribution is the GIST methodology, which is unique in a number of ways, combines the marketing and IT disciplines to continuously address Web site design requirements for both transactional and non-transactional visitors. We create and describe the design methodology - GIST, design artifact?the Web site and its information gathering components, and describe the implementation strategies for a marketing research founda tion to enable GIST.
• Research Communication: GIST is based on the underlying disciplines and follows a rigorous step-by-step approach. At the same time, the resulting artifact is of extremely high relevance to practitioners, as demonstrated through the successful implementation in a Fortune 50 company. As suggested by Hevner et al., we present the technical foundation for successful implementation for design science researchers and the case study to communicate the relevance and importance of this research for a management-oriented audience.

Implications and Conclusion:

• The electronic storefronts of the Web environment present information systems and marketing challenges which, if overcome, can create compe titive advantages.
  • One of the primary advantages is the ability to have a customer-centric site - one that understands and adapts to customers' objectives for the visit.
  • Within the traditional lifecycle approach as well as the existing marketing literature, an existing con ceptual model for designing these customer focused sites did not exist.
  • A review of applicable marketing frameworks on building and solidifying the online customer relationship through customerization (Wind and Rangaswamy 2001), loyalty, segmentation,and retention provide a body of knowledge that has not effectively been translated intoW eb design or ongoing site maintenance that is focused on the customers' needs.
  • It is through these models, applicable to the e-world, that we developed a methodology for the degn and maintenance of customer-centric sites.
• We present a methodology following the design science principles of Hevner et al. (2004) to provide an operational approach for non-transactional Web sites such that they can play a critical role in driving business objectives.
  • The key to our approach is the identification of nanosegments and nanoflows, the specific gap analyses for these groups of visitors, their pre ferred paths, and the consequent actionable findings that are derived from the process.
  • The nano-segments are generated by extensive use of statistical analyses and data mining tools on the combination of online and offline data.
  • The concept of nano-segments vis-a-vis personali zation and microsegments was presented in Figure 1.
  • As demonstrated in Figure 1, the nano-segment approach understands the why (onlinebehavior through rigorous analysis) and the who (understanding from similar behavioral patterns) while minimizing the privacy concerns that are voiced in a pure one-to-one dialogue.
• We also present a business-to-business application, in a large company, as an illustration of how GIST translates the academic research into the practice of developing and maintaining a customer-centric Web site. The proof of concept also serves as a model validation.
  • One of the crucial outcomes from this model validation was if the nanosegment, so defined, is the right unit of analysis forthe design and maintenance activities, given the availability of information technologies that are adequate for analyzing the vast amounts of data.
  • Consequently, important design and con tent factors were identified in the first application of GIST such as:
    • incomplete preliminary targeting
    • inadequate descriptions of the type of loans sought by the Internet customers
    • the discovery of a new segment, intermediaries, which redefined the role of the Web site as well as pointed toward the re quirement of infrastructure enhancements
  • As a result of applying the GIST model, the company experienced a significant increase in their "hit rate," justifying the expense in new Web site design.
• We also briefly relate our approach to the popular usability studies of Web sites (see, for example, Nielsen [2000] and the Web site www.useit.com).
  • Usability studies are an essential part of GIST'S track phase when they are used in conjunction with assessing e-services metrics.
  • After we identify and uncover the relevant nano-segments, usability studies provide additional insights in the redesign and repositioning of the nanoflows that are necessary to align the goals and objectives of each nanosegment with the site.
  • The GIST methodology is much broader in the sense that it discovers the nanosegments and continuously evaluates the designed nanoflows.
  • Usability studies are part of one of the four stages of GIST.
• A natural research follow-up of this work is to evaluate the applicability of the several available statistical and data mining techniques that can be used for the nanosegmentation step of GIST.
  • Because the available data is collected in various ways by different systems, the information architecture necessary for GIST may be at different levels of implementation.
  • For example, in business-to-consumer environments, due to the much richer set of observations about customer profiles and their behavior, different classification techniques might be more appropriate than k-means cluster analysis.
• Another interesting research question is related to the optimal number of nanosegments to satisfy.
  • Depending on the objective function of the company, which may include costs to generate dynamic Web pages, emphasis may be given to better meet the requirements of some and not all identified nanosegments.
  • Also, some nano-segments may not be clearly identified in every cycle of GIST.
  • Knowledge of the underlying business will help better identification of these in the next cycle of the continuous methodology.
• Finally, we emphasize the importance of establishing precise metrics for the continuous evaluation that takes place within the individual nano-segment gap analyses, as well as the overall aggregate performance metrics.
  • This will be pursued through future testing and applications validating the GIST model.
  • These tests are designed to yield a taxonomy or classification structure based on a firm's online information architecture and their e-business marketing strategies.
  • Firms may employ this taxonomy to design their content and ongoing maintenance schedules maximizing their customer-centric sites as an integral component of their competitive arsenal.