Application of knowledge management technology in customer relationship management

& Research Article

Application of Knowledge ManagementTechnology in Customer RelationshipManagement

Ranjit Bose1* and Vijayan Sugumaran2

1Anderson School of Management, University of New Mexico, USA2School of Business Administration, Oakland University, USA

Given the important role being played by knowledge management (KM) systems in the currentcustomer-centric business environment, there is a lack of a simple and overall framework tointegrate the traditional customer relationship management (CRM) functionalities with themanagement and application of the customer-related knowledge, particularly in the contextof marketing decisions. While KM systems manage an organization’s knowledge throughthe process of creating, structuring, disseminating and applying knowledge to enhance orga-nizational performance and create value, traditional CRM have focused on the transactionalexchanges to manage customer interactions. True CRM is possible only by integrating themwith KM systems to create knowledge-enabled CRM processes that allow companies to eval-uate key business measures such as customer satisfaction, customer profitability, or customerloyalty to support their business decisions. Such systems will help marketers address customerneeds based on what the marketers know about their customers, rather than on a mass general-ization of the characteristics of customers. We address this issue in this paper by proposing anintegrated framework for CRM through the application of knowledge management technology.The framework can be the basis for enhancing CRM development. Copyright # 2003 JohnWiley & Sons, Ltd.

INTRODUCTION

CRM is one of the hottest tools in businesstoday. But like total quality management and re-engineering before it, CRM has not always livedup to its hype (Brown, 2000; Swift, 2001). Still, com-panies ignore it at the risk of being left behind.Simply, CRM is a high-tech way of gathering mou-ntains of information about customers, then usingit to make customers happy—or at least a sourceof more business. It is therefore, concerned withunderstanding and influencing customer behavior(Kotler, 2000).

One CRM trailblazer was the gaming companyHarrah’s Entertainment, which has successfullycombined software and human marketing exper-tise to get gamblers into its 25 casinos. Harrah’sdo a thorough, sophisticated analysis of 24 millioncustomers in their database. Harrah’s know—howfrequently customers come, what they play, andthey then provide follow-up with continuous com-munication over the phone, direct mail and e-mailand on their Web site. It allows Harrah’s to be par-ticipatory rather than being simply reactive. Theirtechnologists refer to it as CRM but their managersrefer it as their loyalty program.

Although CRM is the fastest-growing businesstool satisfaction with its use currently ranks quitelow (Winer, 2001). Many companies have startedto realize that they need both the mountains of

Knowledge and Process Management Volume 10 Number 1 pp 3–17 (2003)Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/kpm.163

Copyright # 2003 John Wiley & Sons, Ltd.

*Correspondence to: Dr Ranjit Bose, Anderson School of Man-agement, University of New Mexico, Albuquerque, NM 87131,USA. Email: [email protected]

information on millions of customers as well as anappropriate technical infrastructure coupled withmarketing expertise to use CRM satisfactorily(Zeithaml, 2001). CRM is not necessarily aboutautomating or speeding up existing operationalprocesses; rather, it is about developing and opti-mizing methodologies to intelligently manage cus-tomer relationships. Thus, it is about effectivelymanaging and leveraging customer related infor-mation or knowledge, to better understand andserve customers.

A true CRM solution design requires a complexcombination of many best-of-breed components,including analytical tools, campaign management,and event triggers, combined with the many newcomponents such as collateral management, rule-based workflow management, and integrated chan-nel management needed to achieve a one-to-onemarketing capability. This capability dictates theneed for a single, unified, and comprehensiveview of customers’ needs and preferences acrossall business functions, points of interactions, andaudiences (Shoemaker, 2001; Tiwana, 2001). Addi-tionally, it requires the existence of interfacesbetween non-customer contact systems, such asenterprise resource planning systems (ERP), andoperational and customer contact systems.

As organizations move towards a comprehensivee-business environment, the business processes sup-porting the environment become increasingly,highly knowledge-intensive and therefore, an orga-nization’s long-term success and growth becomedependent on the successful expansion, use, andmanagement of its corporate knowledge across itsbusiness processes (Davenport and Grover, 2001;Liebowitz, 2000). CRM is no exception to this trend,it is moving away from being a transaction-oriented,operational system of the past to a more knowledge-oriented, analytical system of the future that pro-vides the means by which a company can maintaina progressive relationship with a customer acrossthat customer’s lifetime relationship with the com-pany (Gordon, 1998; Kalakota and Robinson,2001). This means having the ability to track andanalyze a range of customer actions and eventsover time, using the information and knowledgefrom operational CRM systems as well as from otherenterprise systems such as KM systems (Wiig, 1999).

Given the important role being played by KMsystems in the current customer-centric environ-ment, there is a need for a simple and integratedframework for the management of customer know-ledge (Winer, 2001). Surprisingly, there is a lack of asimple and comprehensive framework to integratethe traditional CRM functionalities with the man-agement and application of the knowledge, particu-

larly in the context of marketing decisions (Helmkeet al., 2001; Massey et al., 2001; Parasuram and Gre-wal, 2000). While KM systems manage an organiza-tion’s knowledge through the process of creating,structuring, disseminating and applying knowledgeto enhance organizational performance and createvalue (Alavi and Leidner, 2001; Davenport andPrusak, 1998; Liebowitz, 1999; Offsey, 1997), tradi-tional CRM have focused on the transactionalexchanges to manage customer interactions. TrueCRM is possible only by integrating them withKM systems to create knowledge-enabled CRM pro-cesses that allow companies to evaluate key busi-ness measures such as customer satisfaction,customer profitability, or customer loyalty to sup-port their business decisions (Fahey, 2001; Reich-held and Schefter, 2000; Winer, 2001). Suchsystems will help marketers address customerneeds based on what the marketers know abouttheir customers, rather than on a mass generaliza-tion of the characteristics of customers.

We address this issue in this paper by presentingan integrated framework for CRM through theapplication of knowledge management technology.The framework is designed to deliver consistent ser-vice across all touch points and channels by provid-ing: (a) a single view of each customer across theentire enterprise and throughout the customer’s life-cycle; and (b) an architecture that supports and pro-motes knowledge-based, analysis-driven interactionwith each customer. To test the operational feasibil-ity of this framework, a proof-of-concept prototypehas been developed and tested that uses currenttechnologies such as extensible markup language(XML) and intelligent software agents for perform-ing the proposed KM and CRM activities.

Our paper is further organized as follows. First,we present a background literature review onCRM, KM and discuss the uniqueness of ourwork. We then provide the KM capabilities neededfor CRM and the architecture for KM-based CRM.The proof of concept prototype implementationand a demonstration session is then presented. Dis-cussion on the implications as well as limitations ofour research and the future research needs are fol-lowed by the concluding remarks.

BACKGROUND

Customer Relationship Management

CRM is about managing customer knowledge tobetter understand and serve them. It is an umbrellaconcept that places the customer at the center of anorganization. Customer service is an importantcomponent of CRM, however CRM is also

RESEARCH ARTICLE Knowledge and Process Management

4 R. Bose and V. Sugumaran

concerned with coordinating customer relationsacross all business functions, points of interaction,and audiences (Brown, 2000; Day, 2000).

Delivering consistent service across all touchpoints gives companies a strong market advantage.When information or knowledge is fragmentedwithin a company, customer feedback is hard toobtain. As a result, customer service suffers andorganizations fall back on the mass marketing prin-ciple that ‘one-size-fits-all’. One-to-one marketingrequires a comprehensive view of customers’ needsand preferences (Kotler, 2000).

Information technology-driven relationshipmanagement by a firm focuses on obtainingdetailed knowledge about a customer’s behavior,preferences, needs, and buying patterns and onusing that knowledge to set prices, negotiate terms,tailor promotions, add product features, and other-wise customize its entire relationship with eachcustomer (Kohli, 2001; Shoemaker, 2001). Offeringcustomers convenience, personalization and excel-lent service plays a key role in the success and dif-ferentiation of many online businesses (Kalakotaand Robinson, 2001). CRM focuses on providingand maintaining quality service for customers byeffectively communicating and delivering pro-ducts, services, information and solutions toaddress customer problems, wants and needs.

Knowledge management

KM is management of a company’s corporateknowledge and information assets to provide thisknowledge to as many company staff members aspossible as well as its business processes to encou-rage better and more consistent decision-making(Probst et al., 2000). By integrating operationalCRM data with knowledge from around the enter-

prise, companies can make use of the abilities ofanalytical CRM systems, and with them, maketruly customer-centric business decisions. Forexample, companies can proactively offer productsand services that fit a given customer’s needs basedon what the customer has already purchased, orincrease purchase rates by dynamically personaliz-ing content based on Web visitor’s profile, or pro-vide customers in the highest value tier withpersonal representatives who understand their his-tory or preferences.

There is an increased sense of urgency in theinstitutionalization of comprehensive knowledgemanagement programs due to the fact that the Inter-net and the World Wide Web are revolutionizingthe way enterprises do business (Alavi and Leidner,1999; Leebaert, 1998; Liebowitz, 2000; O’Leary,1998). A well-designed KM infrastructure makes iteasier for people to share knowledge during pro-blem solving resulting in reduced operating cost,improved staff productivity, cost avoidance, andsoft benefits such as increasing the knowledgebase, and sharing expertise (Applehans et al., 1999).

The KM framework we present (shown inFigure 1) consists of the following four major pro-cesses: (a) knowledge identification & generation,(b) knowledge codification & storage, (c) knowledgedistribution, and d) knowledge utilization & feed-back. The knowledge identification & generation processincludes recognition and creation of new knowledge.It focuses on determining the relevant customer, pro-cess and domain knowledge needed to successfullycarry out CRM activities and acquiring or generatingthis knowledge by monitoring the activities of custo-mers and other players in the industry.

The knowledge codification & storage process invol-ves converting knowledge into machine-readableform and storing it for future use. In particular, it

Figure 1. Knowledge management framework

Knowledge and Process Management RESEARCH ARTICLE

KM Technology in Customer Relationship Management 5

deals with archiving the new knowledge by addingit to a persistent knowledge repository that can beused by all the stakeholders. This process consistsof mapping the knowledge to appropriate formal-isms, converting it to the internal representationand storing it in the knowledge repository. Currenttechnologies such as XML and the UniversalDescription, Discovery and Integration (UDDI)formalism can be used for internal representationand storage. These approaches facilitate easy searchand retrieval of relevant knowledge from the repo-sitories, and enables the stakeholders to apply thisknowledge in decision-making (David, 1999).

The knowledge distribution process relates to dis-seminating knowledge throughout the organiza-tion and handling requests for specific knowledgeelements that would be useful in working througha specific problem scenario. Knowledge dissemina-tion can employ either ‘push’ or ‘pull’ technologiesdepending upon the organization’s culture andinfrastructure.

The knowledge utilization & feedback process com-prises knowledge deployment and providing feed-back. This process enables the stakeholders toidentify and retrieve relevant knowledge neededfor solving a particular problem. Utilization of thisknowledge in the context of a specific problemmay result in additional knowledge, which can beabstracted out and stored in the knowledge reposi-tory for future use. Stakeholders can provide feed-back regarding the quality of knowledge stored inthe repository as well as how easy or difficult it isto search for relevant knowledge. They can also iden-tify new types of knowledge that need to be gatheredbased on strategic objectives and the changes that aretaking place within the environment.

This research attempts to integrate relevantenabling technologies (Devedic, 1999; Fowler, 2000;Sycara et al., 1996; Wu, 2001) into an environmentthat would support organizational knowledge crea-tion, use, and management. Two such enabling tech-nologies that we discuss are intelligent agents andXML, which are briefly discussed below.

Intelligent agents and KMIntelligent agents are useful in automating repeti-tive tasks, finding and filtering information, andintelligently summarizing complex data (Murchand Johnson, 1999). Just like their human counter-parts, intelligent agents can have the capability tolearn and even make recommendations regardinga particular course of action (Hess et al., 2000;Maes et al., 1999). Intelligent agents can act onbehalf of human users to perform laborious androutine tasks such as locating and accessing neces-sary information, resolving inconsistencies in the

retrieved information, filtering away irrelevantand unwanted information, and integrating infor-mation from heterogeneous information sources.In order to execute tasks on behalf of a businessprocess, computer application, or an individual,agents are designed to be goal driven, i.e. theyare capable of creating an agenda of goals to besatisfied. Agents can be thought of as intelligentcomputerized assistants.

XML and KMExtensible Markup Language or XML is emergingas a fundamental enabling technology for contentmanagement and application integration (Balasu-bramanian and Bashian, 1998; Goldfarb andPrescod, 1998). XML is a set of rules for definingdata structures and thus making it possible for keyelements in a document to be characterized accord-ing to meaning. XML has several valuable character-istics. First, it is a descriptive markup languagerather than a procedural markup language. Hence,it is possible to represent the semantics of an XMLdocument in a straightforward way. Second, it isvendor independent and therefore highly transpor-table between different platforms and systems whilemaintaining data integrity. Third, it is human legi-ble. It is also worth noting that XML has its rootsin SGML (Standard Generalized Markup Language)and adheres to many of its principles.

XML enables us to build a structure around thedocument’s attributes, and RDF (Resource Descrip-tion Framework) allows us to improve searchmechanisms using the semantics of annotations(Decker et al., 2000; Rabarijaona et al., 2000). XMLmakes it possible to deliver information to agentsin a form that allows for automatic processing afterreceipt and therefore distribute the processing loadover a federation of agents that work cooperativelyin problem solving. The set of elements, attributes,and entities that are defined within an XML docu-ment can be formally defined in a document typedefinition (DTD).

We contend that by combining intelligent agentand XML technologies, one could envision aknowledge management environment that sup-ports all phases of the knowledge life cycle,namely, creation, organization, formalization, dis-tribution, application, and evolution.

Our contribution

We present an integrated framework, that aims forknowledge-enabled CRM processes, and which sup-ports and promotes consistent, knowledge-based,analysis-driven interaction with each customer. Maj-ority of today’s CRM systems are focused primarily



on call centers’ operations (Brown, 2000; Masseyet al., 2001; Orzec, 1998). Several software vendorsare active in this field and are offering initial ver-sions of their products. Examples include Macrome-dia (ARIA and LikeMinds product lines), Vignette,Engage, IBM (i.e. net commerce), Mathlogic, Micro-soft (i.e. Site Server Commerce), NetGenesis, and E.piphany. The analytical CRM system that we pro-pose is just emerging (Swift, 2001). It is designedto provide business intelligence by encompassingknowledge management practices and by lever-aging the knowledge gathered from cross-functionalcustomer touch points such as call center, Webaccess, e-mail, and direct sales.

The ability to leverage the knowledge fromcustomer-facing systems for back-office analysishas recently been proven to be directly propor-tional to a company’s success in enhancing custo-mer loyalty (Reichheld and Schefter, 2000).Without this ability, the environment remains dis-connected, and many important business questionscannot be easily answered. For example, a custo-mer service representative scheduling a follow-upcommunication with a customer may not be ableto discern that customer’s value score to determinethe level of service that should be provided, or anaccount representative may have no idea whether akey business customer has responded to certainkey promotions, or a customer support analystmay try in vain to measure complaint historyagainst sales revenue for a given product.

Analytical CRM systems can incorporate severaldifferent types of analytical tools for support per-sonnel. For example, tools for predictive modeling(e.g. behavior prediction uses historical customerbehaviors to foresee future behaviors, using sophis-ticated modeling and data mining techniques) toprovide lists of customers most likely to respondto a given marketing campaign, or purchase-patternrecognition, or enabling marketing and sales staff tocompare customers with like behaviors so they canposition new products to an optimal audience(Berry and Linhoff, 1997; Bose and Sugumaran,1999; Fraternali, 1999). The keys to different typesof analyses, and especially to the actions that result,are (a) knowing a firm’s best customers and itsunprofitable customers, so it can lure the rightones back, and (b) understanding that CRM has towork for customers, not just the company.

KNOWLEDGE MANAGEMENTCAPABILITIES NEEDED FOR CRM

In order to implement knowledge-enabled CRMprocesses, companies need to provide and support

several categories of knowledge management cap-abilities through the deployment and integration ofcurrently available technologies (Gold et al., 2001).The capabilities prescribed in this research are pri-marily intranet and extranet based.

The capabilities framework, presented inFigure 2, is designed around enterprise knowledgeportals. Using a portal architecture allows for acommon interface to knowledge from differentknowledge sources such as documents, applica-tions, and data warehouses (Applehans et al.,1999; Caldwell et al., 2000). The capabilities frame-work is designed to accelerate the penetration ofknowledge management within organizationsbecause the users, who most likely are familiarwith the portal concept through the use of Internetportals such as Yahoo, will expect that the interfacecomponent of the architecture to offer similar cap-abilities for knowledge management, such assearch engines and automatic document summari-zation, across an enterprise-wide collection ofdocuments.

At a high level the framework can be explainedas comprised of two parts. First, it is designed toleverage existing knowledge and to enable creationof new knowledge through a continuous learningprocess denoted by the knowledge learning loops.And second, the rectangular labeled boxes denotethe KM capabilities and a few currently availabletechniques or technologies that can provide them.A brief description of each of the capabilities is pro-vided below.

Presentation involves personalizing both theaccess to and displaying of the results of user inter-actions with the system. It is designed to let everyorganizational user know where to go to find theorganization’s knowledge through a singlebrowser-based point of entry to all informationthat the user may need. Personalization providesthe ability to customize what types of informationare relevant to a user and how that information ispresented.

The personalization function helps personalizecontent and services to deliver tailored content orinformation to users based on several user criteriaor preferences. The primary capabilities of thisfunction include the creation of personalizationprofiles of individual users or groups or depart-ments or divisions, providing personalized naviga-tion, providing personalized notification, and theability to personalize the content categorization.Personalization is often accomplished by usingsoftware agents, commonly called spiders, to getthe information and handle user profiling.

The collaboration function is designed to connectpeople with people through communities of



practices; to preserve discussions; and to stimulatecollaboration by integrating the knowledge reposi-tories and collaboration applications such as work-flow.

The process function allows users to participatein relevant business processes in the context oftheir own roles. Through this function, users haveaccess to knowledge management applicationssuch as knowledge or evidence based decision sup-port system applications that enable increasedresponsiveness to customers and partners.

The publishing and distribution function providesthe means and a platform for users to easily cap-ture and distribute the particular kinds of knowl-edge assets they need to monitor withoutrequiring them to learn complex programmingsyntax. Software agents are used extensively forthis function (Aguirre et al., 2001). These agents

are designed in such a way that users can set upand control them. The users can specify in themthe type of knowledge he or she wants to publish,distribute, and receive. The frequency (by timeand/or quantity) and method (by e-mail or Webpage) are important parameters that should be setup by the users.

The integrated search function is designed toreduce the information overload and usefulnessof search results to the users. Integrated searchesacross all repositories are performed by defaultbut users can also identify the repositories theywant to search such as Web pages, e-mails, and dis-cussions. This function should also provide theability to automate indexing and to crawl fre-quently to keep the index current.

The categorization function allows users tobrowse, create, and manage knowledge categories.

Figure 2. Knowledge management capabilities for CRM



It establishes a process and guidelines for author-ing and publishing knowledge categories by theusers. Business groups or departments or divisionsare made responsible for creating and managingtheir own subject area taxonomies.

The integration function ensures seamless andconsistent navigation among and between theabove functions and knowledge sources such thatall individuals can use the organization’s combinedknowledge and experience in the context of theirown roles.

ARCHITECTURE FOR KM-BASEDCRM SYSTEMS

CRM projects usually fail because they force a lot ofchanges quickly on business units and the resultingapplications often don’t serve customers any better.They also fail to integrate the disparate datasources or provide the right kind of informationto the right people at the right time (Parasuramand Grewal, 2000). Hence, CRM applicationsshould have the capability to not only gather andmake available relevant information in a timelyfashion, but also provide tools for analyzing andsharing the information in a meaningful way andallow managers to act quickly. Knowledge man-agement systems deal with these kinds of issues,particularly, identifying and creating knowledge

elements from various sources, codifying, storingand disseminating knowledge, and utilizing thisknowledge in problem solving (Nissen et al.,2000). Hence, we contend that a KM-based CRMsystem would provide precisely the kinds of cap-abilities needed for a CRM system to be effectivein managing lasting partnerships with valuablecustomers. We envision a KM-based CRM systemwith components that facilitate the easy gatheringand assimilation of customer related informationas well as organizational processes and industrypractices. We propose an architecture for a custo-mer centric CRM system, shown in Figure 3, thatcombines the traditional knowledge managementcapabilities as well as the CRM activities neededfor successful CRM initiatives. The proposed archi-tecture consists of four major components: (a) inter-nal and external data sources, (b) knowledgeacquisition, (c) knowledge repositories, and (d)knowledge utilization. These components arebriefly described in the following paragraphs.

(a) Data sources: Effective customer relationshipmanagement requires different types of infor-mation from a variety of sources. For example,transaction information may be contained inoperational databases, whereas standard oper-ating procedures may be stored in official docu-ments. Data sources may be both internal andexternal to the organization and the CRM

Figure 3. KM-based CRM analytics system architecture



system should have mechanisms to access andretrieve relevant data. For example, the CRMsystem should be capable of gaining access tonot only transaction and customer related infor-mation, but also organizational processes andindustrywide domain information that wouldbe useful in problem solving and strategic deci-sion making activities. Thus, the CRM systemshould have an open architecture that is cap-able of interacting with a wide variety of dataand knowledge sources.

Data needed for CRM analytics is verydiverse and may be unstructured and difficultto manage (e.g. emails, call reports on PDAsetc.). Emerging information technologies canbridge the gap by: (a) defining standard dataformats, such as XML for data presentation orOpen Database Connectivity for database-to-database exchanges, (b) ensuring data integritythrough proven and published processes, (c)establishing data migration processes, such asstoring procedures for graphical data, and (d)choosing CRM analytics tools that supportWeb browser access.

(b) Knowledge acquisition component: This compo-nent is responsible for the early phases ofknowledge management life cycle, whichinvolves identifying, acquiring and storingrelevant knowledge that would be useful inmanaging customers and products and makingmeaningful decisions regarding customer ser-vice and product service offerings. For exam-ple, keeping track of customer histories andcharacteristics would be essential in determin-ing who, and how best to serve the clientalgiven various options. The knowledge acquisi-tion component consists of different agents thatare geared towards acquiring and synthesizinginformation related to various aspects of custo-mer relationship management. These agentsare: (1) Transaction Info Agent, (2) CustomerInfo Agent, (3) Process Info Agent, and (4)Industry Info Agent. The Transaction InfoAgent is responsible for gathering and assimi-lating information regarding what products aparticular customer has bought over a periodof time. This information is obtained by inter-acting with the transaction databases that existwithin the organization. The Customer InfoAgent gathers information related to customerpreferences and characteristics and keeps trackof customer profiles. It is primarily responsiblefor generating a comprehensive picture ofevery customer and determining the value ofeach customer. The Process Info Agent dealswith collecting information related to various

organizational processes, policies and proce-dures that have been established and theirapplicability to different situations. Mostly,standard operating procedures are describedin documents, which are not readily accessibleto users. This agent creates a repository of theseprocesses and policies for everyone to access.The Industry Info Agent is structured to accessdata sources outside the organization to gain anunderstanding of the latest developments thatare taking place in the industry and makingthis knowledge available to decision makers.

(c) Knowledge repositories: This component con-sists of repositories that contain knowledge ele-ments generated by humans as well as theagents that are part of the knowledge acquisi-tion component. These repositories are continu-ally updated as new information becomesavailable. There are four major repositoriesthat are maintained, namely, (a) CustomerTransactions, (b) Customer Profiles, (c) Policiesand Procedures, and (d) Domain Knowledge.The Customer Transaction repository containsparticulars about all the transactions related tocustomers. For each purchasing transaction,information about the products and servicesthat the customer bought, discounts that wereprovided, date of purchase, etc. are maintainedso that the customer representative can searchand retrieve one or more transaction recordsfor a particular customer. The Customer Pro-files repository contains the complete back-ground of each customer including customerhistory and preferences. It also contains custo-mer ratings and as a result a service representa-tive can quickly assess the value of a particularcustomer while interacting with that customer,and make appropriate decisions based on theimportance of the customer. The Policies andProcedures repository contains informationregarding standard procedures and policiesthat have to be followed in handling a particu-lar situation. It also contains taxonomies of pro-duct codes and associated services. The DomainKnowledge repository contains informationabout the industry in general, and the latestdevelopments and trends within that industrythat decision makers have to be aware of,such as changes in governmental regulations,new standards and benchmarks, etc.

(d) Knowledge utilization component: The knowl-edge utilization component is responsible forsupporting the later phases of the KM life cycle,in particular, activities related to searching andretrieving relevant knowledge, as well as shar-ing this knowledge with other stakeholders to



be utilized in different scenarios. It acts as theinterface to knowledge repositories. It enablesstakeholders to search the knowledge reposi-tories for specific information related to theproblem they are solving. This component isalso responsible for content delivery (knowl-edge that may be of interest to certain groups)on a periodic basis. The knowledge utilizationcomponent consists of the following agents: (i)Repository Management Agent, (ii) SituationAnalysis Agent, (iii) Predictive Modeling Agent,and (iv) Marketing Automation Agent.

(i) Repository Management Agent: This agent pro-vides a number of functions for repositorymanagement such as organizing, maintainingand evolving the knowledge repositories. Italso provides mechanisms for browsing theserepositories as well as searching for specificknowledge elements relevant to a particularproblem at hand. This agent is also responsiblefor knowledge dissemination, which includesvarious aspects such as presentation, persona-lization, collaboration, and publishing. Thisagent provides easy access to important andrelevant data, in particular, makes more custo-mer data available to call center operators sothey can solve customer problems on the firstcall. This agent disseminates the informationmined by analysts to the marketing, sales,and front-line customer service people whocould actually use it. It also permits calleridentification linked with customer historiesand characteristics in order to identify mostvaluable customers and provide appropriateservices.

(ii) Situation Analysis Agent: This agent providesmechanisms for the user to undertake problemsolving and decision-making activities. Forexample, a customer service representativemay be faced with an angry customer with acomplaint. The representative can analyze thesituation and reach a resolution quickly basedon the customer profile and transaction his-tory. Similarly, a manager has the ability tosee which specific products in the store areselling well, badly or according to expectedtrends, and to take appropriate actions. Themanager would have the capability to ask sev-eral key questions such as: is the product per-forming badly because of poor displaystandards, poor stock availability or incorrectlocation? Is the product right for the store,does it provide enough profit for the spaceallocated, could another product’s space beenlarged or a new product brought in to pro-

vide better profit for the space? Without thiscapability, store managers may have no wayof identifying the most profitable productsand allocating more time to these profitablelines.

(iii) Predictive Modeling Agent: On the CRM analy-tics side, the biggest disappointment has beenthe failure to integrate business logic into thetools. The Predictive Modeling Agent enablemanagers to conduct meta-analysis and identi-fy areas of strengths and weaknesses. Forexample, they can watch transactions in realtime to spot patterns, such as decreasing trans-action rates or balances for a high value custo-mer that indicate that a customer might soonleave. It enables managers to get a grasp oncustomers’ buying patterns, anticipate trendsand more carefully align inventory to maxi-mize profits in a chain of stores. Most timelyinformation is of little use unless the corporatestrategy aligns with what the customer data isrevealing.

(iv) Marketing Automation Agent: One of the big-gest pitfalls of customer databases is that thebest customers are bothered endlessly—sur-veys, new offers, cross-selling etc. Lack of anintegrated CRM system results in alienatingcustomers by making inappropriate pitchesand ignoring customers with low currentreturns but high potential. Another mistakethat is often made is segmenting customerson the basis of demographics such as age,income, sex or education because this informa-tion is relatively easy to get. But the best CRMsystems will segment customers based on fun-damental values. The proposed KM-basedCRM system will be able to match actual buy-ing information to customer profiles and pre-ferences, which can permit the marketingpeople to really see trends from individualcustomers and develop better marketing cam-paigns.

PROTOTYPE IMPLEMENTATION

A proof-of-concept prototype is currently underdevelopment. This prototype uses the traditionalclient-server architecture, where the client is a sim-ple web browser, using which the user can interactwith the knowledge repositories. The user can alsoperform one or more CRM activities supported bythe Knowledge Utilization component. The agentsthat are part of the Knowledge Acquisition



Component as well as the Knowledge UtilizationComponent have been implemented using JADE(Java Agent DEvelopment Framework) fromCSELT, Turin, Italy (Bellifemine et al., 1999).JADE is a middle-ware product that is used todevelop agent-based applications, which are incompliance with the FIPA specifications for intero-perable intelligent multi-agent systems. JADE isjava-based and provides the infrastructure foragent communication in distributed environments,based on FIPA standards. The reasoning capabilityof the agents has been implemented through JESS,which is an expert system shell written in Java(Friedman-Hill, 2002). The transaction information,customer profiles and preferences, organizationalprocesses and procedure information, as well asthe application domain knowledge are capturedand represented in XML documents with appropri-ate DTDs. These XML documents are stored in thecorresponding knowledge repositories, which havebeen implemented as XML databases using theTamino software (from Software AG—www.softwareag.com). Among other things, Tami-no provides X-Studio, which is a complete suite ofapplication development tools for creating XML-based applications. Tamino XML databases store

data directly in native XML format and providefacilities for fast storage, exchange and retrieval ofXML documents.

Sample session with prototype: The followingparagraphs describe a brief sample session thatprovides a glimpse of some of the functionalitiesof the KM-based CRM System prototype. Whenthe user accesses the CRM system, a login screen,shown in Figure 4, is presented where the usercan type in the userid, password and the usertype. Users are provided different levels of accessto control the evolution of the knowledge reposi-tories. For example, not all users can create newknowledge elements and store them in the reposi-tory or have access to sensitive information. Someof the typical users of the system are customer ser-vice representatives, department heads, divisionmanagers and senior executives. Once the user isauthenticated, depending upon the type of theuser, appropriate menus are presented. Users canalso customize the interface to suite their tastesand preferences.

When the user logs into the system, he or she canperform various knowledge management andCRM activities. For example, if the user type is‘customer service representative,’ he/she can,

Figure 4. Initial screens from the KM-based CRM system



among other things, view customer profiles andhistories as well as perform situation analysis. Fig-ure 4 shows the initial menu that lists the optionsfor carrying out various functions. The user canselect any of the options and click on the Submitbutton to perform that particular operation. The‘Knowledge Acquisition and Repository Manage-ment’ option enables the user to invoke the knowl-edge elicitation process from various sources orperform maintenance operation on one or more ofthe knowledge repositories. The user can explicitlyspecify tasks for the agents that are part of theknowledge acquisition component, or ask them togather information on a continual basis. Theseagents can create new knowledge elements andadd them to the appropriate knowledge repositoryusing predefined ‘repository management’ proce-dures. The ‘Customer History and Profile’ optionfacilitates the user to probe available customerinformation and generate an up-to-date picture ofa particular customer and determine the value ofthat customer. For example, a customer servicerepresentative can pull up the transaction history

of a particular customer and get a sense of thevalue and loyalty of that customer. The customertransaction knowledge repository has been imple-mented as an XML database and can be searchedusing customer id or customer name. Figure 5shows the interface for searching and viewingtransaction histories. By default, the system dis-plays the transaction history of a customer as anXML document (bottom portion of Figure 5), andby clicking on the Display button, the user cansee the HTML rendering of the document usingcascading style sheets.

The user can also browse the organizational pro-cesses and procedures knowledge repository orsearch for specific policies related to a particularsituation by selecting the ‘Policies and Procedures’option shown in Figure 4. The system providesanother panel where the user can specify a few key-words, using which the repository is searched andmatching policies and procedures are displayed tothe user. The ‘Situation Analysis’ option enablesthe user to analyze a particular event or circum-stance based on relevant information, and perform

Figure 5. Viewing transaction information for a specific customer



what-if analysis before reaching a meaningful con-clusion. For example, a customer service represen-tative may be faced with a situation where acustomer is not satisfied with a product or serviceand is calling up and demanding recourse. In thissituation, having quick access to that customer‘shistory and rating, as well as the policies and pro-cedures that dictate how such a case should behandled, can help that representative quicklyresolve this situation to the satisfaction of the cus-tomer and still stay within the parameters that therepresentative has to operate under. A simplesituation analysis interface is shown in Figure 6.The representative can enter customer informationin the ‘Customer Info’ box and some keywordsdescribing the situation in the ‘Situation Info’ boxand get relevant customer information as well asapplicable policies and procedures displayed byclicking on the appropriate buttons. When theuser clicks on the ‘Recommendation’ button, thesystem provides some recommendations based onpre-established rules. When the user clicks on thecustomer profile button after entering the customeridentifier, an appropriate query is generated to

search the transaction and customer profile reposi-tories. The retrieved information is displayed in thewindow shown in the lower part of Figure 6. Simi-larly, when the user enters situation descriptors,those keywords are used in searching the policiesand procedures repository and relevant policyand procedure information is displayed in the low-er window.

The ‘Predictive Modeling’ and ‘Marketing Auto-mation’ options (shown in Figure 4) are utilized bymanagers interested in analyzing the performanceof specific products or services and also under-standing the customer base for developing specificmarketing campaigns and promotions. The ‘SearchKnowledge Repositories’ option (shown inFigure 4) provides ad hoc querying capabilitiesusing which the user can search one or moreknowledge repositories for related information.

DISCUSSION

This section highlights some of the major implica-tions of this research. Our approach to integrating

Figure 6. Situation analysis interface



knowledge management techniques into customerrelationship management activities providesseveral advantages. Individuals, various businessunits, and the organization as a whole can all ben-efit from the proposed integrated KM-based CRMenvironment. At the individual level, customer ser-vice representatives can browse the knowledgerepositories, perform plain-text searches for speci-fic customer information, customer profile and his-tory, and rating. This real time access to relevantinformation enables the representatives to betterserve customers. Different business units can bene-fit from such a system by being able to gain accessto customer and sales information that are gatheredthrough various touch points, as well as the stan-dard policies and procedures that are otherwisenot easily accessible. At the organizational level,our system could be utilized in providing a com-mon infrastructure for carrying out customerrelationship management activities and institutio-nalizing a comprehensive set of CRM policies.

The proof-of-concept prototype we implemen-ted, demonstrates the operational feasibility of theproposed KM-CRM integration framework. Cur-rent technologies such as intelligent agents andXML technologies were selected and used forimplementation because (1) to reduce the cognitiveburden on the user in problem solving and decisionmaking activities, and (2) these technologies facili-tate the easy integration of knowledge manage-ment activities and CRM activities. For example,intelligent agents can be tasked to monitor certaintypes of transactions or search and retrieve specificcustomer related information in real time. XMLtechnology permits easy codification and dissemi-nation of knowledge elements to interested partiesthrough push or pull technologies. In addition, itimproves the interoperability of knowledge ele-ments between different applications. Tradit-ionally, KM tools use proprietary knowledgestructures and internal representations that prohi-bit the exchange of knowledge between variousapplications. In contrast, our system uses XMLrepresentation, which alleviates this problem to agreat extent. Storing customer information in anXML database also facilitates various stakeholdersto view information at different levels of aggrega-tion through specific transformations. For example,customer service representatives can query theXML database for individual customer historiesand profiles, whereas, marketing people can viewcustomer information based on certain ‘value pro-positions’.

While the implemented prototype incorporatesthe necessary functionalities and capabilities toadequately prove the operational feasibility of the

proposed integration framework, it is by no meansa full-blown system. Therefore, the current versionof the prototype has the following limitations. First,while we have developed the DTDs for some of thecommon knowledge elements, much work remainsto be done in order to capture all types of knowl-edge that would be useful in carrying outcomprehensive CRM activities. Second, knowledgerepositories as well as operating procedures evolveover time in an organizational setting. Hence,the prototype needs additional capability to ensurethat the knowledge repositories are consistent withbusiness processes on a continual basis. Third, theprototype currently does not provide applicationinterface to several potential third party softwarethat could be easily utilized in predictive modelingand automating many of the marketing relatedactivities.

Further work is required to bring the prototypeto a full-blown system as well as to address some ofthe issues that arise in integrating knowledge man-agement techniques into customer relationshipmanagement. Our future work on the prototypeincludes incorporating additional components forknowledge acquisition and utilization, and provid-ing APIs for various decision analytic tools forfacilitating the creation of an integrated KM–CRMportal with customizable functionalities. Theresulting full-blown system will be able to supportbetter query facilities for searching knowledgerepositories, particularly, natural language basedinterfaces that provide flexible query mechanisms.Subsequently, field testing and empirical validationof the full-blown system is necessary to evaluate itseffectiveness from the perspective of target users.While the present prototype uses current technolo-gies such as intelligent agents and XML, potentialuse of other enabling technologies like Ontologies,UDDI (Universal Description, Discovery, and Inte-gration), and Web Services need to be investigated.Future research should also address additionalissues related to the integration of KM and CRMactivities such as configuring KM activities to alignwith the overall objectives of CRM initiatives, iden-tifying the types of knowledge needed for specificCRM activities, and managing the evolution of aconsistent set of knowledge repositories.

CONCLUSION

Analytical CRM systems achieve a single, unifiedview of the customer and facilitate a seamlessexchange between customers and corporations.However, a single view of customers requirestightly integrated applications both within the



realm of CRM applications and back-end technolo-gies, such as knowledge management. Organiza-tional knowledge creates value in use. A keychallenge in the application of knowledge is trans-ferring it from where it was created or captured towhere it is needed and should be used.

We tried to address the issue in this research bydeveloping a simple and overall framework to inte-grate the traditional CRM functionalities with themanagement and application of knowledge in thecontext of marketing decisions. The operationalfeasibility of the framework was tested through aproof-of-concept prototype, which was built usingintelligent agent and XML technologies. We con-tend that the framework can be the basis for enhan-cing CRM system functionalities and development.

REFERENCES

Aguirre JL, Brena R, Cantu FJ. 2001. Multiagent-basedknowledge networks. Expert systems With Applications20(1): 65–75.

Alavi M, Leidner DE. 1999. Knowledge management sys-tems: issues, challenges, and benefits. Communicationsof the AIS 1(7).

Alavi M, Leidner DE. 2001. Review: knowledge manage-ment and knowledge management systems: concep-tual foundations and research issues. MIS Quarterly25(1): 107–136.

Applehans W, Globe A, Laugero G. 1999. ManagingKnowledge: A Practical Web-Based Approach. Addison-Wesley: New York.

Balasubramanian V, Bashian A. 1998. Document manage-ment and web technologies: Alice marries the MadHatter. Communications of the ACM 41(7): 107–114.

Bellifemine F, Poggi A, Rimassa G. 1999. JADE—A FIPA-compliant agent framework. Proceedings of PAAM’99,London, 97–108.

Berry MJA, Linhoff G. 1997. Data Mining Techniques: ForMarketing, Sales, and Customer Support. John Wiley:New York.

Bose R, Sugumaran V. 1999. Application of intelligentagent technology for managerial data analysis andmining. Database for Advances in Information Systems30(1): 77–94.

Brown SA. 2000. Customer Relationship Management: AStrategic Imperative in the World of E-Business. JohnWiley: New York.

Caldwell N, Clarkson PJ, Rodgers P, Huxor A. 2000.Web-based knowledge management for distributeddesign. IEEE Intelligent Systems 15(3): 40–47.

Davenport TH, Grover V. 2001. General perspectiveson knowledge management: fostering a research agen-da. Journal of Management Information Systems 18(1):5–21.

Davenport TH, Prusak L. 1998. Working Knowledge: HowOrganizations Manage What They Know. Harvard Busi-ness School Press: Cambridge, MA.

David M. 1999. SQL-based XML structure data access.Web Techniques June: 67–72.

Day GS. 2000. Managing marketing relationships. Journalof the Academy of Marketing Science 28(1): 24–31.

Decker S, Melnik S, Harmelen FV, Fensel D, Klein M,Broekstra J, Erdmann, M, Horrocks I. 2000. The seman-tic web: the roles of XML and RDF. IEEE Internet Com-puting 4(5): 63–74.

Devedzic V. 1999. A survey of modern knowledge mod-eling techniques. Expert Systems With Applications 17(4):275–294.

Fahey L. 2001 Linking E-business and operating pro-cesses: the role of knowledge management. IBM Sys-tems Journal 40(4): 889–907.

Fowler A. 2000. The role of AI-based technology in sup-port of the knowledge management value activitycycle. Journal of Strategic Information Systems 9(2/3):107–128.

Fraternali P. 1999. Tools and approaches for developingdata-intensive web applications. ACM Computing Sur-veys 31(3): 227–263.

Friedman-Hill E. 2002. Jess, the expert system shell.Sandia National Laboratories, Albuquerque, NM,URL: http://herzberg.ca.sandia.gov/jess

Gold A, Malhotra A, Segars A. 2001. Knowledge manage-ment: an organizational capabilities perspective. Jour-nal of Management Information Systems 18(1): 185–214.

Goldfarb CF, Prescod P. 1998. The XML Handbook.Prentice Hall: Upper Saddle River, NJ.

Gordon I. 1998. Relationship Marketing: New Strategies,Techniques to Win the Customers You Want and KeepThem Forever. John Wiley: New York.

Helmke S, Dangelmaier W, Uebel MF. 2001. CRM-systems as technology enabler for a customer-orientedknowledge management. PICMET ’01—Portland Inter-national Conference on Management of Engineering andTechnology 1.

Hess TJ, Rees LP, Rakes T R. 2000. Using autonomoussoftware agents to create the next generation of deci-sion support systems. Decision Sciences 31(1): 1–31.

Kalakota R, Robinson M. 2001. E-Business 2.0: Roadmapfor Success. Addison-Wesley: Boston, MA.

Kohli R. 2001. Managing customer relationships throughE-business decision support applications: a case of hos-pital–physician collaboration. Decision Support Systems32(2).

Kotler P. 2000. Marketing Management. Prentice-Hall.Upper Saddle River, NJ.

Leebaert D. 1998. The Future of the Electronic Marketplace.MIT Press: Boston, MA,

Liebowitz J. 1999. Information Technology Management—AKnowledge Repository. CRC Press: Boca Raton, FL.

Liebowitz J. 2000. Building Organizational Intelligence—AKnowledge Management Primer. CRC Press: Boca Raton,FL.

Maes P, Guttman RH, Moukas AG. 1999. Agents that buyand sell. Communications of ACM 42(3): 81–87.

Massey AP, Montoya-Weiss MM, Holcom K. 2001. Re-engineering the customer relationship: leveragingknowledge assets at IBM. Decision Support Systems 32:155–170.

Murch R, Johnson T. 1999. Intelligent Software Agents. Pre-ntice Hall: Upper Saddle River, NJ.

Nissen M, Kamel M, Sengupta K. 2000. Integrated analy-sis and design of knowledge systems and processes.Information Resources Management Journal 13(1): 24–43.

Offsey S. 1997. Knowledge management: linking peopleto knowledge for bottom line results. Journal of Knowl-edge Management 1(2): 113–122.

O’Leary DE. 1998. Enterprise knowledge management.IEEE Computer March.



Orzec D. 1998. Call centers take to the web. DatamationJune.

Parasuram A, Grewal D. 2000. The impact of technologyon the quality–value–loyalty chain: a research agenda.Journal of the Academy of Marketing Science 28(1): 168–175.

Probst G, Raub S, Romhardt K. 2000. Managing Knowledge:Building Blocks for Success. John Wiley: Chichester.

Rabarijaona A, Dieng R, Corby O, Ouaddari R. 2000.Building and searching an XML-based corporate mem-ory. IEEE Intelligent Systems 15(3): 56–63.

Reichheld F, Schefter P. 2000. E-loyalty—your secretweapon on the web. Harvard Business Review July–August.

Shoemaker ME. 2001. A framework for examining IT-enabled market relationships. The Journal of PersonalSelling & Sales Management 21(2): 177–185.

Swift RS. 2001. Accelerating Customer Relationships: UsingCRM and Relationship Technologies. Prentice Hall: UpperSaddle River, NJ.

Sycara K, Pannu A, Williamson M, Zeng D, Decker K.1996. Distributed intelligent agents. IEEE Expert 11(6):36–45.

Tiwana A. 2001. The Essential Guide to Knowledge Manage-ment: E-Business and CRM Applications. Prentice Hall:Upper Saddle River, NJ.

Wiig KM. 1999. What future knowledge managementusers may expect. Journal of Knowledge Management3(2): 155–166.

Winer RS. 2001. A framework for customer relationshipmanagement. California Management Review 43(4): 89–107.

Wu DJ. 2001. Software agents for knowledge mana-gement: coordination in multi-agent supply chainsand auctions. Expert Systems With Applications 20(1):51–64.

Zeithaml VA. 2001. The customer pyramid: creating andserving profitable customers. California ManagementReview 43(4): 118–145.



Documents

Application of knowledge management technology in customer relationship management