best practices in enterprise knowledge managementhcmgt004/kmwhitepapernov06.pdf · the benefits of...

Best Practices in Enterprise Knowledge Management

KMWorld November/December 2006

Andy Moore . . . . . . . . . . . . . . . . . . . . . . . 2 Passing the Cringe Test: Has KM Made It to Prime Time?Once upon a time, saying the words “knowledge management” was the fastest way to get thrown out of a meeting. And not only would you be ejected, you would never be invited back. . . .

Paul Sonderegger, Endeca . . . . . . . . . . . . 4 People Judge Relevance. Machines Calculate Evidence.Does the man make the machine? Or the machine make the man? A pilot can't fly without aplane. A doctor can’t diagnose without a centrifuge. . . .

Daryl Orts, Noetix . . . . . . . . . . . . . . . . . . 6 How to Correctly Design and Implement a DashboardA dashboard is a vital tool for monitoring the daily health of your organization. From a sin-gle interface, decision makers have access to key performance indicators (KPIs). . . .

Hadley Reynolds and Silvija Seres, FAST . . . . 8 Building the Search Center of ExcellenceSearch is strategic; however, the strategic potential of search is not captured by the act ofacquiring a powerful search platform alone. . . .

InQuira . . . . . . . . . . . . . . . . . . . . . . . . . 10 Implementing KM: Practitioners Share Best PracticesMost customer service organizations today will admit that effective knowledge transfer is the most crucial element to resolving customer problems. . . .

Brandon Lackey, BEA, and . . . . . . . . . . . 12 Mistakes to Avoid and Principles for SuccessMany organizations today still struggle to get value from their knowledge management (KM)efforts. Even companies that have been benchmarked as the “best practice” can easily fall. . . .

Jeff Dirks, SchemaLogic . . . . . . . . . . . . . 14 Knowledge is Power: Best Practices in Semantics ManagementKnowledge is power. The meaning of the phrase is even more applicable today as companiesaround the world are dealing with an overwhelming amount of data. . . .

Anand Chopra, KANA . . . . . . . . . . . . . . . . 16 Answering Customers’ Questions the Intelligent Way Enterprises face a difficult challenge when simultaneously improving the quality of service and reducing costs. More products and rapid change substantially increase the amounts of information. . . .

Gordon Taylor, TOWER Software . . . . . 17 Knowledge at the Point of DecisionA small, two-legged robot stands atop a glass-topped coffee table. On its two dimensionalworld, it has to contend with a potted plant, an old TV Guide and several coffee cups. . . .

Laurent Simoneau, Coveo . . . . . . . . . . 18 Enterprise Search: The Foundation for Risk ManagementIf you can’t find it, you can’t manage it. Unfortunately, many executives have discovered thistoo late in the game—when they face legal troubles or a corporate scandal. . . .

Jon Parsons, XyEnterprise . . . . . . . . . . . 19 Realizing Measurable ROI with Multi-Language Content Management What is content management worth? That can be a philosophical question. If you ask a content author, you’ll hear about the way content creation and review is made easier. . . .

Dan Dube, DocZone.com . . . . . . . . . . . . 20 Hosted XML Content Management: Is It Right for You?Publishing organizations have long recognized the value of migrating content to XML to attainthe benefits of content reuse, reduced localization costs and single-source publishing. . . .

Gregory F. Roberts, Lockheed Martin . . . 21 Who is that “he?”—Using Pronouns and Anaphors in Text ExtractionText extraction is a powerful tool to find and categorize elements in unstructured documents.These elements, or entities, are connected together to form the relationships, facts and events. . . .

Andrew Cohen, KNOVA Software . . . . . . 22 Beyond First Call ResolutionKnowledge management (KM) initiatives are one way of improving technical support organizations, driving support margins through efficiency and increased customer loyalty. . . .

Chip Gettinger, Astoria Software . . . . . . 23 The Hidden Costs of Product Information PublishingManufacturing enterprises have placed a sustained focus on information management solutionsto support and augment the design, development and production process. . . .

Premium Sponsors

Michael Behounek, Emerja

crete,” insisted Sally Hicks, marketingmanager at Noetix. “Our biggest dashboardcustomer is Florida Rock. Their name saysit all—they move rocks. It’s not sexy. Butthey allow their customers to log in througha secure dashboard and get a few basicpieces of data, and it’s saving hundreds ofman-hours per week. Their customersdon’t know the terminology of ‘knowledgemanagement.’ They just want a copy of aninvoice.”

“We are able to now talk about knowl-edge management without the audiencerolling their eyes,” said Sivija Seres,FAST’s VP of strategic business develop-ment, “but we have learned a lot. Peoplehave been told one thing at a marketinglevel, but it’s difficult to achieve thosethings when it scales greatly and gets morecomplex, and more features creep up intothe original specification,” said Silvija.“Recently, the technology—not the core,but the functionality—and our understand-ing of how to set things up from the begin-ning has reached a level that we can bereally proud of.”

Unmet expectations is a common themein technology deployments, of course. Butmaybe it’s worse in KM, because the goalsare often so ambiguous. “The problem withasking users to describe what they want islike the old saw about Henry Ford,”remarked Paul Sonderegger. “If Ford hadasked people what they wanted, they wouldhave said ‘faster horses.’ Luckily, there is arising generation of managers who have agreater baseline of familiarity with technol-ogy than the departing generation. They’restill not technologists, but they have newideas about what they want from theirinfrastructure.”

Lowell Anderson, VP marketing atSchemaLogic, agreed with that assessment.“There’s definitely an increased awarenessof the capabilities that a focused KM strat-egy can bring to an organization. As we become familiar with collaborationtechniques, people are becoming morecomfortable with moving away from fixedstandards as a way to govern knowledge.Open collaboration effectively becomes thestandard,” said Lowell.

The KM FormulaIf KM is now accepted among polite

company, what’s holding up its widespreadadoption? “We always come back to thesame fundamental thing: people can’t getthe information they need to do their jobs,”said Sally Hicks. “Even if IT is helping getinformation to people, they don’t get itquickly enough, or it changes. Sometimesthe users don’t even know what it is theywant; they just know they need somethingin order to make a decision.”

The theme of “information+decision=knowledge management” is practically universal. “What’s the connection betweeninformation and action?” asked Paul, entirelyrhetorically. “It’s the assessment of the prob-lem and the appropriateness of available solutions. Before you can do anything logisti-cally—like shipping something from oneplace to another to fix a problem—someonehas to assess the problem, and become satis-fied they have identified the right solution.And that’s ALL about decision-making. It’sthe information being brought to the personthat allows us to talk about the application of knowledge.”

Of course, as is often the case, it’s easi-er said than done. “As customers spend sig-nificant amounts of money deployinginformation-access networks, they are real-izing they aren’t working as well as they

Passing the Cringe TestHas Knowledge Management Made It to Prime Time?

Once upon a time, saying the words“knowledge management” was the fastestway to get thrown out of a meeting. And notonly would you be ejected, you would neverbe invited back.

That was then, but this is now. I asked apanel of experts not only whether KM wasfinally ready to pass the “cringe test,” butwhat it would take to propel KM to evengreater prominence as an enterprise set ofsolutions. And I was surprised—andpleased—at the reactions.

“One of the big failures of knowledgemanagement, and the reason it had the rep-utation it had, is that it tries to solve prob-lems at too much of a macro level,” saidBrandon Lackey. Brandon is the globalsolutions director for BEA Systems. “Theytried to build a database of everyone’sknowledge, and it can’t be done. The suc-cessful systems may go in under the basisof knowledge management, but theyaddress very specific business problems.”

So can you “talk about KM” duringthese meetings? “You can, in pockets,”noted Paul Sonderegger, the principalstrategist and “evangelist” at Endeca.“There are companies that are moredependent than others on the movement ofinformation in order to affect the bottomline. Take a professional service organiza-tion—a consulting company, for instance.Its whole ability to leverage the work theirconsultants do depends on their ability totake information others have created andapply it in another context. In that kind ofenvironment, yes...you can absolutely talkabout knowledge management.”

Tim Shetler, vice president, marketingat InQuira, thought his company knew thedifficulty in the terminology: “We deliber-ately named our product ‘InformationManager’ to avoid any issues with the term‘knowledge management.’ We discoveredthey don’t exist! We thought there wouldbe some tarnish around the term. But everyRFP I look at uses the phrase ‘knowledgemanagement.’ The end users don’t havethat baggage.”

It’s probably a matter of whom you ask,and what type of organization they workfor. “The term ‘knowledge management’doesn’t make sense to a guy pouring con-

November/December 2006S2 KMWorld

Andy Moore is a 25-year publishingprofessional, editorand writer whoconcentrates onbusiness processimprovement throughdocument andcontent management.As a publicationeditor, Moore most recently waseditor-in-chief and

co-publisher of KMWorld Magazine. He is nowpublisher of KMWorld Magazine and its relatedonline publications.

Moore acts as chair for the “KMWorld BestPractices White Papers,” overseeing editorialcontent, conducting market research and writingthe opening essays for each of the white papers in the series.

He has been fortunate enough to cover emergingareas of applied technology for much of hiscareer, ranging from telecom and networkingthrough to information management. In this role,he has been pleased to witness first-hand thedecade’s most significant business andorganizational revolution: the drive to leverageorganizational knowledge assets (documents,records, information and object repositories) toimprove performance and improve lives.

Moore is based in Camden, Maine, and can bereached at [email protected].

Andy MooreBy Andy Moore, Editorial Director, KMWorld Specialty Publishing Group

November/December 2006 S3

possibly could,” said Lowell. “They areconnecting different repositories and con-tent silos, but they don’t have a commonunderstanding of the terminologies thatdescribe the information residing in thosecontent silos. There are also complex inter-relationships among those terms they haveno way to model.”

Sally Hicks agreed: “The terminologyissue is huge. If you’re working with adatabase, and there isn’t a term that exactlymatches ‘table 1’ in the database, there isabsolutely no way you’re going to get whatyou need. In many cases, the user isn’t evensure he HAS the information, much less beable to find it using strict database terms.”

And it doesn’t appear that the averagetechnology tools provide much help; amore integrated solution is required. “A lotof customers feel that ‘search’ by itself isnot as big a problem as content or knowl-edge,” said InQuira’s Tim Shetler. “Most ofour customers now come to us looking forboth. It’s a much broader solution area.”

“A useful lens to look at this through isthe long tail,” said Paul. “Within an enter-prise, there will be a number of requests forinformation that everyone asks all thetime—total blockbusters. Then there’s thebody of requests that are kind of in the mid-dle, relatively common but not block-busters. But then you’ve got an almost infi-nite number of requests that occur seldom-ly... maybe only once. So any kind of ‘mas-ter schema’ will fail.”

Getting in the DoorPerhaps because of the complexity; per-

haps because of the ambiguity of the benefit,KM is a tough sell. “Selling knowledge man-agement systems is harder than selling sys-tems that help people make money,” saidSilvija. “You have to provide a very strongROI argument. You have to help customersunderstand how it will either make them moreeffective, or save money,” she said.

“The difficulty begins when you want toroll it out. You have to convince the IT personat every division or office. For them this is justa headache; they have to replace somethingthat’s already in place and working withsomething that will take time to get right. Inorder to grow a system to a very large scalewith very advanced features, you need to havea nicely staged process, where they first getsomething that works really well at a basiclevel, then add new features and grow the sys-tem in a controlled way.”

“When it works right,” added BrandonLackey, “IT recognizes that the businessproblem has to be solved by the line-of-business stakeholders. IT should coordi-nate, and see themselves as the enablers. Somuch of knowledge management is peopleand process; if the business says to IT‘Make me a repository,’ IT should push

back and say ‘No, this is so much morethan that. You need to make sure theprocesses are open and the right people arein place.’” All agreed that Brandon’s viewis correct, but utopian. “Business wants aquick fix, and that’s been the failure ofknowledge management: there isn’t one.”

“IT will end up owning and running thesystem, so it’s OK for them to be incharge,” added Sally. “But they need tohave the input and buy-in from business,and listen. And the end users also have totake the time to work with IT and ask forexactly what they want up front.”

“The more ambitious of our customersknow they have to view it as cross-functionaland cross-divisional. They see it as a platform,and on top of that they keep growing proj-ects,” explained Silvija. “They are starting‘centers of excellence’ in greater numbers.But in order for them to succeed, you needsupport from quite a high level.”

What’s Next?Tim Shetler detects a “mood shift”

among KM system users and purchasers:“The resurgence of KM is driven by thedistribution of workers, and work, bothgeographically and over time. For compa-nies like that, knowledge management issomething you have to do just to run thebusiness. But there’s another change: peo-ple are looking for ways to affect their topline. Sure, people are still looking for waysto cut costs, and get ROIs from cost cen-ters, such as their call-center operations.But people are tired of doing that; they’relooking for ways to raise the top line—‘Give me a 2% increase in marketshare.’”He continued: “An enterprise knowledge-sharing initiative has a lot of hard-to-quan-tify, soft-dollar benefits. That’s hard to justify. But if you can demonstrate a hard-dollar return, it’s much easier to spreadwhen the case is proven.”

The most interesting turn in our con-versations revolved around the opportuni-

ties for KM to leverage emerging Webtechnologies and concepts. “Certainorganizations, such as telcos and mediacompanies, are very keen to figure out howto make money from Web 2.0,” declaredSilvija, “and apply the whole social aspectof information. We need to provide themwith the tools to position themselves inthat space.”

Brandon Lackey agreed: “I’m excitedabout Web 2.0, because it builds communitiesof like-minded contributors who share implic-it knowledge, but in doing so it’s being cap-tured and made searchable. So while they’resolving specific business problems, they’realso building a valuable knowledgebase...justin a subtle and implicit way.”

“People are putting KM systems in placewhich are quite capable at organizing and dis-tributing knowledge,” added Lowell. “Buthow do they enrich those systems in real timewith knowledge from the employees’ minds?There’s a tremendous amount of interest inusing the concepts of Web 2.0 to use collabo-ration to enrich and expand the informationwithin these systems.”

It’s popular to think of KM technolo-gies as being “transparent” and totallyautomated, but some of the panel thoughtjust the opposite. Sally said, “What’s dif-ferent now from five or 10 years ago isthat users want to ‘do it themselves.’ Backthen, they just said: ‘Give me a report,give me a number.’ Now they want to be empowered.”

“People are concerned with the qualityof information. And they have becomespoiled, because there are so many moresolutions. How do you provide them withthe information they want, even if theycan’t express what it is they want?” won-dered Silvija.

I guess that remains to be seen. It’s cer-tainly true that a “one-size-fits-all/it slices, itdices” solution is probably a thing of the past.Even with search, it is rare to find a vendorthat only uses one type of technology. It’s acombination of text with semantic under-standing on top of a taxonomy...as KM getsbetter and enters the mainstream, it also hasbecome extraordinarily more complex.

That’s what the KMWorld White Papersare for. It is no longer possible to explainKM in a simple “blurb” or describe in acolorful brochure. Information manage-ment solutions are not only complex, theyare usually multi-dimensional, solvingmany different problems, depending onhow they are applied and where the needsof the organization lie. And don’t forget:solving one problem only opens the oppor-tunity to face the next one. But that’s whatknowledge management solutions havealways done best. ❚

KMWorld

“Every RFP uses the

phrase ‘knowledge

management.’ The

end users don’t have

that baggage.”

Is Relevance Irrelevant?The familiar framing of the information-

access problem is how to help people findthe right information at the right time. Butwe need some measure of “rightness” beforewe can characterize the tools we need.“Relevance” is the technical term for thecentral measure of “rightness” in informa-tion retrieval. Unfortunately, relevance defieslogical rules. You know this if a searchengine ever returned to you a document thatwas “72% relevant.” This is like a flight com-puter saying the plane is 72% on course.

Relevance is subjective because it’s rela-tive to the person’s goal. For example, if I tellyou that the movie tonight is at 7:30 PM,is that relevant? If we’re going to the movies, then yes. If not, no. How can a piece

of information be both relevant and irrele-vant? It depends on what the user is trying toaccomplish. Relevance is in the eye of thebeholder.

Just because relevance is subjectivedoesn’t mean we must give up hope in find-ing it. On the contrary, it must be the central focus of information access tech-nology. But it does reveal why we should-n’t ask the machine to determine relevance for us. It can’t. Fortunately,people are experts at judging relevance—and with help from machines, we can do

better than without. The central function ofinformation access technology is to produceevidence that informs human judgment.We summarize this as:

◆ People judge relevance◆ Machines calculate evidence

Context Informs RelevanceSo what can a technology do that would

help people better determine relevance?First, let’s look at how context helps peopledetermine meaning. Take this simpleexchange:

Host: “Would you like some dessert?”Guest: “It would fill me up.”

How does the host figure out if the guestwants dessert or not? If the host knows that hisguest doesn’t like to sleep on a full stomach,then the answer is no. If the host knows thathis guest likes to eat until satisfied, the answeris yes. This context lives outside the literaldialogue in the mind of the host. It completesthe proper meaning of the guest’s statements.But how does the guest know the host will make the proper interpretation? He’sactually relying on the host to have this context in mind and, therefore, interpret hisstatements accurately.

This is how human communicationworks. It’s based on the resolution of ambi-guities through best efforts and educatedguesses. But both parties have to be workingwith the richest context-processing machineavailable—the human mind. Both guest andhost use observations about the current environment, knowledge of the audience and personal experience to attempt to communicate effectively. The vast majorityof the time it works. And the application ofthese contextual facts, taken from the envi-ronment and the head, is the key.

People Judge Relevance.Machines CalculateEvidence.

Does the man make the machine? Or the machine make the man?

A pilot can’t fly without a plane. A doctorcan’t diagnose without a centrifuge. A pho-tographer can’t create without a camera. Andexcept in the simplest circumstances, the re-verse is also true. But why? Couldn’t youbreak down flying, diagnosis and photogra-phy into a set of rules that would allow themachine to make the same decisions the per-son would have made if he had been there?No. There is something more to the humandecision-making process than simple logicalrules can account for.

This fact is a central problem for tra-ditional information-access technologies,including keyword search, databases and

business intelligence (BI). Each of these“machines” intends to help an individualfind the information she needs to make adecision at a unique point in time. This isone of the most difficult technologicalproblems of our age, and one of the mostvaluable to solve, but too often we’reasking the centrifuge to do the work ofthe doctor. Instead we should considereach person the expert and provide moreeffective tools for reaching their goals.But how do we evaluate which tools arethe right ones for the job?


Paul Sonderegger isEndeca’s chiefstrategist. Beforejoining Endeca,Sonderegger was ananalyst at ForresterResearch for sixyears. His researchfocused onexperience designand searchtechnology.Sonderegger’s

research into search included informationretrieval theories and technologies. But it wasexamining those approaches through the lens ofuser experience that showed which onesactually help people find what they’re lookingfor. In addition to publishing numerous reportson these topics, Sonderegger helped hundredsof executives apply research to solve realbusiness problems.

Paul Sonderegger

By Paul Sonderegger, Chief Strategist, Endeca

“We should consider each person the

expert and provide more effective tools for

reaching their goals.”


If we are to make humans the determi-nant of relevance in an information accessapplication, it is context that helps themjudge information. But how do we get fromcontent to context?

Relationships in the Content InformContext in the Head

In the man/machine exchange of infor-mation access, only one party is reasoningwith the flexibility and inventiveness of ahuman being. To make the machine play ahelpful role in the dialogue, there are twobasic approaches:

1. Mimic human judgment with software rules

2. Use software rules to inform human judgment

The first works well in constrained, staticuser scenarios. For example, a digital camerasensing low light will leave the “shutter” openlonger, just as a photographer would. Butthere’s no setting for capturing, say, a child’sjoy of accomplishment. And if there weresuch a setting, would you trust it on the day ofyour child’s big game? Search enginesattempting to calculate relevance are over-reaching just as much. These algorithms arereally just counting—whether it’s the numberof times the query term appears in the targetdocuments, the number of links using thatterm which point to a given Web page, orsomething based on probabilistic or other sta-tistical calculations. In each case, the raw cal-culations fail to capture what the indexedinformation is about. Worse, the mechanism isa black box. When the user gets results thatappear irrelevant to him, it’s hard to tell whyhis best-attempt query failed.

The second approach works well indynamic, unpredictable scenarios. Think ofa doctor in an emergency room. A patientarrives complaining of weakness, shortnessof breath and numbness in his extremities.A blood test shows low oxygen saturationand a chest x-ray shows fluid in the lungs.Neither of the machines that produce thisdata can draw the conclusion that thepatient was poisoned. But the doctor can.The tools provide the doctor with facts thatbetter inform her judgment. But it’s thedoctor who puts the whole picture together.In information-access technologies, data-bases and BI tools are much closer to thisapproach. They store data in specific struc-tures, allowing them to maintain the factu-al relationships inside. The key is to pointthe machine at the smallest indivisible unitof information and then give the user thetools to see those pieces as a whole picture.

That unit is the facet—an explicit charac-teristic of a thing. Facets are everywhere—metadata on documents, user-contributedtags and fields in a database record are all

facets. And in a strange twist, the full-textindex of a document can be thought of as justa dynamic facet of that file.

But don’t relational databases work thisway today? Not quite. People easily conceiveof dimensions that cut across a “rectangular”database. Take a customer database; it’sstraightforward to find out which customersbought a particular product in the past year.But it’s much more difficult to find out whichcombinations of products were bought in thelast year by customers in a particular region—a sensible question, based on dimensions thatcut across the original database structure.That’s what BI is for. Unfortunately, BI canonly support the questions the developersknew ahead of time would be asked. But peo-ple are much more flexible than that. Theycome up with unforeseen questions andunpredictable ways to express otherwise mun-dane requests. And, they want to take intoaccount information in different forms fromrecords to documents. That’s what search isfor. But, of course, traditional search enginesdon’t maintain relationships from the systemsthey index. And we’re back where we started.

A New Architecture CalculatesRelationships Among Relationships

Just as modern planes, medical tech-nologies and cameras give pilots, doctorsand photographers more flexibility andcontrol, a modern information-access plat-form must do the same for employees, cus-tomers and partners. Such a machine must:

◆ Accept messy, real-world enterprise information. The platform must be able toindex enterprise data and content thatcomes in all its different formats, sizes and quality.

◆ Preserve all the relationships in theoriginal systems. The relationships in thedatabases, enterprise apps and documentsare pre-existing investments the platformmust exploit.

◆ Calculate all the dimensional relation-ships. The facets in each record and doc-ument are the basis for connections acrossotherwise completely different assets.

◆ Guide users through constantly shiftingcontexts. Each time the user takes a step

through the app—whether a keywordquery, a navigation selection, a circled re-gion on a map—the platform must showhim the results and all the possible, butonly the valid next steps.

These requirements call for an architec-tural capability called adaptivity. Adaptivityis the dynamic calculation of relationshipsamong relationships in the current results setbased on the possibilities in the data, theuser’s actions and any business rules.Information access applications built onsuch a platform show a user all the results toa query plus all the information about thoseresults, creating greater context for theuser’s judgments. For example, a productengineer looking for titanium bolts for alightweight chassis design searches for “tita-nium bolts.” The results include all the tita-nium bolts, of course, as well as all thedimensions by which this list of bolts couldbe refined—length, weight, finish, threadpitch and so on—plus the quality-assurancereports including data on titanium bolts,with the appropriate refinement dimensions.And as the engineer selects refinements orsearches within the results, or both, all theinformation on the screen dynamicallyupdates, driven entirely on the possibilitiesthat exist in the data. The result is the richestpossible presentation of the evidence, guid-ing the engineer to his goal.

Machines should not pretend to dothings they can’t. And they can’t judge rel-evance for a particular user with a specificgoal. Instead, the machine should calculatethe relationships among relationships in thedata and content, fueling the context fordecisions, making experts of us all. ❚

Endeca, headquartered in Cambridge, MA is a next-gen-eration information access company uniting the ease ofsearch with the analytical power of business intelligence.The Endeca Information Access Platform combinespatented intellectual property, breakthrough science anda deep focus on user experience to help people find,analyze and understand information in ways neverbefore possible. Leading global organizations like ABNAMRO, Boeing and Cox Newspapers rely on Endeca to increase revenue, reduce costs and streamline operations through better information access.

KMWorld

“Adaptivity is the dynamic calculation

of relationships among relationships in the

current results set.”

your dashboard project by striking a bal-ance between the primary user’s needs andwhat you can afford to deliver.

Requirements gathering and prototype—Interview the key stakeholders to determinetheir needs and expectations. To keep thedashboard project within scope, map theseneeds and expectations to the pre-establishedKPIs. To increase the likelihood that the

final dashboard will meet users’expectations,take advantage of available tools and technologies that lend themselves well to prototyping.

2. Design: Once the team approves the dashboard’s content and appearance,the next step is to incorporate major design aspects:

◆ Refine the user interface and control flow;◆ Confirm the data sources for each

data element;

◆ Determine how to “persist” data when his-torical trending information is desired, butunavailable from the transaction database;

◆ Define the queries needed to retrieve eachdata element; and

◆ Determine drill paths.

3. Build and validate: The “real”development begins at this stage of theproject. Several tasks occur here, typicallyin parallel and closely coordinated witheach other.

Front-end implementation—Create thedashboard user interface. Evaluate whatgraph and chart types best represent thedata to be displayed and make decisionsregarding grouping data to provide thegreatest visibility for cross-analysis.

Query implementation—Create thequeries to retrieve the necessary informationfrom the appropriate databases. This step canbe particularly complex and time consum-ing, especially if there are multiple datasources for the various dashboard elementsincluding data from customized enterpriseapplications for ERP, CRM or SCM.

Configure scheduling, refresh and secu-rity—To ensure the dashboard content isup to date, queries need to be configured torun regularly. At the same time, it is impor-tant to establish and implement securityrules to display the appropriate informationfor users with different levels of access.

Dashboard validation—As with anysoftware project, when the effort reaches“code complete,” both the technical teamand the primary users must test the dash-board to ensure it meets the requirementsoutlined in the project plan.

4. Deploy and maintain: Once thedashboard has been built and tested, it isthen deployed into production and securityrequirements are implemented. With the

How to Correctly Design and Implement a DashboardA Step-by-Step Guide

Adashboard is a vital tool for monitoring thedaily health of your organization. From a sin-gle interface, decision makers have access tokey performance indicators (KPIs)—action-able information that can be used to effectivelyguide and track business performance.

At a high level, it may seem relativelyeasy to build a dashboard. Companies thatfeel they have a good handle on whichperformance indicators are of strategicimportance to the organization may thinkcollecting, summarizing and consolidatingthe supporting data shouldn’t be that diffi-cult. However, such oversimplificationcan lead to a failed project before it evergets off the ground.

The successful implementation of adashboard is complex and requires astep-by-step process: a methodology thatconsiders all aspects of the project lifecycle. This series of tasks—plan, design,build and deploy—will be similar,regardless of the technology or vendorchosen. When comparing proposals frommultiple vendors or the cost of a “do-it-yourself” project, it is important toinclude all of these steps. Correctlydesigned and implemented, a dashboardhas the potential to bring immediate andconsiderable return on investment (ROI)to your organization.

1. Plan: Dashboard developmentbegins in the planning phase. Identify theproject team members, their roles andoverall project objectives. When workingwithin a tight timeline, populating thedashboard is the most critical area of con-cern. Take care not to underestimate thecomplexity of the databases in which thedata resides. Accessing the data from a myriad of tables requires technicalresources with detailed knowledge of theunderlying table structure and consider-able SQL skill. Define the project budgetand take into consideration the workrequired to create the custom queries forthe desired metrics. Set realistic goals for


Daryl Orts isresponsible formanaging thegrowth anddevelopment of theNoetix Dashboardoffering, bringingmore than 10 yearsof technology andleadershipexperience toNoetix. Orts joinedNoetix in early 2001

and managed the development of NoetixViewsfor Oracle Applications for two years. He alsomanaged the development and launch of thefirst two releases of Noetix for Siebel CRM. Priorto joining Noetix, Orts was vice president ofapplications R&D for Onyx Software as well asvice president of engineering for LuminousCorporation. Earlier in his career, Orts heldsoftware management and developmentpositions at Adobe Systems, Frame Technologyand IBM.

Daryl Orts

By Daryl Orts, VP of Advanced Solutions, Noetix Corp.

“It may seem

relatively easy to

build a dashboard.

However, such

oversimplification

can lead to a failed

project.”


dashboard in production, or “live,” stepsmust also be taken to provide for ongoingmaintenance. Over time, requirements andexpectations for the dashboard will changeand the dashboard solution should be flex-ible and open to allow for such inevitableenhancement requests.

Final Note: Build vs. BuyBuilding and deploying an executive

dashboard takes time, regardless of the

vendor or technology that is chosen.Creating the graphical front end is rela-tively quick and easy, but that’s merely theshell of the dashboard. What you actuallysee on your desktop pales in comparison tothe hidden effort—80% of the complexitylies beneath the surface. All of the taskslisted above require planning, organiza-tion, coordination, scheduling and solidproject management. When comparingproposals or considering a “build vs. buy”decision for deploying a dashboard, it is

imperative to ensure that the entire scopeof the project is considered. ❚

Noetix provides business intelligence tools that enablemore than 1,300 customers worldwide to quickly andcost-effectively access the enterprise application datathey need to make important business decisions. Unlikemost BI tools that require weeks of extensive manualmapping to be set up and maintained, Noetix usespatented technology to automatically discover and produce metadata based on customers’ specific implementations of Oracle Applications or Siebel CRM.For more information please visit www.noetix.com.

KMWorld

Founded in 1945, Florida Rock Industries, Inc., is one ofthe nation’s leading producers of construction aggregates,ready-mixed concrete, concrete block, Portland cement andpre-stressed concrete. The company is also at the forefrontof technology in the construction materials industry.

When Florida Rock receives a customer order, itassigns a “ticket” to each request for materials. The ticket acts as the proof of delivery. It is signed by the customer when the shipment is received and scannedinto Florida Rock’s electronic document retrieval system.Once the signed ticket enters the system, Florida Rockadds the associated customer order to an invoice. Therecan be one ticket or a hundred tickets on an invoice, sotying the ticket back to the invoice quickly and efficientlyis critical for providing the kind of quality customer service for which Florida Rock is known. The companycan field hundreds of calls per week from its more than8,000 customers looking for information regarding theirinvoices. The process involved a customer going througha telephone answering service to reach the correct person, who would then query an Oracle database orFlorida Rock’s electronic document retrieval system toget to the needed information. This image or copy wouldthen be emailed, faxed or mailed to the customer.Florida Rock also saw the need to make it more efficientboth for its customers and for its credit and accountsreceivables (AR) departments, who spent several hoursa day just supplying lost copies of documents.

Addressing the Challenge

The solution involved a creative way to use the NoetixDashboard as a customer self-service tool. Instead ofhaving to call for information, customers are able to signon to a secure system that takes them directly to a custom-tailored Noetix Dashboard. It shows them theirAR buckets; allows them to view their most recent 50open invoices and all the associated tickets, and lets

them see receipts or checks that have been posted totheir accounts. Finally, it gives them a list of local FloridaRock locations if they need to do more business.Essentially, Florida Rock’s customers are getting a greatdeal of information about their business in a matter of seconds.

In addition, customers can link directly from theirinvoices to Florida Rock’s electronic document retrievalsystem and pull out a PDF with the exact image of thedocument, or drill down into the tickets that made upthat shipment or particular invoice, as well as make acopy of their most current statement.

The Noetix Dashboard provides Florida Rock customers with a self-service tool to facilitate a more efficient business partnership. It is much easier for themto monitor their accounts and ensure their invoices arepaid on time, cutting back on overdue payments. “Ourcustomers require that all of the signed tickets be filed,”said Dave DeVore, manager, application development,Florida Rock. “If one ticket is misplaced and there is noproof of delivery, payment is withheld on an invoice thatcould be worth several thousand dollars. The easy-to-useNoetix Dashboard will help shorten the payment cycle bygiving customers immediate access to all of the ticketsand invoices associated with each account. We want ourcustomers to have control over their businesses and getthe information they’re after in a timely manner. Logginginto these dashboards is certainly a lot quicker and easier than calling, going through the various levels, andthen waiting for a fax or image to come through.”

In addition to the customer self-service tool, Florida Rockuses Noetix Dashboard internally. The company’s creditdepartment staff uses a dashboard to get quick customersnapshots and access to associated statements, invoicesand tickets without having to go into Oracle ERP. FloridaRock also intends to roll out dashboards to its executives inHR payroll and receivables to monitor business operationsfrom a single, intuitive view.

Florida Rock Institutes Customer Self-Service Dashboards

search. Forrester Research, for example,has consistently reported breakdowns insite-search quality. Recent research showsthat 58% of 211 websites reviewed throughmid-2006 failed to meet basic criteria forsite search engine and search interfacequality. Failure rates for clarity and presen-tation of navigation options were in thesame range or higher. At the same time, thefirm’s demographic research finds thatfindability and navigation are even moreimportant to online site visitors than thequality of information on the site or therange of functions available.

Looking at information breakdowninside the organization, IDC Research hasfound consistently that the cost of wastedtime on the part of professionals searchingbut not finding information is a major

continuing cost to organizations. The mostrecent 2006 “Hidden Costs of InformationWork” report suggests that this costamounts to $5.3 million annually for anenterprise with 1,000 information workers.

Delphi Group reported that more than50% of professionals surveyed report beingeither dissatisfied or very dissatisfied withthe search experience in their firms, whileonly 15% reported that their firms had anenterprise search strategy in place.

These research examples show that theissues with achieving high-quality searchgo much deeper than a selection of tech-nology. Some of the problem is clearlyrelated to the legacy of basic keywordsearch deployments whose fatal lack of accuracy and relevance continues to disappoint users. Most modern platformofferings combine families of advancedlinguistic and statistical functions that aremore than adequate to deliver highly accu-rate and contextually significant results in a rich analytic framework with suggestiveand intuitive navigation options. We maintain that many of the issues of search quality arise not from technology limita-tions, but from the unnecessarily limitedimplementation practices which most firmshave resorted to in deploying search.

Fortunately there is a constructive solutionto the challenges described above; we are see-ing impressive results at a number of firmswho are making search quality a priority.Businesses as diverse as Merrill Lynch,Pfizer, McGraw-Hill, Autotrader.com andYouTube are dramatically raising the level ofthe search experience they offer their onlineaudiences. They are replacing “one-size-fits-all” thinking with a management process thatsecures business acumen and measuredinvestment strategies at the center of thesearch deployment. The new focus is ondeveloping core organizational resources andtailored governance capabilities that willdeliver business value across multiplesearch-powered applications.

Search Quality DriversBefore we describe the approach in

more detail, let’s take a closer look at thekinds of competitive and business driversthat lead these pioneer firms to deliverbest-in-class search.

The Internet has made everyone moredemanding when it comes to search per-formance and intelligence. Customers andemployees have all become acclimated tothe apparent effortlessness of Web searchon MSN or Yahoo! or Google.

Self-service is no longer just for shopsor gasoline “service stations.” Today it isalso the accepted access model for infor-mation. Customers and employees nowrequire, as well as expect, self-service toolsable to mine all the information sourcesthey should have access to and to deliverrelevant results in a familiar and comfort-able environment.

Organizations that can deliver the rightinformation at the right time with the rightsearch behavior reap dividends fromincreased online sales and from empow-ered employees. In order to gain thesereturns, information access needs to gobeyond the “one box/one button” paradigmand adjust user experiences to match their

A Sustainable Advantage

Building the SearchCenter of Excellence

Search is strategic; however, the strategicpotential of search is not captured by the act of acquiring a powerful search platformalone. Pioneering firms are now developinga new kind of management approach to help deliver maximum value across multiple search-driven applications: the“search center of excellence.” It is a structured approach, utilizing a focusedcross-functional team, and it is emerging asa practical tool to drive search innovationand deliver high quality online experiences.

This is the age of search; search is becoming the de facto infrastructure forfinding and delivering information. It isubiquitous in new online business applica-tions, driving revenue and capturing opera-tional efficiencies inside the organization.Any organization whose operations touchthe Internet, or important digital informa-tion in general, is finding that delivering better search is good for business.

Yet despite the scale and importance ofthis trend, many companies can’t seem toget out of their own way as they beginusing search. For example, many firmshave fallen into what we refer to as the“one-size-fits-all” technology purchasesyndrome. In this mode, the enterprisesearch problem is seen (at least by thesponsors) as solved as soon as new “enter-prise” software is installed on a productionserver. In such cases, however, the value ofthe solution often fails to impress usersinside the company or customers and part-ners outside, because it simply does notseem to “get” their particular business sit-uation. This is because the core of all suc-cessful search experiences is built onunderstanding the enterprise business con-text and the knowledge drivers that powereach specific set of business interactions.

One indicator of the challenges posedby this current state of the practice is thatvirtually all researchers into search qualitycontinue to report user frustration in both external and internal applications of


By Hadley Reynolds, VP Center for Search Innovation and Silvija Seres, VP Strategic Market Development, FAST

“Many firms

have fallen into the

‘one-size-fits-all’

technology purchase

syndrome.”


roles, the context of their questions, theirvocabulary and their purchase or work pat-terns. Users do not want to know about themultitude of different formats being con-sumed, analyzed, contextualized and per-sonalized for consumption by the searchplatform—they just want the system “towork” across this universe of information,with the most suggestive and relevantresults. The best of these results are, in fact,delivered by composite and intelligentbusiness applications, built on the searchplatform and engineered to unify views ofan arbitrarily complex information “space.”

With this new class of intelligent busi-ness applications, “search” is moving far“beyond the box” and taking up a role thatis as central to today’s Internet-connectedbusinesses as relational databases and ERP systems were to the pre-Web era of IT. In the most advanced enterprise and commerce implementations today,“search” acts as the crucial business infor-mation filter—bringing to life the “longtail” value resident in both enterprise andInternet information.

Taking advantage of these new capabil-ities in search platforms requires an evolu-tion in internal processes and practices foraligning business goals with technologymanagement. It requires a new collectionof skill sets for developing compositeapplications with disparate data. It requiresa change in thinking from data models toconsumption paradigms.

Enter the Search Center of ExcellenceIn adopting a “search center of excel-

lence” (COE) approach, organizations aremoving search from the level of technologydetail to the level of business innovationand strategy. With a strategic approach andexecutive-level support, they are adopting a centralized management capability to

turbocharge a diversity of search projectsacross the enterprise.

The development process for the COEtakes place on a number of levels. In orderfor the COE to succeed in its managementand strategy dimensions, it must first andforemost gather sponsorship and authorityat the senior-executive level. The COE gov-ernance model establishes the interactionprotocols between the COE and the variousbusiness units and technology supportgroups of the enterprise. In order to alignsearch projects with business objectivesand to leverage the benefits of knowledgesharing, experiential learning and technicalsearch expertise, the COE unifies execu-tive-level interactions among all the busi-ness units who are or will be making use ofsearch technologies.

Within the operations of the COE itself,the key success factors are: (1.) ensuring thatappropriate roles are identified to support theanticipated activities of COE projects (seesidebar, “Typical Roles in a Search Project”)and (2.) ensuring that the competency mod-els and interaction patterns for those roles arethoughtfully specified.

Search technology makes uniquedemands across the entire spectrum of tra-ditional IT roles, from systems analyst toarchitect to developer to database adminis-trator to user-interface designer. It alsointroduces non-traditional knowledge engi-neering and customer-experience manage-ment elements to projects and programs.

The COE leverages resources fromacross the firm and potentially across thecustomer and partner universes. In order todeliver high business value and quality userexperiences, the COE incorporates inputand participation from line-of-businessmanagers, business-process designers andbusiness analysts, human-factors experts,business-intelligence analysts, merchandis-ers, marketers and other stakeholders ofsearch applications.

The COE practice brings together peo-ple with deep business-domain expertise,broad search-applications experience,cutting-edge software infrastructureknowledge, complex project-managementskills and demonstrated facility in knowl-edge transfer. This group has the ability toact as a central point of contact to facilitatecollaboration between lines of business,functional specialties and customer,service provider or partner resources. Itmay provide the resources to staff each ofthese components of search projects: appli-cation architecture and design, projectmethodology, best practices and stan-dards, user interface design, educationprograms, support services and analyticsfor continuous improvement.

By putting a dedicated team in place,companies adopting the COE process gainthe ability to:

◆ Identify core patterns of search success;◆ Share best practices and facilitate innovation

in next practices; and◆ Leverage search technology, knowledge

engineering and search infrastructureskills across the enterprise.

The Search Center of Excellence Practice

Using the COE to create a repeatableprocess, common business rules, standardbest practices and custom methods andcomponents tailored to the business willdrive down the cost and improve the suc-cess rate of implementing search projects.

By providing a project office for search,the COE practice can integrate training programs, business consulting and projectportfolio prioritization, best practices exam-ples, advanced solutions “tiger teams,”implementation services, applicationmonitoring and continuous improvementservices. These capabilities deployed withinthe context of an organization’s business priorities offer a fast-track approach to highsearch quality. In our view, it’s time to adoptthis approach to driving business innovationwith quality search. ❚

Hadley Reynolds works in Boston as a vice president andthe director of the Center for SearchInnovation at FAST. He has been anindustry thought leader for over adecade, researching, speaking and writing on new trends in search technology and business practice.

Silvija Seres works in Oslo as a vice president of strategic market development with FAST. She holds an MBA from INSEAD in France and hasextensive scientific background in algorithm design and optimization,with a Ph.D. and Prize Fellowship fromOxford University.

KMWorld

1. Executive sponsor—budgets andoverall success of project.

2. Business owner—representation ofbusiness/user community and cleardefinition and communication of theirrequirements to the project.

3. Program/project manager—all projectplanning, resources, communicationsand deliverables.

4. Information architect—content planningand management (e.g. meta tags andtaxonomies).

5. User interface engineers—design,development and integration of searchfront end with existing applications.

6. Hardware engineer—all hardwareand O/S installs, in addition to DNS,DB or other software.

7. Network engineer—network config-urations as required for the imple-mentation.

8. Operations—daily operations ofsearch solution, including all first-linesupport.

Hadley Reynolds

Silvija Seres

Typical Roles in a Search Project

stakeholders and end-users. Follow a commu-nication plan that keeps all parties includedand responding. “You are going to be takingpeople outside of their comfort zone and theyneed to understand what you are doing. Youcan never over-communicate with a projectlike this,” said McBride.

2. Recruit the right people for the proj-ect. Identify and recruit the right people withthe right skill sets from the start, making suretheir skill sets align with the assigned tasksand responsibilities. Maintain a consistentvision throughout the implementation project.A knowledge manager at a leading mobilecommunications company suggests that prac-titioners ask themselves “whether the peoplewho defined the specifications have beeninvolved all the way through the implementa-tion. Sometimes as people change in a project,it is easy to lose sight of what you originallythought versus what the new folks thought asthey were introduced into the project.”

3. Solicit end-user input in the solutiondesign. Involve end users in content identifi-cation, design and testing. Consider involvingdifferent representatives from differentdepartments. “At Pitney Bowes, we created arole called ‘content ambassadors,’ whichincluded representatives from differentdepartments within the contact centers. Acontent ambassador’s job was to help identi-fy sources of content; validate, enter and testcontent; help design the user experience; andprovide feedback on which informationshould be made public for employees andcustomers. Without a doubt, their contribu-tions led to greater user adoption,” said PitneyBowes’ McBride.

4. Encourage user adoption with incen-tives. Employees may be reluctant to moveout of their comfort zones or embrace newprocesses. Help employees understand thedesired results and how they will be meas-ured. Then, develop ways to recognize andreward individuals who adapt to the changesin the system to encourage its use.

“Rewarding and recognizing employeeefforts to share knowledge is a powerfulway to encourage this activity. Considerperiodic bonuses to individuals with exem-plary efforts, peer recognition combinedwith cash, and project work available toindividuals who actively share knowledgeas three incentive strategies,” suggestsLadd Bodem, principal and co-founder ofmarket research firm ServiceXRG.

5. Identify content gaps and dupli-cates, and scrub your content early.Eliminate redundancy and pieces of contentthat overlap. Although content scrubbing isan ongoing process, more scrubbing in theearly phases of the KM project can helpmake content more usable and help improveuser adoption since users are able to quicklylocate the right content and do not have togo through duplicate content.

“Our content had always been organizedby department,” commented McBride. “Forthe first time, a centralized knowledgebaseprovided visibility to content overlapsbetween teams. Additional time to test andscrub responses that overlapped would havebeen beneficial in reducing the amount ofredundant or conflicting content.”

6. Define ROI measurement andreporting requirements early in theprocess. McBride emphasizes, “Make sureyou understand exactly what you are goingto look for to get to your ROI and definethose reporting requirements as early as youpossibly can.” Once live with the new systemand process, expect business stakeholders toask for reports comparing performance toexpected ROI from the project.

According to a recently published reportby ServiceXRG titled “KnowledgeManagement—Strategies. Benchmarks andBest Practice,” “there are two distinct types ofknowledge management measurements; onetype is focused on measuring the efficiencyof the KM processes such as content cover-age and quality content, while the other looksat the impact from knowledge management.Some impact metrics to consider include:measuring deflection, staffing, change in firstcontact resolution rates, resolution times,agent productivity and overall success rate.”

7. Don’t underestimate the impact oftangential benefits. For example, JodiMcBride explains how at Pitney Bowes, “wedidn’t anticipate how, having informationavailable from a single source, was going tochange how we trained. Our focus this yearhas been on redesigning the classroom expe-rience to incorporate the use of KIP (PitneyBowes’ knowledgebase) and create moreinteractive training materials that betterengage the learner across multiple dimen-sions. This is a dramatic shift away from typ-ical lecture-based training programs. Thisshift in the classroom is not something weanticipated when we went live, but has beena positive experience nonetheless.”

8. Know the needs of your end users.Understand how language componentsimpact search accuracy. Identify user searchbehaviors and consider search rules toimprove the user’s experience. “Understandhow people use the resources they have available today. Understand the slang for keyconcepts, and how people from differentdepartments may approach the same infor-mation but from different perspectives. These

Implementing KM Practitioners Share Best Practices

Most customer service organizations to-day will admit that effective knowledgetransfer is the most crucial element to resolving customer problems. When donecorrectly, knowledge transfer acceleratesproblem resolution processes, fuels cus-tomer satisfaction and leads to greater or-ganizational efficiency. Organizations investin processes and technologies that enablethem to create, manage and publish knowl-edge, and that allow them to find, retrieveand share the enterprise’s knowledge acrossall support channels. Without formal knowl-edge management processes, companieswould be unable to share knowledge withtheir customers, partners and employees.

When organizations consider investing inan enterprise-level knowledge management(KM) initiative, they generally conduct adetailed analysis of how to effectively tacklea project of such magnitude. Companieswant to know how to best prepare for suchprojects, what pitfalls to avoid, and how tomeasure results that are generated from theirKM implementation. They scrutinize internalshort-term and long-term needs, organiza-tional objectives and resources, customerneeds, system requirements and technologyimplications. Companies conduct knowledgeassessments, system assessments and devel-op a roadmap by which to guide the initia-tive. They consider the role of and impact onstakeholders, end users, partners, internalteams, customers and outside vendors. Theytry to anticipate and quantify the impact ofthe knowledge management project in theirbusiness processes, operations and ROI.

No amount of analysis and planning,however, will uncover every potential chal-lenge or roadblock. Here, practitionerswho have been through successful large-scale knowledge management initiativesshare their insights and lessons learned.

Best Practices1. Over-communicate. “You can never

underestimate the change managementrequirements that go along with projects likethese,” according to Jodi McBride, director ofknowledge and training services at PitneyBowes. “You need a shared vision, supportedand driven throughout the company from the highest levels of the organization.”Communicate regularly throughout all the phases of the project with business


By InQuira


observations will provide excellent input forcreating the first round of search rules foraccuracy and efficiency,” said McBride.

9. Keep the end goal clearly in mind.Periodically evaluate how well you areadhering to your original specifications. Itis important to make sure that decisions on

whether to expand the effort or to maintainthe effort are based on your business needsand the timing you are trying to achieve.

10. Consider the impact on existingbusiness processes. Identify and under-stand all the existing processes that will nolonger work with the new system so that

these can be addressed. A knowledge man-ager of a leading mobile communicationscompany notes that, “you need to under-stand the processes that will change andwhat is going to break or be different asyou go into the implementation phase ofyour investigation.” ❚

KMWorld

The InQuira 8 suite of business applicationsimproves companies’ Web self-service andassisted-service resolution processes. Withapplications for knowledge management, Webself-service, agent-assisted service, emaildeflection and customer experience management and analysis, InQuira 8 deliversa complete solution for resolving customersupport issues online and in the contact center—applications that can infer the user’sneeds and intentions and provide the rightinformation, tools, recommendations andassistance in an efficient, orchestrated interaction that will resolve more issues anddeliver greater satisfaction and higher ROI.

Knowledge management. Many cus-tomer support issues require research toresolve, either by the agent in the contact cen-ter, or by the customer in the self-service chan-nel. A core function of support organizations isto capture enterprise knowledge to createcontent designed to resolve customer issues.InQuira 8 provides full knowledge management capabilities, including:

◆ Capturing content and content requestsfrom within the resolution process;

◆ Authoring process that removes the unnec-essary tagging and increases content reuse;

◆ Full version control, with the ability to revertback to a previous version, compare versionsor view versions side-by-side;

◆ Robust publishing workflow to ensurecontent is effectively managed throughuser-defined stages of development andpublication;

◆ Task management to ensure that eachtask is being completed by someone with the right role and skill;

◆ Configurable email notifications for alltasks in the system directing attention tocomplete work;

◆ Multi-lingual content and translation workflow to update and distribute con-tent in multiple languages;

◆ Embedded natural language search (InQuiraIntelligent Search) for more effective and useful retrieval of information from within theknowledgebase and forum content;

◆ Discussion forums for community-basedsupport;

◆ Embedded reputation models to supportperformance evaluation for knowledge engineers;

◆ Tokenization to secure content and sectionsof content for specific user audiences; and

◆ Configurable subscriptions to categoriesof content and specific content items, including newsletters and content fromspecific authors.

Web self service. Customers and compa-nies increasingly demand more sophisticatedand effective Web self-service capabilities toresolve customer support issues. InQuira 8delivers that. Key capabilities include:

◆ Unique ability to determine user intent fromsearch and navigation behavior, and use thatinsight to manage the customer experienceintent-by-intent, empowering companies todeliver a personalized resolution experiencebased on an understanding of what the customer is trying to accomplish;

◆ Packaged horizontal and verticalized Webapplications to accelerate implementation;

◆ Industry-specific “intent libraries” to mapthe resolution experience to pre-definedcustomer needs;

◆ Intelligent search and retrieval capabilitiesbased on patented natural language processing technology;

◆ Industry dictionaries to improve search effectiveness; and

◆ Diagnostic process wizards to troubleshootand resolve customer problems.

Agent-assisted service. Contact centermanagers are under constant pressure todeliver effective, loyalty-inducing customerservice at reasonable cost. InQuira 8 for agent-assisted service includes the followingproductivity-enhancing capabilities:

◆ Embedded intelligent search, retrieval andnavigation from within the agent cockpit;

◆ Integration into leading CRM packages, including Siebel and Clarify;

◆ Ability to associate retrieved content tocase resolution;

◆ Web self-service session history capturedon service requests escalated to agentsvia email;

◆ Integrated links to trigger knowledge creationworkflows from specific cases; and

◆ Collaboration.

Email deflection. When customers areunable to resolve their own problems from acompany’s website, they will often submittheir questions to the company through email.InQuira 8 for email deflection:◆ Intercepts Web-submitted emails and

uses the subject line to search and re-trieve appropriate information to resolvethe customer issue;

◆ Deflects inbound service requests when a customer clicks through the offered solutions and abandons the submissionprocess;

◆ Passes service requests directly into theERMS when an appropriate solution is notfound; and

◆ Captures Web self-service session historyand passes it with the service request to theagent in the contact center, allowing theagent to pick up where the customer left off.

Experience management and analysis.Companies recognize that customer service isan iterative business process based on continuous analysis and refinement. InQuira 8offers several capabilities to manage the customer experience and measure the effectiveness of the resolution process and the creation and use of knowledge assets.Capabilities include:

◆ Intent libraries for retail banking, telecom andautomotive industries that automate themapping of searches and navigation behavior to industry intents. InQuira enablestargeted and managed responses at the intent level, resulting in a 10:1 reduction inthe number of managed responses userswould have to define;

◆ The ability to guide an interaction and present additional content and offers that result in higher resolution rates; and

◆ Out-of-the-box data warehouse, star schemaand reports to analyze user interactions. In-Quira Analytics includes full ability to drill intoreports to look at relationships of data for in-depth analysis of results. The analysis thatcan be performed using InQuira analytics includes conversion analysis, case escalationand email ROI, navigation usage, content gapanalysis, content usage, content authoring,customer feedback, surveys and process wizard usage.

About InQuira

technology cost assumption was found tobe off, and the original business justifica-tion was now ambiguous. The team beganto build a new business case with a viableROI but a different business sponsor wasnow required. The critical business ownerdeclined to participate and the project waseventually cancelled. If a business case hadbeen done before starting the project, thecorrect audience, business owner(s) andtechnology could have been anticipated.

3. Deploy KM Enterprisewide.Knowledge in most situations is contextual.Implementing an approach across the entireenterprise poses significant risk of failure.According to research by Robert FrancisGroup, up to 85% of all KM initiatives fail toachieve their business objectives. For exam-ple, a large energy company’s IT group wasasked to identify a technology solution forcollaboration, in both teams and large groups,as part of the KM effort. Several technologieswere evaluated and one was selected for the50,000 enterprise users. Although the tech-nology was best-of-breed, the standaloneapplication was deployed to the entire enter-prise. Adoption was slow. The implementa-tion team did not answer the fundamentaluser question: “What’s in this for me?” Theresult was a costly re-launch. This time, how-ever, it was done on a project basis aroundcore business groups with critical business

challenges. In addition, the stand-alone appli-cation with other project-specific applica-tions was surfaced on a portal platform. Useradoption rate more than doubled.

4. Assume knowledge is only in docu-ments and data. If “content is king,” thenthe king is dead. Most companies aredrowning in content and most of it is irrel-evant. The most valuable knowledge in anorganization is “implicit knowledge.” Aspart of a KM initiative, a VP of an engi-neering firm launched a project to updateand maintain engineering standards.Believing that this explicit knowledge wasthe key to knowledge longevity and trans-ferability, the firm bought and installed acontent management system. The systemsuffered from low adoption. Further explo-ration revealed that the less-experiencedengineers had trouble applying the infor-mation to their specific problems, and themore experienced engineers only rarelyneeded it. A better solution would be toprovide a continual method of capturingreal-time knowledge from threaded discus-sions or instant messages, and indexing thesalient points for quick retrieval and evalu-ation. Implicit knowledge would then be a

result of the collaboration and made explicitand usable in context.

5. Not integrating unstructured con-tent with the structured data. Since mostcompanies have terabytes of unstructuredcontent, they rely on search engines toindex the documents and users to find theinformation by keywords. Unfortunately,the user is often unaware of additionalstructured data that relates to this unstruc-tured content. Most documents do not con-tain the unique identifier that unlocks thedoor to the databases, applications and sys-tems that contain a wealth of other relevantcontent. Most deployments do not take intoaccount the relationships between the dataand applications. Portals can help bridgethis gap more effectively by linking disparate data and content sources throughan interface to seamlessly present theaggregated results.

Knowledge Management through Portals

Mistakes to Avoid andPrinciples for Success

Many organizations today still struggle toget value from their knowledge manage-ment (KM) efforts. Even companies thathave been benchmarked as the “best prac-tice” can easily fall from the top. So, howcan you get sustained value from KM ini-tiatives? There are several key factors forvaluable KM deployments. But first, let’slearn from what has not worked. Here arethe eight most common KM mistakes.

KM Mistakes to Avoid1. Build something and hope that

they’ll come. This could be said for almostany technology project, but it’s particular-ly common in KM. Why? Management islooking for the quick-fix, a silver-bullettechnology. It’s faster and easier not to getthe users involved. A global mid-sizedcompany embarked on a project to help theorganization efficiently match expertisewith their customer support needs. TheirIT group built an expert locator system thatwas implemented using a traditionalchange management approach. In a projectstatus report to management, three monthslater, they found out only 72 out of the2,500 target users had completed their pro-files. Six months of desk-pounding bymanagement got the total up to over 200users. Subject matter experts claimed itprovided no value, and it took too muchtime to update. It was also not tied to anyother data sources that could be leveragedto make the users’ jobs easier and to helpthem keep the information updated. It wasnot until a portal was implemented that thesystem began to reach critical mass.

2. Implement technology tools withouta business owner or a specific businessproblem. A Fortune 100 company felt ithad a reasonable understanding of thebusiness problem. A project team wascommissioned to produce a new enterpriseframework and a pilot for a community ofpractice. In the design phase, the team real-ized that the scope and the expected targetusers were off mark. So they adjusted the design. One month later, the original


By Brandon Lackey, Global Solutions Director, BEA and Michael Behounek, Managing Partner, Emerja LLC

“If a business case had been done

before starting, the correct audience, business

owners and technology could have

been anticipated.”


6. No perceived value to contributors forsharing knowledge. Often, organizations willexpect contributors to share informationbased on benevolence or by mandate.Unfortunately, business users can be stubbornabout performing tasks that they perceive asno-value-added. A mid-size professionalservices company had identified their stan-dards and practices as a core intellectual assetand critical knowledge to the company. Theproblem was the standards were not beingmaintained by the senior consultants.Executives believed the solution was to makeit easier for the consultants to contribute bestpractices by exposing it through the portal.Unfortunately, no additional time was allot-ted and no recognition provided to the con-tributor as part of the process. Given thechoice to work on an urgent and challengingproject or update documentation, the userschose not to contribute. Although manage-ment said it was very important, their behav-ior was to reward the “fire-fighting” work,and not to recognize the captured knowledgenecessary to prevent the fire.

7. Selecting portal technology that istoo difficult to maintain and use. If adeployment gets bogged down in complexportal or application customizations, spon-sors soon lose interest and the project isbound to fail. The portal technology shouldhave integrated, yet easy-to-use, compo-nents to produce successful KM communi-ties, such as: content directory, contentmanagement, collaboration tools, searchcapabilities, personalization, applicationintegration and business process manage-ment. These components should be easilyavailable, and integrated right after the ini-tial framework is established.

8. Failing to use business measures todrive improvement. Departments within anorganization (HR, IT, etc.) often find it diffi-cult to quantify impacts from a KM project. Ifyou can’t quantify the business value, theproject is guaranteed to fail. Companies onceviewed as the “best” at KM have greatlydiminished their efforts over time or stoppedsupporting it mostly because of a lack ofquantifiable business results. In other cases,where a business value is calculated but

cannot be substantiated, the result is similar. Ifthe supposed $200+ million per year savingsresulting from KM efforts cannot be support-ed by financial results, projects are eventuallyterminated. Sustained improvement andmomentum requires an ongoing cycle offinancial measurement for KM efforts, andshould be planned for during the design stage.

KM Principles for SuccessAvoiding these eight common mistakes

can help you reduce your risk of failure inKM. But what should you focus on?Results for the organization can beachieved following these simple practices:

Solve real business problems. This maysound like a no-brainer, but in the throes ofa technology roll-out, managers may losetrack of the underlying business problemsto be solved. Clearly stating the problemshas further benefits in scoping the require-ments, managing deliverables and selectingquantifiable metrics. In addition, objectivesand benefits that are clear to users have apositive effect on the adoption process.

Secure executive commitment to KM.Executive support is critical since itinvolves the assignment of people and achange in process. Gaining long-term com-mitment from executive sponsors comesfrom solving their business challenges,delivering business value and, ultimately,gaining their trust. Keep them informedand have them play a visible leadershiprole in the launch.

Own the process within a business unitor organization. Unlike many supportfunctions, KM needs to be embedded in thebusiness unit or support function. Having it“outside” the groups delivering the valuewill only hamper widespread use. A smallKM core team within the business unit isrequired to help facilitate the solution andhelp ensure similar processes and scalabil-ity between projects.

Emphasize people and process.Unfortunately, in many cases, projects aredriven by the technology. One secret to get-ting people to participate and collaborate isto ensure they get value from doing so.

This can take many forms—streamlinebusiness processes, reduce their workloador enable them to spend time on job-relat-ed initiatives with people inside and out-side the organization.

Embrace technology as a key enabler.One of the key struggles for many compa-nies is the variety of databases, content andapplications. KM solutions deliveredthrough portal technology can provide theinitiative to tap these disparate systems anddeliver information to the user in context.When combined with content delivery,search, business process management andcollaboration tools for sharing and captur-ing implicit knowledge, portals can deliversignificant value for KM.

Integrate the process into the employ-ees’ workflow. Weave KM processes intothe normal workflows and, where possible,simplify and optimize for the particularuser. If the solution is extra work or a bolt-on, user adoption will remain low.

Capture business metrics to determinesuccess, drive improvement, and communi-cate your achievement. Too often, the onlycaptured metrics are for portal or applicationusage. Executives and users want to seevalue that relates to the business proposition,which can involve tools like business activi-ty monitoring and analytics. For example, aportal community of practice to supporttechnician repairs would track and seeimprovement in failed repairs, average repairtimes and repairs per technician with highportal site usage. Having limited or no busi-ness metrics weakens justification during thebudget cycle or when a change in manage-ment requires a review of the project value.

We learn as much from our mistakes asfrom our successes. The challenge is totake what others have learned, to avoid thesame mistakes and to produce a consistent,repeatable KM program that delivers significant business value. ❚

Brandon Lackey is global solutions director at BEA. Heassists organizations in aligning busi-ness objectives with their enterpriseportal strategies. Before joining BEA,Lackey led the development ofHalliburton’s worldwide portal thatgrew to serve more that 6,000 clientsand 20,000 employees. DuringLackey’s tenure, that portal receivedawards from Network World,the DM Review and InfoWorld. Contact:[email protected].

Michael Behounek is Emerja LLC’s man-aging partner. Before forming Emerja,Behounek had 23 years in the energyindustry. In 2001, he led the develop-ment of Halliburton’s highly successfulKM effort that resulted in a 563% returnover a three-year period. He continuesto help companies deliver breakthroughperformance by using collaboration to solve complex business problems.

Contact: [email protected].

KMWorld

“Executive support is critical

since it involves the assignment of people and

a change in process.”

Brandon Lackey

Michael Behounek

lifecycle changes—a name, a feature, acolor or price, for example—then search-ing and finding all the information acrossthe enterprise related to that product can bea chore, if not impossible. Business seman-tics management provides organizationswith a dynamic registry to model, govern,publish and collaborate around a consensusof terms and term relationships that definethe products, services and organizationalknowledge of the enterprise.

Business semantics management isemerging as a core component of service-oriented architecture (SOA), which distributesapplications that perform services on demand.For many experts, SOA is the future of software, where applications, used internallyor externally, are delivered as services.Companies such as SchemaLogic ofKirkland, WA, have developed a businesssemantics management solution that modelscomplex business semantics relationships and provides those models to subscribing sys-tems as a service within the context of anSOA. Referred to as “semantics-as-a-service,”the solution provides a single source for creating, managing and distributing corporate semantics.

Semantics-as-a-service can be deliveredas a real-time service to enable organizationsto instantly update their semantics models as changes happen, or manage their

semantics with regular and consistentupdates. Semantics-as-a-service can bedeployed behind the firewall or as a Web-based application. As enterprises movetowards SOA and Web services environ-ments, semantics-as-a-service has emergedas a business-critical solution that can help toenhance knowledge management. The evolu-tion of semantics management from a manu-al time-consuming, cumbersome process toreal-time through semantics-as-a-service is arevolutionary concept that is changing theway companies are managing their informa-tion assets and their businesses.

Increasingly, companies are finding thatdeveloping a truly effective enterprise lexi-con requires the participation of all stake-holders in a company, from the purchasingmanager to the CTO. Semantics-as-a-service allows organizations to collaborateand leverage the knowledge and expertise ofa variety of sources, including subject-matterexperts, business users and domain expertsto develop a semantics model that describesthe corporate lexicon, provides up-to-theminute know-how and makes knowledgeaccessible across the organization.SchemaLogic’s centrally managed solu-tion and Web-based collaboration servicehelps to develop corporate semantics modelsthat can be used enterprisewide to describetheir business, products, services and overallcorporate knowledge.

Knowledge is PowerBest Practices in Semantics Management

Knowledge is power. Though said in the16th century by author and philosopher, SirFrancis Bacon, the meaning of the phrase iseven more applicable today as companiesaround the world are dealing with an overwhelming amount of data stored on com-puters, servers, flash drives, PDAs and otherstorage devices scattered throughout the enterprise. The information, email messages,documents, spreadsheets, presentations,graphs, photographs and other content trappedin information silos across the enterprise, con-tain a significant amount of knowledge thatcan benefit the organization. Unfortunately formany companies, much of this information remains untapped, unsearchable and trappedin disjointed systems.

Enterprises have deployed a complexweb of information management systems,applications, operating systems and data-bases, to manage and deploy data and content. However, the resulting informa-tion-access systems are unable to exchangeinformation in an efficient, dynamic fashion because they lack a commonunderstanding of the business semanticsthat describe the information contained inthe various content silos. Marketing mayhave one way to describe a new product,while IT, shipping, public relations andresearch and development may each haveothers. When something in the product


Jeff Dirks has morethan 20 years ofsuccess in bothearly-stagecompanies andestablished marketleaders. Dirks bringsan extensivebackground inoperationalexecution, spanningstrategy, enterprisesales and services,

distributed enterprise-system softwareengineering and financial operations.

From 1998 to 2003, Dirks served in variousexecutive capacities at CapitalStream, includingpresident and chief operating officer, where hearchitected the business and technical strategy that sparked company growth andmarket-share leadership.

Prior to CapitalStream, Dirks was vice presidentof product delivery at HK Systems, a leadingsupply chain management software company.From 1992 to 1998, Dirks rapidly advanced atFourGen software, progressing fromdevelopment team lead, group productmanager and director of consulting beforeserving as vice president of engineering forGeneral Electric Supply’s worldwide supply chainmanagement upgrade.

Jeff DirksBy Jeff Dirks, President and Chief Executive Officer, SchemaLogic


With semantics-as-a-service, organiza-tions can:◆ Build a common framework to facilitate

collaboration and governance for all usesacross the enterprise;

◆ Control the power of business lexicon/information;

◆ Unlock the corporate assets of brands,products, services, human capital andknow-how for competitive advantage;

◆ Leverage communities of participation tomake intelligent business decisions basedon improved data quality, collaborationand business agility; and

◆ Manage the corporate memory of businessdecisions for compliance.

IBM: Managing Human ResourcesIBM is well known for its worldwide

team of 190,000 employees spreadthroughout more than 160 countries. Asone of the world’s largest organizations,IBM is also one of the richest sources ofdata and human capital. With thousands ofemployees spread across scores of coun-tries, the ability to connect the company’shuman resources with the specific and relevant job listing data is vital to helpingIBM and its service contract outsourcingoperate more efficiently.

With no easy way to catalog employeesand sort them into a glossary of technicalskills for easy access, project managers had adifficult time finding the required obscure orunique skills within IBM’s talent pool, andas a result, IBM recognized an over-relianceon contractors, which were creating marginproblems for service contract budgets. Thecompany initiated a worldwide collaborationeffort for its enterprise taxonomy in whichthe semantics that describe employees’skills, roles and categories are described in a corporate-wide semantics model. Theresulting expertise taxonomy is published toIBM’s corporate expertise portal to enableorganizations across IBM to locate and manage corporate expertise as an efficientand dynamic process.

As a result of initiating this effort to bet-ter manage the information about its consult-ants and employees, IBM was able to reduceits dependence on outside contractors by 7%.The company now has worldwide collabora-tive management for its global enterprisetaxonomy, making it easier to find the rightperson or skill set for specific assignments.The company also increased the utilizationrate for consultants, and improved the prof-itability of service engagements. IBM’sOpportunity Marketplace, a Web-based serv-ice, which employs the enterprise taxonomyand features a searchable database of itsglobal talent pool, saves IBM an estimated$680,000 a year.

Kellysearch: Managing SearchWhile IBM focused on using business

semantics management to manage its collective knowledge about its humanresources, Kellysearch, the largest busi-ness-to-business (B2B) search portal inEurope, was faced with a growing taxono-my that its infrastructure could not adequately manage.

The company’s growing taxonomy includ-ed more than 200,000 phrases and 400,000sub-phrases that were attached to more than 2million businesses. Unfortunately, the compa-ny’s current infrastructure and manual taxono-my processes were not capable of scaling tomanage the growing amounts of data, whichresulted in an increase in time-to-market fordeveloping and publishing advertiser informa-tion to the Web.

Kellysearch, a division of ReedElsevier, needed a knowledge managementsolution that could drive queue time down and improve the speed and quality of search results in order to increase market share and the revenue stream of itsonline model. Kellysearch turned to thesemantics-as-a-service model to linksearch with content to drive revenues andmove the classical advertising print busi-ness to the paid search market on theInternet.

Kellysearch now has the capability toestablish and build corporate semantics by initially harvesting data from Kellysearch’smultiple company and product listings. Theinformation is then rationalized so that it isestablished, evolved and distributed acrossKellysearch. The variances in naming terminology among companies are quicklyresolved to a common set of terminology,making it easier to match global B2B queries for specific goods and services. Byimplementing a business semantics solution,embedded knowledge is monetized via an improved search and ad-revenue model.Kellysearch now has the ability to quickly expand internationally to acceleraterevenue; business processes have been streamlined and content contributors empow-ered; and enhanced search results haveincreased the quantity and quality of leads topaid advertisers.

In both the IBM and Kellysearch exam-ples, information that has typically been difficult to manage or out of reach, is nowavailable to everyone, driven by semantics-as-a-service. Delivered through an SOAframework, semantics-as-a-service providesorganizations with the flexibility and scala-bility to meet the demands of their growingbusinesses, manage the contributions fromall areas of the enterprise and increase thespeed of change.

Knowledge can be a variety of things todifferent companies. For some, it’s themagic sauce or company trade secrets; forothers it is the intellectual property foundin approved patents. Today’s enterprisesneed a common semantics model to man-age information across the enterprise.Semantics-as-a-service is emerging as abusiness-critical application, and for com-panies like IBM and Kellysearch, hasbecome a mission-critical application fortheir services-oriented architectures.

More than 400 years from its origin, thephrase “knowledge is power” couldn’t betruer. The ability to unlock the corporateassets of human capital and products forcompetitive advantage; to make businessdecisions based on improved data quality,collaboration and agility; and to better manage information and knowledge for thebenefit of the entire enterprise is true power. ❚

SchemaLogic is a leading provider of business semanticsmanagement (BSM) solutions. BSM provides a frame-work that enables companies to model the structures andrelationships of business semantics that define corporateknowledge and content. SchemaLogic facilitates dynamicchanges to the business semantic models through a Web-based governance and collaboration process thatenables participation across organizational, corporate and industry boundaries.

SchemaLogic enables enterprises to increase competitiveadvantage and reduce operational costs through betterinformation management, increased data quality andmore agile, intelligent business decisions. SchemaLogichas licensed its solution to some of the best known andlargest companies in the world.

For more information about SchemaLogic, call 425-885-9695 or visit www.schemalogic.com

KMWorld

“Information-access systems

are unable to exchange information

because they lack a common understanding of

the business semantics.”

sophisticated search and retrieval method-ologies that guide users through the issue resolution process so first-time self-service users and highly experi-enced agents can quickly find the right information including:

◆ Case-based reasoning combines NLQwith clarifying questions. The user entersa text string that yields a solution set anda series of targeted questions to furthernarrow the problem definition. Based onthe user’s answers, case-based reasoningnarrows and reorders the solutions in order of relevance.

◆ Decision trees guide users through struc-tured diagnostic scripts. Each time theuser answers a question, the decision treedynamically presents new questions andnarrows down the number of possible solutions until the most appropriate solution is identified.

◆ Expert modeling ranks potential solutionsin order of relevance to the problem as determined by subject matter experts. Expertmodels define precise relationships betweenproblems, causes and solutions.

Using this wide range of search andretrieval capabilities, enterprises can offermultiple levels of guidance and allow usersto select those techniques that best matchtheir skills and preferences. However, thesearch methodologies should be blendedseamlessly so that users do not have tomake a conscious decision about whichmethod to use. For example, users canbegin a session with a text string search,and then be presented with a series of ques-tions that combine aspects of case-basedreasoning, decision trees and expert model-ing to quickly refine the general descrip-tion into a specific problem description.KANA customer John Harrigan ofSiemens states, “What attracted us was thatthese are the kind of tools that you can get

the average user up to speed on withoutrequiring much background knowledge ofthe actual system.”

Reporting and tracking capabilitiesshould complement these methodologies todynamically score and rank potential solu-tions by popularity based on users’ experi-ences. Users provide feedback with eachsearch on the helpfulness and accuracy of thesolution, which is incorporated into futuresearch results so that solutions are scoredhigher or lower on subsequent similarqueries. Paul Kinsella, VP worldwide cus-tomer response at Creative agrees, stating,“with the knowledgebase feedback mecha-nism, we can update and revise content toreflect customer choices and preferences.”

As the management of and access toknowledge grows in importance, so doesthe ability to deploy a comprehensive, yetmaintainable knowledge managementsolution that increases customer accept-ance of self-service, enables call centeragents to answer inquiries more quickly,accurately and consistently and enhancesthe value and use of information stored inenterprise systems. By providing multiplesearch methodologies, enterprises canempower users of all levels to diagnose andresolve problems with ease. ❚

Anand Chopra brings 10 years of sales, marketing andconsulting experience to KANA. As director of productmarketing, Chopra is responsible for driving effectivemarketing programs and creating strategies for advancing the KANA eService Solutions in the marketplace. Prior to joining KANA in 2003, he heldinstrumental positions at leading companies such asOracle, Commerce One and Ernst & Young, LLP.

Answering Customers’Questions the Intelligent WayThe Importance of Multiple Search Methodologies

Enterprises face a difficult challenge whenit comes to simultaneously improving thequality of customer service and reducingservice costs. More products, growing prod-uct complexity and rapid change substan-tially increase the amounts of informationrequired to answer customer questions andtroubleshoot problems. Paradoxically, thisgrowth of information availability increasesthe difficulty of finding relevant solutions.

For enterprises to improve self-serviceadoption rates, increase call center effi-ciency and improve response accuracy,they need solutions that help agents, cus-tomers, partners and suppliers findanswers more efficiently. The traditionalmethods of search and retrieval use key-word, simple text and Natural LanguageQuery (NLQ). In many cases, search-basedknowledge management solutions empha-size their ability to sift through multipleenterprise systems to present results. Thistypically generates long hit lists with manyirrelevant entries. Another failing of this approach is that results are presentedindiscriminately. The user does not know ifthe information is accurate or current,increasing the possibility of an incorrect or out-of-date answer. Unfortunately, thesemethods are best suited to expert users whoare familiar with the content and terminologyand know which words will most quicklyyield a correct answer. Novice users withoutdomain expertise cannot easily apply the ter-minology precision these techniques requireand, frequently, people need guidance to findthe answers to a question.

Call center agents, customers and partners can all benefit from knowledgemanagement solutions that organize andstructure access to information, with intelligent guidance that matches eachuser’s level of sophistication and skill. Inaddition to keyword, text and NLQ search,a knowledgebase should deliver a set of


With more than 30years of experiencein management ofenterprise softwarecompanies, MichaelFields hasspearheadedsuccessful salesorganizations at anumber of largecorporations,including OracleU.S.A, where heserved as president,

and Applied Data Research and BurroughsCorporation. In addition, he was a founder,chairman of the board of directors and chiefexecutive officer of OpenVision Technologies,Inc., which was acquired by Veritas in 1997.Currently, Fields serves on the board of directorsof Imation Corporation and ViaNovus. He hasalso served on the advisory board of the FordMotor Company Customer Service Divisionfrom 1999 through 2001.

Michael FieldsChairman and CEO

By Anand Chopra, Director, Product Marketing, KANA


most enterprise architects are including acentral structured repository as part of theirinformation architecture. ECM systems,built on solid data-storage solutions, arethe platforms that facilitate these soundinformation management policies.

At the heart of these information sys-tems, is metadata—data stored about thedata you store. By monitoring, storing andindexing specific information about yourbusiness content, ECM vendors allow theircustomers the ability to easily find anypiece of information, and its relevant busi-ness context, quickly and efficiently. Thesesystems are built on information manage-ment policy and principles that have beenaround for a long time.

So, if your organization has a sound ECMpolicy and system in place, it’s not likely to

fall off the coffee table because of poor quality information. The next generation ofenterprise systems will focus on how to man-age the analysis of that information base tosupport your decision-making process.

The DIKW Model is an informationhierarchy that’s frequently cited when try-ing to address this problem. The model wasoriginally recorded in a 1932 poem calledThe Rock from TS Eliot:

Where is the life we have lost in living?Where is the wisdom we have lost in

knowledge? Where is the knowledge we have lost in

information?In the modern, slightly less poignant

implementation of the DIKW model, wefind four layers:

Data;Information;Knowledge; andWisdom.

Nowadays, thanks to advances in datastorage, the science of information manage-ment and the implementation of these systemsin ECM products, the transition from data toinformation is largely a solved problem.

Getting from information to knowledge ismuch more difficult. Knowledge includes the“how” aspect of a problem. Returning to ourrobot, it’s the analysis of the information thattells it “how” to proceed.

Current efforts at solving this problem arevaried, and you'll probably recognize them asthe more modern features provided by ECMvendors: collaboration, which allows peopleto discuss and share information in order tofacilitate progress; and workflow, where pre-scriptive, best-practice knowledge as definedby a business process analyst provides “how”information. Content management tools,like blogs and wikis, all provide additional published content around a topic—morepublished analysis to help people decidewhich step to take next.

Tools like these are striving to bridgethe conceptual void between informationand knowledge. While the jury is still outon how effective they are, the challengeis considerable. The next time you needto evaluate a system for inclusion in yourenterprise architecture, consider howwell it bridges this gap. Think like arobot. Do I have the right informationavailable? Will this system enable me tomake better decisions? Without a carefulapproach to both aspects, you could endup on the carpet. ❚

TOWER Software, a leading enterprise content manage-ment (ECM) provider to government and regulatedindustries, delivers award-winning information manage-ment solutions. Our product, TRIM Context 6, enablesorganizations to have compliant, secure and accurateinformation available to make confident business deci-sions. TRIM Context 6 won AIIM E-DOC Magazine 2006Best of Show award for ECM suites.

What a Robot Really Wants

Knowledge at the Pointof Decision

A small, two-legged robot stands atop aglass-topped coffee table. On its two dimen-sional world, it has to contend with a pottedplant, an old TV Guide and several coffee cups,along with the ever-present danger of plum-meting over the side onto the carpet below. Asthe robot navigates its way around, a constantevaluation process occurs inside its software“brain.” First, information is collected throughits sensors. Secondly, that information is ana-lyzed, using a decision tree to determine theoptimal course of action.

What do the adventures of this robot haveto do with knowledge management? It’s moreimportant than you might think. You see, rightbefore the robot takes its next step—once thissimple two-stage process is completed—therobot could be said to “know” something. It’scollected all available information and

analyzed it. Knowledge is created throughanalysis of information.

So the effectiveness of our robotfriend—or if you like, how “smart” it is—depends directly on two things: the accura-cy and relevance of the information sup-plied; and the effectiveness of the evalua-tion process. Poor information, throughfaulty sensors or too few sensors, willresult in an inaccurate picture being fed tothe decision-making processes. Pooranalysis will lead to bad decisions, regard-less of the quality of information supplied.

Now I’m sure you saw this analogycoming, but face it—your enterprise isexactly the same. To create a smart enter-prise, you need to have a stable, reliableinformation base and the analysis tools thatallow you to create valuable knowledge—knowledge that fosters good decisions.

Information management has beenrefined over the years, to the point where

KMWorld

Gordon Taylor is asenior technicalconsultant forTOWER SoftwareNorth America,makers of TRIMContext. For thepast 10 years, he’sbeen working onsolving enterpriseinformationproblems across theprivate and public

sectors, as a software developer, solutionsarchitect and research and development projectmanager. Since joining TOWER five years ago,his work has ranged from developing ECMsoftware to planning and configuring large ECMrollouts for government and regulated industry.He lives in Virginia with his wife and fourchildren and no robots.

Gordon Taylor

By Gordon Taylor, Senior Technical Consultant, TOWER Software

“Poor analysis will lead to bad decisions, regardless of the quality of

information supplied.”

enterprise search has a strong emphasis on knowledge management, intellectual property, e-discovery and compliance, itbecomes the foundation for comprehensiverisk management.

Whether it is to meet regulatory compli-ance requirements from Sarbanes-Oxley(SOX), to the Health Insurance Portabilityand Accountability Act (HIPAA) or torespond faster to e-discovery requests, com-panies of all sizes today need to be able tosearch through literally gigabytes and ter-abytes of stored data wherever it resides. Inother words, companies have to find whatthey are looking for, every time, all the time—their future may depend on it.

The Role of Search in ComplianceFor the California Conservation and

Liquidation Office (CLO), a division of the

California Department of Insurance respon-sible for regulating all insurance companiesoperating in California, the compliance taskrequires managing insolvency, asset distribu-tion, claimant lawsuits and disbursements.The office also has to routinely determine ifan insurance company can be rehabilitatedand continue its operations. This requires theCLO to keep extensive forms, policies andchange control documentation on an intranetavailable to both employees and state auditors. And to ensure compliance, it mustkeep all processes and change controls current, and make sure each step in any procedure is well-documented and imple-mented without deviation.

According to CLO’s Mohammed Mojabi,the IT networks operations manager, “As ahighly regulated entity, we are in a constantstate of audit so we need a search engine thatis highly secure, highly accurate, yet easy touse for both users and auditors.”

With CES, users and auditors quicklyfind current forms and documented proce-dures, reducing the time it takes to find andact on information by up to 35%. In addition to the time and money saved, CESminimizes compliance violations andresults in far more efficient audits. ❚

Find information. Understand Information. Act—muchfaster. Based on industry standard ASP and .NET technologies, and winner of the 2006 Microsoft PartnerRegional Winning Customer Award, Coveo EnterpriseSearch delivers the best value in the marketplace without-of-the-box document level security, unparalleledaccuracy, consumer style ease of use, and an implementation cycle of less than 24 hours. Whether it’sto meet regulatory compliance, improve customerresponse, protect intellectual property or improve organizational efficiencies up to 35%, Coveo EnterpriseSearch enables organizations to find, understand andtake action on critical information located anywhere in the enterprise.

Enterprise Search: The Foundation for Risk Management

If you can’t find it, you can’t manage it. Unfortunately, many executives have dis-covered this too late in the game—whenthey face regulatory compliance penalties,legal troubles or a corporate scandal. Today,there is no question that managing securityrisks and complying with government reg-ulations related to information security canbe a daunting task for any organization. Infact, many enterprises now have a mandateto implement technologies and applicationsthat ensure all employees, not just corporatemanagement, can access appropriate information securely.

Having a comprehensive, highly secureenterprise search capability—one that fillsthe gap between specialized search sys-tems and Web-focused search tools—canbe a key business asset, and is essential toeffective knowledge management for cor-porations and government entities. When


Laurent Simoneauhas more than adecade ofexperiencedeveloping andbringing to marketinnovative searchproducts. He comesto Coveo fromCopernic, where hewas the chiefoperating officer andresponsible for

orchestrating the company’s enterprise productsdivision spin-off into Coveo. Prior to being COO,Laurent served as CTO of Copernic for six years,during which he guided product strategy anddirected the research and development of core technologies.

Laurent Simoneau

By Laurent Simoneau, CEO, Coveo Solutions, Inc.

In today’s litigious world, compa-nies must respond not just quickly, butquickly enough to litigation requestsfor electronic evidence...or face heavycourt-imposed sanctions. The problem?e-discovery is literally like finding aneedle in a haystack.

ProSearch Strategies, a discovery,analytics and workplace tools companyfocused on the legal market in areas oflitigation, due diligence and compliance,faces these issues every day. During itsdue-diligence process, it follows a highlyaccurate automated process to minimizethe manual review of documents—whichis time-consuming, error-prone andexpensive. For a recent case, ProSearchhad to search through some 27terabytes of unstructured data and need-ed an accurate technology solution thatwould enable it to hone in on the rightdata, quickly. ProSearch Strategies settled

on a combination of applications includ-ing Microsoft SharePoint, SQL BusinessAnalytics and Coveo Enterprise Search(CES) for back-end processing on thedata. CES is the foundation applicationthat allows ProSearch Strategies toaccess volumes of information in a manageable, structured format.

“We’re using software-as-a-serviceand SharePoint as the secure portalstructure on the front-end, and SQLbusiness analytics and Coveo for back-end processing on the data. Thissolution has enabled analysts to consolidate and eliminate time-consuming analytic tasks, decreasingprocessing time by 30%, and allowedresearcher-reviewers to make moredocument decisions per hour, whilemaintaining the ease of use for casualsearchers,” says Trevor Allen, CIO ofProSearch Strategies.

The Role of Search in Litigation


assigned cost that can be readily determinedfor your business. And, typically, translationcosts are rising—not because the per-pagerates are increasing, but because more andmore content needs to be translated. Product lifecycles grow shorter. Many morevariations of products are introduced. Newproducts are added to those already there.The increasing amount of content requiredto support these products means thatmore must be translated and, inevitably,translation expenses rise.

So, in calculating the ROI on introducingcontent management, you CAN get an objec-tive metric to start with—the amount you cur-rently spend on translation. This starting pointis good for either projecting what you will saveover a given period or analyzing the results ofa pilot project designed to show what you canachieve. But still, that’s only half the story.How do you measure what your savings are,or what they could be?

Translate Once, Translate RightThe answer lies in the ability of multi-

language content management to control whatgets translated. If you have content that evolvesfrom one product release to the next, chancesare not everything in the content related to thatproduct changes. Some things do. Maybe manythings do. But not everything.

The key to controlling translation costsis to translate only those things that havechanged. Your savings come from not trans-lating those things that haven’t. Factor in thesavings in time and resources and that number grows exponentially.

But in order for this approach to work,your content must be managed in granularunits, sometimes called “minimum revisableunits.” An XML-based approach like theDarwin Information Typing Architecture(DITA) calls them “topics.” Once your con-tent is authored in such units, a content man-agement system can determine when one ofthose units changes. When that content goesthrough the review and approval cycle andbecomes ready for the next edition, a tool in

the content management system can identi-fy all the related units in other languages thatmust now be retranslated to capture the newor changed content. If the base unit is brandnew, the system can also determine the needfor creating new objects in the target lan-guages. Such a content management systemcan then create a project and send the newcontent along with the older translated ver-sion off to the translation process.

When the translated units return, they areput in the repository and managed with allthe others so that when an information setin the target language is published, it is easyto verify that all units are at the current ver-sion and in synch with the source language.

This approach to controlling translationcosts is demonstrable and in use today. Howdoes this approach provide a metric formeasuring ROI? By using it, you can read-ily determine the percentage of content thatis reused and not translated. The translationcosts that would have been expended pro-vide one good measure of your ROI. If thatalone justifies your investment cost, then theother less readily measurable benefits ofmulti-language content management comealong for free. ❚

XyEnterprise is a leader in content management andmulti-channel delivery, providing solutions to industryleaders in technology, financial services, publishing,manufacturing, government and aerospace/defense.The company continues to innovate to deliver productsthat support an evolving marketplace—including distributed workflows, multi-language content manage-ment, interactive electronic delivery and standards suchas DITA and S1000D. Their solutions help workgroups ofall sizes simplify and expedite the automated creation,management, delivery and reuse of content across the enterprise. [email protected]

Realizing MeasurableROI with Multi-LanguageContent Management

What is content management worth?

That can be a philosophical question. Ifyou ask a content author, you’ll hear aboutthe way content creation and review is madeeasier and more efficient. If you ask a proj-ect manager, you’ll learn that content man-agement makes it possible to identify bot-tlenecks and monitor the progress ofcomponent pieces in an information set. Ifyou ask a production person, you’ll discoverthat delivering to different data formats isnow a more automated and reliable process.Ask an IT person and they’ll talk about in-tegration with existing enterprise infra-structure. While all of these answers aretrue, none of them really assigns a dollarvalue to content management. Finding aconcrete way to measure ROI for contentmanagement can be a challenge.

One place to look for a metric to meas-ure the value of content management is inthe rising costs associated with deliveringproduct-related content in many differentlanguages. The global economy is here.Products that were once marketed and soldin areas that required only one or two lan-guages are now being offered worldwide.Suddenly the Pacific Rim, Eastern Europe,India, China, and Brazil are growing markets for many kinds of products andservices. To reach them and sell to them suc-cessfully, content such as marketing andsales information, product documentation,services manuals, warranty policies andmany other kinds of content must be avail-able in each of the many languages spokenthere. Translation is the key. And time-to-mar-ket is critical. The required content is alreadyin one language. It just needs to be put intothe languages needed for your new markets.

The Rising Cost of TranslationBut translation is expensive. Whether

you translate in-house or use a translationservice, whether the process is aided by a translation memory system or not, no matter how it gets done, translation has an

KMWorld

Jon Parsons hasmore than 20 yearsof experienceautomating thecreation,management anddelivery of contentin multiple forms.Currently he worksin productmarketing atXyEnterprise. Prior tothat, he was a writer,

editor, tools developer and publishing consultantfor a large computer manufacturer. Long anadvocate of generic mark-up and an enthusiastfor XML, he has served on the board of directorsof OASIS, the Organization for the Advancementof Structured Information Standards, and is afrequent speaker at industry events.

Jon Parsons

By Jon Parsons, XyEnterprise

solution, the customer pays a subscriptionfee to use licenses on the system. The vendor has to perform to the specificationsof a service level agreement (SLA), orthere are typically financial penalties topay. SaaS is gaining acceptance as an alternative business model, led by the popularity of applications like WebEx andSalesforce.com.

Any organization can benefit from anSaaS business model, regardless of its size:

◆ Small-to-midsize business: For small-to-midsize businesses, SaaS allows accessto software that might otherwise be toocostly or complex to implement or support.

◆ Enterprise: For larger organizations,SaaS allows departments to avoid havingto make large capital expenditures andhaving to pay for internal support costs.Large corporate environments typicallyturn to SaaS to support short-term projects, software that will only be usedoccasionally or by a small number of em-ployees and for applications that need tobe available outside of a firewall to part-ners, contractors, suppliers or customers.

Traditional CMS vendors typically chargemost, or all, of the purchase price at the time

a contract is signed, before the system is eveninstalled. Usually, the customer is responsiblefor the system deployment (often workingwith a consulting/integration firm). The ven-dor charges 18% to 20% annually for soft-ware support, and is not accountable forimplementation failure, even if the system isnever actually used in production.

By contrast, an SaaS vendor is respon-sible for configuring the environment anddelivering a turnkey application. Licensefees to an SaaS vendor do not start until thesystem is production-ready, and there arefinancial penalties for failure to meet themetrics in the service level agreement.

Many “traditional” CMS vendors are con-sidering (or announcing) that they will nowsupport a hosted model as an alternative deliv-ery mechanism. Most of these companies willstruggle, because they will now be held moreaccountable for a successful productionimplementation and will have a difficult timeadjusting to having to wait for payment. It willalso be very hard for these companies to give up their ongoing profitable maintenancerevenue. (For example, 45% of Oracle’s revenue comes from maintenance.)

Implementing or upgrading a contentmanagement environment is a significant andrisky undertaking, and there are many optionsavailable for consideration. But if you properly define your business needs, stickwith solutions that conform to standards, startwith a small pilot project and look for rapidROI models (such as SaaS), your chances forsuccess will increase dramatically. Choosewisely—the rewards are well worth it. ❚

DocZone.com is a privately held company headquarteredin Heemstede, the Netherlands, with a wholly owned USsubsidiary headquartered in Bedford, New Hampshire.DocZone.com has direct employees in the Netherlandsand US, as well as a close network of development and integration partners. DocZone.com is a KMWorldTrend-Setting Product of 2006

Hosted XML ContentManagement: Is It Right for You?

Publishing organizations have long rec-ognized the value of migrating content toXML to attain the benefits of content reuse,reduced localization costs and single-sourcepublishing. But, many of these organiza-tions have never been able to justify the highcost and long implementation cycles re-quired to install an in-house XML contentmanagement system (CMS).

Recently, a new alternative hasemerged—the “Software as a Service (SaaS)”model, which offers a hosted XML contentmanagement environment on a subscriptionbasis. According to research firm InfoTrends,more than 40% of their survey respondentswould either “prefer” or “definitely consider”a hosted content management solution.

Software as a Service:What is It; Why Should I Care?

The SaaS business model is essentiallydesigned to offer a full-featured solution ina hosted environment. The software appli-cation sits in a centralized, secure data center and is served up to end users com-pletely via a browser. Rather than buyingand implementing an expensive in-house


Dan Dube is themanaging directorof US operations forDocZone.com.Dube has 20 yearsof experience inbusiness process re-engineering andimplementation ofstandards-basedcontentmanagementsystems, including

three years as the founder and president ofLighthouse Solutions, an XML systemsintegration company. He led the successfuldeployment of approximately 50 XML-basedcontent management and publishing systemsaround the world in a variety of industries, anddesigned and managed the implementation of XML-based automated localization systemsthat are currently used at several Global 2000 companies.

Dan Dube

By Dan Dube, Managing Director, US Operations, DocZone.com

DocZone.com provides the first commercially available XML content management platform available exclusively with the SaaS “on-demand” businessmodel. Our customer base spans many industries, from automotive to hardware/software manufacturers to healthcare solution providers to utilities.Some examples include:

◆ A European automotive company is using DocZone to facilitate the creation, localization and automated publishing of glove box manuals in up to 30 languages, including bidirectional languages such as Arabic;

◆ A global healthcare company is implementing DocZone to manage the editorial,localization and single-source publishing of technical manuals, Web-based training materials and HTML help from the same set of source content; and

◆ A localization provider is using the DocZone platform to facilitate its translationand content optimization services to its end-user clients, allowing them to passon significant savings for translation and desktop publishing of multilingual content and making it a more competitive player in the localization industry.

Example of an SaaS CMS: DocZone.com


typically other entities or noun phrases.Look again at our previous example.

he[Ali Ghufron] was the operationalchief of the organization

An anaphor reference capability canenhance a text extraction tool. Relating ref-erents to their anaphors provides additionalinformation when discovering relation-ships, facts and events. Now consider theprevious example when one knows the referent to the organization.

he[Ali Ghufron] was the operationalchief of the organization [JemaahIslamiah]

A relationship that Ali Ghufron is theoperational chief of the Jemaah Islamiah isnow apparent. But is there any more infor-mation we can glean from this example?

Inferable AttributesInferable attributes are those things

about an entity that can be determined fromother bits of the document or are implicit inthe entity itself. For example, the gender ofa name like John Smith can be inferred tobe male simply because one knows that thename John tends to refer to males.However, some names are ambiguouswhen it comes to their attributes. From ourexample, the gender of the name Ali isambiguous. The name is used for bothmales and females. One could merge theattributes of all occurrences of an entitytogether to make a composite entity. Welearned that the pronoun he is linkedtogether with Ali Ghufron in our previous

example. Since he is a pronoun that refersto males, one can assume that the entity Ali Ghufron is also male. Reconsider our previous example.

he[Ali Ghufron, GENDER=Male] was the operational chief of theorganization[Jemaah Islamiah]

We now know that the male Ali Ghufronis the operational chief of the JemaahIslamiah, even though the relationship wasnever explicitly stated in our document.

Structured Language

In this article, I showed how leveragingpronominal information can provide moremeaningful results to a text extraction tool.I also demonstrated how anaphors addvalue to text extraction by making relation-ships become more apparent. Finally, Idescribed how a text extraction tool coulduse inferable attributes, either implied fromthe entity itself or by merging the attributesof all the various occurrences of the entity,to further enhance text extraction. Theseare just a few examples of how the under-lying structures of natural language can be used by text extraction tools to enhancethe value of extracted information for real-world applications. ❚

Headquartered in Herndon, VA, Lockheed Martin’sIntegrated Systems & Solutions (IS&S) was formed inJune 2003, in response to the increasing demand forsolutions that promise a comprehensive, real-time information picture for faster, better informed decisions.Developed with more than 20 years of Lockheed Martinexperience, AeroText is a high-performance data extraction engine and development environment thatworldwide companies and governments use to find andcorrelate relevant information in text documents.

Who is That “he?” Using Pronouns and Anaphors in Text Extraction

Text extraction is a powerful tool to find andcategorize elements in unstructured docu-ments. These elements, or entities, are con-nected together to form the relationships, factsand events in a document. Oftentimes, the sur-face forms of an entity are not sufficient to cap-ture and glean all of the necessary informationfor further analysis. Text extraction needs to beable to capture and link three underlying piecesof information to rightly categorize elementsin a document: pronouns, anaphors and infer-able attributes. This article will first describehow leveraging pronominal information im-proves text extraction. Second, I’ll explain howanaphors are useful to text extraction. Finally,this article will describe how inferable attrib-utes are used to further enhance text extraction.

PronounsVery often in a document, a relationship

occurs that cannot be determined solely fromits surface structure. Pronouns, such as he, sheand it, are only valuable if they can be relatedback to the entities which they refer. Considerthe following example.

he was the operational chief of theorganization

This phrase has little value for textextraction unless one can determine whothe he refers to. Once a text extraction toolcan determine pronominal reference,relationships between entities becomemore apparent. Now consider the previousexample when one knows the pronominalreferent.

he[Ali Ghufron] was the operationalchief of the organization

This gets us closer to understanding,but we still need to know who the organi-zation references.

AnaphorsAnaphors are referents to some other

entity within a document. Another type ofanaphora that is not discussed in this articleis exophora. Exophora is when the referentrequires real-world knowledge and lies out-side of the document. While pronouns men-tioned above are also anaphors, referents are

KMWorld

Gregory Robertsholds a bachelor’sdegree in historicallinguistics from theUniversity of Illinoisand a master’sdegree insociolinguistics fromGeorgetownUniversity. Over thelast 10 years, he hasbeen helping todevelop extraction

capabilities for most of the industry’s majorextraction tools. His expertise has been appliedto both government and commercial markets,both in and outside the United States. For thelast six years, Roberts has been expanding theextraction capabilities of AeroText for LockheedMartin, having played an active role in the tool’sinitial creation. He is now one of LockheedMartin’s leading experts in knowledgemanagement and data extraction.

Gregory F. RobertsBy Gregory F. Roberts, International Sales & Marketing, AeroText,

Lockheed Martin Integrated Systems and Solutions

“An anaphor

reference capability

can enhance a text

extraction tool.”

get the appropriate information from thecustomer? What is the average elapsedtime that cases are open, and what are thecut-off points for customer delight and cus-tomer pain?

3. Perform a diagnostic: We often findthat the multiple iterations required beforetroubleshooting can get started do not need

to be handled by an expert. In the vastmajority of cases, there are technologysolutions. In our experience, we can auto-mate the collections of the knowledgerequired to initiate troubleshooting. Self-service wizards can walk the customer orfront-line agent through the critical infor-mation required to initiate troubleshooting.

4. Set clear goals: Monitor the underlyingprocesses. Monitor your customers’ andagents’quantity and success with the wizards,the elapsed time cases are open and how oftenthe users (customers or agents) get the correct

log info to the engineer before they start thetroubleshooting process.

5. Develop your future state: Only forsome of the problems can wizards collectlog and configuration information.Specifying the set of applicable problemsfor which this is possible is a pre-requisiteto quantifying the benefit.

6. Quantify the benefit: Develop wordequations to specify the savings. For exam-ple, the potential savings from collectinglogs automatically is (% of time engineersspend collecting sys files) X (% of time sysfiles are needed) X (% of time the collec-tion could be completed in a wizard) X (%of time cases are submitted online). Theresult of this is an efficiency improvementfor your engineers. Similar word equationsassess the reduction in elapsed time casesare open.

7. Run a continuous improvementprocess: Using technology to solve prob-lems requires ongoing effort. Make sureyou are assessing which problems are likelyto apply to this process. Set 30- to 90- and120-day assessment periods and add/modify/delete wizards as necessary.

You can see that low FCR is a proxy fora host of underlying processes. Measuringyour process today and theorizing how yournew processes will help you close gaps thatunderlie problems can help remedy coreproblems. Continuing to improve and benchmark processes allows you to knowhow changes improve your organization.Over time incrementally improving andbenchmarking improvements will help youreach your customer satisfaction and efficiency goals. ❚

Andrew Cohen, Ph.D. is the senior director of businessconsulting at KNOVA Software. His team is responsiblefor business strategy and process re-engineering in pre-sales, implementation and post-implementationservices. Prior to joining KNOVA, Cohen ran the knowledgemanagement and distributed learning research team atIBM Research’s Cambridge lab. Cohen received a Ph.D.in cognition and learning from the University of Torontoand has a Masters of Science and BA in physics.

KNOVA Software is a leading provider of intelligent customer experience solutions that maximize the valueof every interaction throughout the customer lifecycle.Built on an adaptive search and knowledge managementplatform, KNOVA’s suite of applications helps companiesincrease revenues, reduce service costs and improvecustomer satisfaction. Industry leaders including AOL,Ford, HP, Novell, Reuters, McAfee and H&R Block rely onKNOVA’s award-winning applications to power an intelligent customer experience on their Web sites, andwithin their contact centers. For more information, visit www.knova.com.

Beyond First CallResolutionDiagnostic and Measurement Practices for KM

Knowledge management (KM) initiativesare one way of improving technical supportorganizations, driving support marginsthrough efficiency and increased customerloyalty. However, it is one thing to imple-ment different practices in KM in your organization and another to know that thepractices are contributing to your organiza-tion’s bottom line. In what follows, we outline best diagnostic and measurementpractices for judging the effects of new KMpractices.

Increasing “first contact resolution”(FCR) is often one of the first goals speci-fied in a project. Organizations oftenbelieve a significant reason their satisfac-tion is low and costs are high is that callsor cases are not solved on the first contact.They regularly point to long elapsed timesfor cases, poor satisfaction scores and lowered support margins as indicators thatthey need to improve the expertise of theirfront-line support personnel and theexplicit knowledge that supports theirefforts. As we shall see below, this is onlypartly true. Pursuing this path alone maynot fully address the FCR problem and thecorresponding cost/satisfaction issue.

Following common best practices, let’sexamine the situation:

1. Map and model current process: Forcomplex technical support centers, there aremultiple interactions or activities, oftenincluding three or more touch points with thecustomer, before any troubleshooting starts.Support engineers may require logs, configu-ration and sample files just to start theirassessments. Adding expensive experts toyour front line will only exacerbate the problem. The experts will have to go throughthe same set of iterations and activities withcustomers before they get started.

2. Establish metrics: Measuring firstcall resolution alone does not give you thewhole picture. In this case, you need tomeasure the number and content of theiterations. How much time and over howmany iterations does it take an engineer to


By Andrew Cohen, Ph.D., Sr. Director of Business Consulting, KNOVA Software

“Increasing

first contact

resolution is often

one of the first goals

specified in

a project.”


exponentially. The volume of content rises asproduct lines are extended, the velocity ofcontent accelerates as product lifecyclesshorten and the variability in informationgrows as products are customized and internationalized. The strain on the organiza-tion’s product information processes is apparent in squeezed product documentationdeadlines and missed launch dates, informa-tion inaccuracies that trigger legal liabilityand increased redundancy costs as content is updated and transformed for multiple languages and delivery formats.

Enterprise customers now identify andmeasure where value and costs are createdor diminished throughout the product infor-mation lifecycle. From original content cre-ation, to review and translation cycles, todocument assembly and delivery, there arefour specific areas identified within organi-zations where content value and cost islikely to be measured: re-use of content; col-laboration and review workflow cycles; lan-guage translation processes; and publishing.

What is clear is that the more an organi-zation can eliminate a redundant, standaloneapproach to product information processes,the better the cost-efficiencies across the entire content lifecycle. For many, the old model of writing and locking valuable

content in static documents stored on file-servers in a business group is crippling in an age when agility to respond is critical to success.

A Standards-based ApproachAstoria Software has helped leading or-

ganizations realize efficiency and cost-savingsin how they create, manage and deliver theirproduct information. Most start with the adoption of a single-source informationmodel, relying on a centrally managed content repository. This centralization of con-tent in an XML repository delivers flexibilityas an organization creates, tracks, updates,translates and publishes content and managesits re-use over time. Besides automation tools,organizations are adopting information stan-dards such as the OASIS Darwin Information Typing Architecture (DITA). This standardsupports a single-source information modelwith the creation and management of content“topics” that are easily re-used. Content topics, along with metadata and product con-tent attributes are managed independently, butmaintain relationships with each other. DITA’sgranular topic-based approach, in concert withan XML content repository, delivers inherentflexibility in the creation of documents,and supports efficiencies throughout the author-to-publish process.

As demand for new products continues toput pressure on content groups, the need formore cost-effective management of the infor-mation lifecycle becomes critical to businessefficiency. Solutions such as Astoria’s havehelped Fortune 500 companies streamline theirproduct information processes and reducecosts, from $3 million to $300,000 in docu-mentation translation costs alone for one Fortune 500 manufacturer. It starts with an organization’s capability to re-use content, toeffectively collaborate, and to automate multi-language, multi-format publishing. Finally, itmeans adopting a single-source informationmodel and content management solutions thatcan effectively manage the entire process in ameasurable and cost-controlled way. ❚

Chip Gettinger is a long-time executive in the informa-tion publishing industry and is a regular speaker at con-tent management and publishing industry conferences.As vice president of services and support for AstoriaSoftware, he advises customers on best practices inproduct information management. Gettinger is alsoinvolved with OASIS, the XML standards board, and isleading the first DITA specialization committee for med-ical manufacturing. Astoria, based in San Mateo, CA, andfounded in 1994, is a leading provider of XML contentmanagement solutions for the dynamic publishing ofproduct documentation and content.

To learn more about how Astoria has helped leadingorganizations including Siemens Medical Solutions,Texas Instruments, NCR Teradata and more, visitwww.astoriasoftware.com/kmworld, or via email [email protected].

The Hidden Costs ofProduct InformationPublishing

Manufacturing enterprises have placed asustained focus on information managementsolutions to support and augment the de-sign, development and production process.The associated product and manufacturinginformation that parallels the productionprocess is closely managed, tracked and reviewed to optimize the overall productlifecycle opportunity.

Increasingly, many organizations are track-ing and measuring the product informationchain beyond the direct manufacturing process.The findings? This extended chain is muchlonger, broader and costlier, with a decentral-ized network of product information creatorsand consumers. The critical nature of this chainis evident as most content flows directly to thecreation of customer-facing product informa-tion, from sales and marketing content, to prod-uct documentation and operations manuals, andto customer service and support information.Some estimate the current costs of content creation, management and delivery throughoutthe organization accounts for up to 6% of thecosts of goods sold for any one product.

Volume, Velocity and VariabilityAs companies add to and customize prod-

uct lines, the associated content costs grow

KMWorld

By Chip Gettinger, Vice President, Customer Service and Support, Astoria Software

A manufacturer’s technical publication departments are most affected by thevelocity of new product innovation, and the volume of demand for new contentdelivered in more formats and languages.

To address this need, Astoria Software introduces a new end-to-end XMLcontent management solution now available as an on-demand offering. This best-in-class solution delivers an XML authoring tool, an XML content managementrepository, workflow, a built-in DITA Workbench for simple transition to the DITA standard and a composition engine to output to any delivery format—print or digital. This entire solution is accessed from the Web via a third-party hosting provider,eliminating the need for any hardware or services investment, and supported andserviced by Astoria. For more information, visit Astoriasoftware.com/OnDemand

Technical Document Publishing On-Demand

November/December 2006

Produced by:

KMWorld MagazineSpecialty Publishing Group

For information on participating in the next white paper in the “Best Practices” series, contact:[email protected] or [email protected] • 207.338.9870

Kathryn Rogals Paul Rosenlund Andy Moore207-338-9870 207-338-9870 [email protected] [email protected] [email protected]

For more information on the companies who contributed to this white paper, visit their websites or contact them directly:

www.kmworld.com

Coveo Solutions Inc.120 Hawthorne Avenue, Suite 100Palo Alto CA 94301

PH: 800.635.5476FAX: 650.475.8024Contact: [email protected]: www.coveo.com

BEA Systems, Inc.500 Sansome StreetSan Francisco CA 95111

PH: 800.817.4BEA (US toll free)Contact: [email protected]: www.bea.com

Endeca Technologies, Inc.55 Cambridge ParkwayCambridge MA 02142

PH: 617.577.7999FAX: 617.577.7766Contact: [email protected]: www.endeca.com

Fast Search & Transfer, Inc.117 Kendrick Street, Suite 100Needham MA 02494

PH: 888.871.3839FAX: 781.304.2410Contact: [email protected]: www.fastsearch.com

InQuira Inc.851 Traeger Avenue, Suite 125San Bruno CA 94066

PH: 650.246.5000FAX: 650.246.5036Contact: www.inquira.com/contact.aspWeb: www.inquira.com

DocZone.com21 McAfee Farm RoadBedford NH 03110

PH: 603.488.5008FAX: 603.488.5009Contact: [email protected]: www.doczone.com

KNOVA Software10201 Torre Avenue, Suite 350Cupertino CA 95014

PH: 800.572.5748Contact: [email protected]: www.KNOVA.com

Noetix Corp.5010 148th Avenue NE, Suite 100Redmond WA 98052

PH: 866.4NOETIXFAX: 425.372.2942Contact: [email protected]: www.noetix.com

KANA Software181 Constitution DriveMenlo Park CA 94025

PH: 800.737.8738Contact: [email protected]: www.kana.com

SchemaLogic Inc.620 Kirkland Way, Suite 100Kirkland WA 98033

PH: 425.885.9695FAX: 425.883.0117Contact: [email protected]: www.schemalogic.com

XyEnterprise101 Edgewater DriveWakefield MA 01880

PH: 781.756.4400FAX: 781.756.4330Contact: [email protected]: www.xyenterprise.com

Lockheed Martin CorporationIntegrated Systems & Solutions13560 Dulles Technology Dr.Herndon VA 20171

PH: 703.466.1268Contact: [email protected]: www.aerotext.com

TOWER Software12012 Sunset Hills RoadTwo Discovery Square, Suite 510Reston VA 20190

PH: 800. 255.9914 or 703.476.4203FAX: 703.437.9006Contact: [email protected]: www.towersoft.com

Astoria Software66 Bovet Road, Suite 280San Mateo CA 94402

PH: 650.357.7477FAX: 650.357.7677Contact: [email protected]: www.astoriasoftware.com

best practices in enterprise knowledge managementhcmgt004/kmwhitepapernov06.pdf · the benefits of...

Documents