What to Read Next
Getting more value from knowledge — especially from a firm’s own hard-won knowledge — is one of the central challenges facing companies today. Many organizations have approached this problem in recent years by making big investments in IT systems, especially content repositories and databases. For example, 70% of organizations with more than 10,000 employees have more than 100 separate content repositories.1 Despite such investments, many companies have not seen much in the way of returns.
A particularly egregious example is a Fortune 50 manufacturing firm that put more than 3 million documents, previously dispersed in more than 2,000 systems, into one database. In the first three years alone, the project’s budget came to more than $22 million. Unfortunately, the system mostly created confusion, as no changes were made in the way documents were presented to employees or to help them navigate the new system. The anticipated benefits from the substantial investment went largely unrealized.
As bad as this story sounds, it is not uncommon. A research report by analysts at International Data Group’s IDC estimated that an organization employing 1,000 knowledge workers might easily incur a cost of more than $6 million per year in lost productivity as employees fail to find existing knowledge they need, waste time searching for nonexistent knowledge and recreate knowledge that is available but could not be located. Imagine the impact on an organization with 50,000 or more employees.
The challenge in reducing such waste and increasing productivity is to help employees find what they need, when they need it. The good news is that companies can easily do better than to employ the blunt tools they have used so far, and they can do so with incremental investments. A good place to start is by considering how the innovative giants of the Internet — Google, eBay and Amazon.com — have built business models around attracting and retaining customers. A big part of their success has come from their ability to make it easy for customers to find what they are looking for, to browse for products and services, and to evaluate potential purchases.
These are exactly the things that are hard to do in most companies. That is, employees find that it is not intuitive to search for information in company repositories; they cannot easily browse within categories of knowledge; and they are not given the context they need in order to evaluate the quality of the knowledge they do find. This is why many knowledge workers give up in their efforts to find content that may help them to serve clients better or to conduct innovative research.
The Internet giants can serve as models for companies that want to more effectively “market” knowledge to their employees. If organizations apply the basic, proven approaches of these online businesses to capture the attention of their employees, they should be able improve their ROI on sunk IT costs while increasing knowledge-worker productivity.2
Organizational efforts to make knowledge more valuable rarely begin with assessments of employee (“customer”) needs. Many organizations assume their employees are a captive audience willing to seek out the content they need, regardless of where, or in how many repositories, it is stored. Yet for many employees, the cost of finding and qualifying knowledge in a repository often exceeds the benefits — even if the additional knowledge could help them do their jobs better.
Employees face three common frustrations in their efforts to access the content they need: finding, browsing and qualifying knowledge. And knowledge-worker experiences and frustrations are virtually universal in nature, regardless of an organization’s industry, location or status as nonprofit or for-profit. The first frustration comes from having to navigate through multiple repositories, a time-consuming and often fruitless chore. The second frustration is that when possibly useful content is retrieved —through a keyword search, for example, or by browsing the corporate intranet — search results are often confusing and the information is presented in formats that are not easily comparable. Finally, content in a system is often redundant or outdated, making it all the more difficult for employees to assess the relevance of what they find.
Consider a scientist at a pharmaceutical company who spends hours looking for the results of tests on a compound he is screening for effectiveness in treating a cardiovascular disease. He is confused by what he finds because he doesn’t know which results are most current, isn’t sure whether he has all the results from the different therapeutic areas that might be conducting tests (or have already done so) on the same compound, and can’t easily compare the results since the reports contain different information about the screens. As a result, the company’s collective experience and expertise are not brought to bear in this instance, and the costs to customers and the firm of this and similar experiences are enormous.
To overcome problems of this magnitude, employees must believe that the benefit of finding and applying knowledge is greater than the investment of time needed to do so. By emulating certain approaches taken by the Internet giants, organizations can significantly increase the value of their existing knowledge by making the search process worthwhile.3 More specifically, companies should undertake three actions: They should provide one-stop access to content, design a dynamic classification approach and consistent content formats, and entice employees to easily find the knowledge they need and also to locate the unexpected.4 While each may seem simple in isolation, the power resides in their combined impact on employees’ experiences and the fact that they are based on proven successes and adopted behaviors in the marketplace — indeed, a large percentage of employees in any company will already be familiar with these Internet sites, making the adoption of their principles relatively easy.
Google, eBay and Amazon may not think of what they are doing as making knowledge easier to find and use, but their approaches are the essence of creating value from knowledge: They create a satisfying customer experience. While the three companies all address the three knowledge challenges we have identified, each is particularly strong in one of the design attributes.
One-Stop Access to Content
The explosive growth of Google is testament to the importance of the first design attribute — simple one-stop search functionality. Google has proved that people place a great deal of value on a streamlined ability to search when dealing with massive amounts of information. With an estimated 200 million searches logged daily (accounting for 40% of all searches on the public Internet), Google is the most popular Internet search engine, attracting 66 million unique visitors per month.5
Imagine the impact on productivity if organizations could apply one-stop functionality to internal knowledge searches. Employee frustration over having to deal with multiple repositories would vanish. The possibility that they would fail to find information because they didn’t even know of the existence of a particular database would likewise disappear. In addition, many employees at any given organization are likely to have used Google in the past and value their ability to search using free-form phrases in the white bar prominently centered on its main page. Comprehensive one-stop search functionality in an organization can save people a great deal of time and give employees the confidence that they are not missing important content that may be hidden away in other databases. See sidebar.
Since most companies have multiple repositories that are not integrated, employees conducting white-bar searches generally fail to access the full scope of the company’s codified knowledge. In many companies, more than half of all content cannot be found from a simple search on the intranet portal. And employees typically are unaware of the existence of content that they are not searching for specifically but that might be valuable to them. This problem often stems from multiple divisions or groups creating their own portals or separate unintegrated Web sites on a single portal.
In addition to the problem of unintegrated repositories, companies frequently rely on technology alone to solve the search problems associated with poor back-end design. While search-engine technologies have become more sophisticated, unless they are customized they are unlikely to solve all a company’s search challenges. To get better results from these investments, companies should design the search functionality with content meta-data or tags on documents, which include keywords, abstracts, author name and document date. Better metadata leads to better search results. Implementing these types of changes takes some effort but does not typically require major capital outlays for technology, since most large companies have already invested in basic portal technology.
Dynamic Classification, Consistent Formats
People don’t always know what they don’t know. That’s why taxonomies — classifications of items in an ordered system that indicates natural relationships — are so important: They help searchers who are browsing for information and don’t know exactly what they are looking for, leading them to interesting content in the process.
The value of knowledge within organizations would increase significantly if it were always classified according to a well-designed taxonomy. The key is to establish an organizing structure that is based on how work is done and what work needs to be done. Knowledge, after all, follows the rules of supply and demand — the good material is used, the rest is not. (Looking at which documents get downloaded is one way to see this phenomenon in action.) But while knowledge grows and evolves with the dynamic needs of an organization, few structures and approaches exist to accommodate this growth.
Many companies have no taxonomy at all, and documents are haphazardly organized without regard to the way employees use knowledge. A good taxonomy has three main characteristics. First, it is easily recognizable by authors as a comprehensive representation of their expertise. Second, it organizes knowledge the way people actually use it. Third, it allows for growth in breadth and depth of knowledge.
Although we do not often hear about this element of its online marketplace, eBay has developed a dynamic navigational taxonomy. When eBay began, a small set of goods was sold through its site; now one can find thousands of categories there. EBay’s classification system has developed based on what its customers want to buy and sell. The structure and taxonomy have grown over time and continue to grow to meet customer needs.
Another important method of improving the browsing experience is through consistency of formats. Again, when formats look the same, people can find what they are looking for more quickly and will spend less time analyzing results. When standard formats do not exist, material is often represented inconsistently, causing system users to spend time collecting and analyzing multiple representations of similar knowledge. The lack of templates containing standardized information can also be damaging. For example, when searching for an industry analysis, a researcher may find it in two different formats that contain different, not easily comparable, information. Obviously, it will take more time to analyze the reports and could lead to a misinterpretation of the knowledge.
EBay, again, is a model for companies seeking to put their knowledge into consistent formats. Within its taxonomy, it has standard formats for goods or services that further help shoppers and sellers easily navigate through hundreds of thousands of products. For each product, a shopper knows the asking price and what others have bid. He can view a photo, read a description and get information about the seller. These are all critical pieces of knowledge that help buyers make their decisions and sellers market their products.
With a quick review of formats currently in use, organizations can begin to identify common formats and focus on those that are preferred by employees, are most helpful and are most often used. They can then begin to institutionalize common formats more broadly across the organization or within specific departments. This approach to organizing knowledge should reduce rework, allow organizations to build on their growing knowledge and limit the need for employees to adapt their behavior to a complicated new process. See sidebar.
Helping Employees Create Value From Internal Knowledge
Search results from corporate portals very often deliver bland lists of documents requiring employees to scan long lists or open multiple documents to determine which ones are most relevant for their purposes. Companies seeking to extract more value from their knowledge should pay attention to the way Amazon helps people find things, including the unexpected.
Amazon knows that having the proper tools to help shoppers answer questions they have when searching for products has a direct impact on sales. The retailer makes it easy for customers to quickly identify products they may want to buy based on their relevance. For example, suppose you want to buy a birthday present for your father, a prolific reader. You can quickly identify a good gift using the multiple search criteria and the product snapshot. You can tell the system some of the books your father has liked in the past and it will suggest others, based on a purchasing algorithm showing that customers who bought certain books also tended to buy certain others. Or you can look at customer advice, a function that lets customers recommend a book (or another product) in addition to, or instead of, the one you are thinking of purchasing. The product snapshot for books includes list price and Amazon price (so you can see how much you save), information about a book’s availability and much more.
Amazon also provides a great deal of information that helps shoppers assess the quality of the products they are viewing. For example, they can get a quick snapshot of the feedback on a product by looking at the average score from the total of all customer reviews. Perhaps the most enlightening way to judge a product’s quality is by reading the reviews, which include both individual customer opinions and reviews published elsewhere. If a product has a large number of reviews, Amazon highlights certain “spotlight reviews” to indicate those that have been deemed most helpful by customers.6 Another proxy for quality is sales rank, which also appears with the product snapshot.
Taking a page from Amazon’s book, companies could provide tools and categorizations to help employees determine the relevance and quality of the knowledge they find, including
- Relevance ratings that allow employees to sort content results by relevance percentages (based on a keyword, for example), date created or date modified, department or group that created the content, and whether it is historical or archived versus active or new knowledge;
- Quality ratings that allow employees to know if the content passes certain hurdles (is it deemed clear, useful and current, for example) or ratings that allow employees to describe the situations in which they found the content most helpful;
- The display of the first several sentences of the abstract in the search results to help employees quickly scan more detail;
- A snapshot of the content, including length, author(s), content summary, type of content (proposal, case study, evaluation of a process), whom to contact for more information if not the author (or if the author has left the organization), and estimated “shelf life” of the content (after which it should be reviewed again or archived);
- A link to an author profile providing employees with quick access to related material by that author.
These types of changes can likely be made on a company’s existing technology platform without major (or in many cases, any) technology-infrastructure investments. Most large companies have a technology base in place. The changes needed are to the design and use of the system and require mostly labor and, in some cases, limited capital investments for add-on software capabilities.
Design Attributes in Action
The cost to implement these design attributes would be incremental compared with the IT investments many organizations have already made, but the payoff could be substantial. Consider the potential impact these changes could have on the research and development process in a pharmaceutical company. Bringing a new drug to market costs about $900 million, and first-mover advantage can have an enormous impact on the company’s revenues. Consequently, pharmaceutical companies have a strong economic imperative to get as much value as they can from the knowledge in their R&D function and to get it quickly.
Scientists testing a compound for possible use in treating a disease frequently run different screens to determine its effectiveness. When this information is captured, it might reside in multiple locations. For example, biologists might have one database and chemists another. Or the results might be in documents that only a project team or therapeutic area (such as pulmonary or cardiovascular) conducting research on the compound has access to. Consequently, other scientists at the firm may waste time searching for the research results from these compound screens. Worse, they might not be able to find any results and might unnecessarily — and at great cost to the company — repeat a screen. Applying the first design principle from Google — providing one-stop search access to the knowledge — could reduce a significant amount of time and effort spent gathering such information.
Another problem facing scientists in this scenario is that the results from compound screens are often inconsistently captured. For example, they may contain little in the way of standardized information about the compound’s solubility, molecular weight, availability and cost, how the screen was conducted, and where it does and does not work. Poorly designed classification systems and the lack of standard formats for capturing content make it difficult for scientists to browse results to determine a compound’s potential efficacy in treating a targeted disease. A dynamic classification system and standard formats could significantly reduce the time wasted by scientists.
For instance, a pharmaceutical company might design a system in which standardized results from a compound screen —with inputs that include assumptions, testing data results and application scenarios — are all located in a single repository that can be easily updated. In addition to ensuring analytical consistency in the screening process, such an approach would reduce the duplication of work by enabling scientists to determine which tests have already been run on a compound by others. It would also increase the speed of decisions about promising compounds and avoid the costly advancement of those with significant problems, such as toxic side effects.
Finally, R&D scientists often struggle with search results that are outdated or contain partial results that overlap with other research findings, making it difficult to know what is actually known about a compound or whether it merits further development. Talking to the scientists who performed a particular screen might help shed light on the approach used and the results achieved, but the names are often not included in the database. Here, too, researchers spend too much time assessing the relevance and quality of research that has already been conducted. Applying Amazon’s design principles could markedly improve the situation. For example, it might be useful to create different relevance categories for scientists to use when searching, such as the date the screen was performed and the therapeutic area in which it was tested. Content snapshots with a summary and information about who was involved and who to contact would also be valuable, especially if they were linked to a profile of the relevant experts.
In all three situations, the cost of employees not being able to find and qualify existing knowledge is high. Researchers might misinterpret information and push ahead with a compound that is doomed to fail, or they might terminate research on a promising compound. Imagine, for example, if the severe side effects of a compound that one team had documented were not found by another team exploring the potential of that compound for a different disease target. Development time could be unnecessarily extended, at huge cost to the company.
Applying the three design principles embodied by Google, eBay and Amazon can help companies reduce search costs and avoid the opportunity costs associated with failure to find the best available knowledge — costs that include stifled innovation, decreased productivity and lost revenue opportunities.7 The good news is that the principles can be implemented without the multimillion-dollar investments that many large companies initially made at the beginning of the knowledge-management era. Organizations can improve employee productivity with incremental investments by taking to heart these simple lessons from three Internet giants about how to get more value from their knowledge.
1. Evidence of substantial corporate spending on IT for knowledge management has been well documented. Companies spent $2.7 billion in 2002 on knowledge-management software, with document-and content-management software capturing the top share. Worldwide spending on knowledge-management services totaled $2.6 billion in 2001, much of it on documentation efforts. The software market is expected to increase to $4.8 billion in 2007, and spending on services is expected to increase to $4.6 billion in 2004. In addition, 85% of the Global 2000 are expected to have selected a portal platform by the end of 2004. Employee portals represented companies’ largest expected software spending priority in 2003, with content and document management remaining a close second. Sources for these figures: S. Feldman and C. Sherman, “Industry Developments and Models: The High Cost of Not Finding Information,” IDC, Framingham, Massachusetts, April 2003; B. McDonough, “Market Analysis: Worldwide Knowledge Management Software Forecast, 2002–2007,” IDC, May 2003; C. Philips and R. Rathman, “Morgan Stanley CIO Survey Series: Release 3.8,” Morgan Stanley, Dec. 9, 2002; L. Conigliaro and R. Sherlund, “IT Spending Survey,” Goldman Sachs, Jan. 2, 2003; C. Roth, “Enterprise Portal Frameworks: META Spectrum Evaluation,” META Group, Inc., Stamford, Connecticut, Oct. 20, 2002.
2. Many approaches to creating more value from knowledge have been addressed in the management and academic literature. These include transferring and applying tacit knowledge and expertise; acquiring, creating and codifying new knowledge; and organizing existing knowledge, both tacit and codified. We focus on a subset of these issues — knowledge that has already been codified and placed in a repository. This knowledge represents significant sunk costs by organizations that we believe could be made much more valuable with incremental investments. The management of codified knowledge has been discussed in M. Zack, “Managing Codified Knowledge,” Sloan Management Review 40 (summer 1999): 45–58; R. Cowan, P.A. David and D. Foray, “The Explicit Economics of Knowledge Codification and Tacitness,” Industrial and Corporate Change 9 (2000): 211–253; and R. Cowan and D. Foray, “The Economics of Codification and the Diffusion of Knowledge,” Industrial and Corporate Change 6 (1997): 595–622. However, little attention has been given to the relationship between codified knowledge and knowledge-worker productivity in particular and to illustrating by analogy the ways in which nonintuitive case studies have relevance for managers trying to increase the value of their organizations’ documents.
3. An implicit assumption is that existing documents represent some or most of the codified knowledge a company would want employees to find and apply. If this were not the case, organizations would want to devote more time to identifying, codifying and prioritizing the most important knowledge.
4. These actions build on the significant contributions in the literature to the design of information- and knowledge-based products, including M.H. Meyer and M. Zack, “The Design of Information Products,” Sloan Management Review 37 (spring 1996): 43–59; and the more popular views of information design by E. Tufte (all published by Graphics Press, Cheshire, Connecticut): “The Visual Display of Quantitative Information” (2001), “Visual and Statistical Thinking: Displays of Evidence for Decision-Making” (1997), and “Envisioning Information” (1990).
5. D. Hochman, “In Searching We Trust,” New York Times, Sunday, March 14, 2004, sec. 9, p. 1.
6. EBay is similarly successful helping customers qualify potential sellers, which is critical given that their business model is based on trust. Every eBay member has a profile with basic information about the member and a list of feedback left by their trading partners from previous transactions. Learning to trust a potential seller (or buyer) has a lot to do with what their past customers or sellers have said about them. For each transaction, only the buyer and seller can rate each other by leaving feedback, which consists of a positive, negative or neutral rating and a short comment. The feedback gives other eBay shoppers and sellers a good idea of what to expect when dealing with that person.
7. Consider an example: If average employee salaries are $80,000 and an employee searches for knowledge on average twice daily, then a five-minute savings in search time could result in 40 hours of time saved over the course of a year per employee (assuming 240 working days). This represents over $15 million for an employee base of 10,000 workers and 400,000 hours of time that could be better spent on more productive activities than searching for knowledge.