Gastbeitrag von Richard Reed
Richard Reed ist Mehrheitsaktionär und managing director im Bereich Entwicklung und Marktforschung der Coextant Systems International AG, Stuttgart, die sich auf automatisierte Content und Knowledge-Management-Systeme spezialisiert hat.
( http://www.coextant.de )
Critical Points to Consider
Most organizations these days are considering or actively pursuing knowledge management solutions. When it comes to requirements, such solutions have many aspects in common. For example,
(1) sharing critical corporate knowledge
(2) providing just-in-time access to critical knowledge, and
(3) making such knowledge searchable from a single entry point.
As our business we've learned quite a lot about what works and what doesn't work, and about where strategic failure points can occur. In this essay I will share the most critical concepts so that you can draw on and consider these experiences in the hope that they will help you avoid some of the slippery slopes found in such implementation efforts.
The Information You Need in the Format You Need at the Time You Need
One of the central themes in today's Knowledge Management solution thinking is to centralize all contributed content in master "libraries" of information that are all available in the same place for searching. The trouble with this thinking is that it assumes all content is the same, all purpose is the same, and that all content consumers have the same needs and abilities. Information in the form of short documents, long manuals, slideshows, images, spreadsheets, and various multimedia formats is all "dropped" into centralized libraries and consumers access this content in each document's native format.
But this is clearly not workable in an organization that needs to be competitive. Obvious drawbacks are:
(1) content consumers are required to have a comprehensive set of authoring applications installed on their local computers,
(2) search result hit lists can take readers to long documents that they need to search manually once they have been opened, and
(3) a single format for a specific document is not useful in every situation, wasting the valuable time of staff members who are trying to complete mission-critical business processes.
As an example of the last issue, consider that a 300 page manual will not provide an on-line reader with rapid access to specific information contained in it if the document is released as a long source file or PDF, or that a hypertext version of a procedure is useless to a user who must take a properly printed procedure into a lab where computer access is scarce or not possible. If a single piece of knowledge is useful to many types of users, it is likely the case that a different format of the document is required to optimize access to the information depending on the situation. Without providing these formats, the problem of making information accessible has been solved (or the reader is forced to do additional work), but the real goal of providing just-in-time, rapid access to information that is required to complete a task has not.
One critical requirement of any knowledge management solution, then, must be that users can locate the information they need, when they need it, and in the format they need it to be in to allow them to complete their current task without delays resulting from time wasted reading through documents to find what they are looking for.
Centralization of the Content Process Can Strangle A Solution
Some organizations recognize this last requirement as essential for implementing a process that provides true value to the business process. But frequently the implementation approach that is selected is the one that is the fastest to completion, without regard to the impact of that decision. This "fast" implementation is to centralize the knowledge repository in a single location, often IT, and designate a small staff to be responsible for making key pieces of information available in multiple formats to meet the just-in-time requirements of content consumers.
In short, this approach makes a knowledge management solution "pregnant with failure." It will only be a matter of time before that baby is born. I will explain this in more detail. It is interesting to recognize that the same concept is the reason why so many organizations without web content management automation in place have failing web site initiatives.
If a single unit of information requires 10 short but concrete manual steps every time a new version is made available, then revising that piece of information has an initially invisible human resource cost in addition to the base cost required to keep the content current through authoring. Invisible because the cost seems so low:
(1) get the document
(2) open it in the word processor
(3) create a PDF
(4) attach the PDF into the solution
(5) fill in search metadata
(6) save the document as HTML
(7) post the HTML document
(8) fill in the metadata for the HTML
(9) review the PDF and HTML in context of the knowledge base or web site to make sure there were no conversion problems and links work
(10) push an approval button to post these items for availability.
In a situation where you have as little as 5000 documents, some of which are regularly revised, a solution that requires a small team to perform these steps with each new document or revision creates a clear bottleneck to keeping content up-to-date (this is why many web site solutions fail and web sites become out-of-date so quickly). It is critical to consider the impact of such effort rather than dismissing it as negligible because it adds up quickly. There is also the additional problem that the centralized team cannot be expected to understand how each document is used and what actions should be taken to make it properly useful. But also with this amount of material, an expectation that the content consumers should have to do additional work all the time to access the content they need is also clearly unacceptable.
Another critical requirement of any knowledge management solution, then, is that the process cannot be centralized unless it is automated, and in many cases it should not be centralized at all because it sets up the conditions for various forms of bottleneck failure in the future, and ignores the "localized" and correct treatment of the knowledge managed by the solution.
The Smartest Way to Provide Superior Intelligence is to Implement Massive Dumbness
Knowledge Management must not be looked at as a technological solution, but rather as a content solution. When you look at it from this perspective, it's fairly obvious that with many business units in the organization, each knows best how its content is created and in what circumstances it is used. The secret to implementing a successful KM initiative is to recognize and leverage this expertise, for it allows each distributed KM solution to implement as much or little intelligence about its content as is appropriate. This which is better than applying one standard to all content, because in most cases that makes it necessary to engineer all content equally. A process that is not only costly but which in most cases provides little benefit.
If you compare a centralized approach with a decentralized approach with this thinking in mind, you should reach the conclusion that a decentralized approach to implementation of KM is indicated. Once a number of web-enabled, decentralized "mini-KM" solutions have been created, centralized approaches can be used to provide navigational possibilities for standard appearance and searching without introducing the risks of content scaling or inappropriate treatment of content.
But obviously decentralization requires a lower-level of complexity at each point. Business units are usually not sufficiently equipped nor interested in running a complex KM process. In addition, for each mini-KM solution to use usable from a thin navigation framework provided by the organization, each has to be reasonably "dumb" about its user interfaces and not be technically complex to integrate. So the solution in each business unit needs to be very easy to use and administer. This means it must provide reasonable amounts of automation that function on author-supplied metadata, but require content authors and content consumers to understand little else about the solution. The solution must make technology invisible. Remember that "users want to do nothing."
A final critical KM implementation point is then to implement a number of easily plannable, implementable, and maintainable "points of light," rather than to implement a single blazing sun that requires content fuel to burn properly.
KM isn't just about "getting documents out there." KM can be expensive. To create a successful KM solution, it is crucial to ensure that if the money will be spent, that the organization will reap benefits that will truly help increase its competitive ability. It's not worth spending the money unless the benefits are delivered.
This means essential KM solution goals must include:
| || || |
Content consumers get what they need,
when they need it, in the format they need it, just-in-time
Content authors focus on keeping content up-to-date using standard authoring tools and "do nothing else"
KM implementations start with "being dumb" and implement additional organization intelligence only for specific areas and specific content
The architecture should be distributed rather than centralized
Automation should be used to create the required just-in-time benefits
Standards should be documented for making individual KM efforts capable of being integrated into the organization's navigation framework
The above information should be very helpful for you to consider additional issues before you get started with your KM efforts. Look before you leap, and best wishes for a successful project!