cpt chart The field of Knowledge Management (KM) has existed for about a decade, and after an initial flurry of enthusiasm (at one point six of the top ten business best-sellers were about KM) it has fallen into disarray. Part of the problem is that the field has been dominated by three largely disconnected groups:

  • Academics & business gurus, who write about theory that is too general and abstract to have much practical application, 
  • Knowledge managers and project managers who cobble together pragmatic custom applications, often in an undisciplined and unsustainable way, applications that are often abandoned as needs, roles and technologies change, and
  • IT managers who, with the best of intentions, buy and install commercial ‘KM tools’ that never get much front-line take-up

Ten years later many organizations have little to show for large investments in promising KM tools, projects and infrastructure. What went wrong, and is it too late to save KM from the scrap heap of failed management fads?

I believe one cause of the failure of KM was the attempt to build generalized tools that were expected to have application in almost all industries and business processes. Such tools work well enough in the old financial information (FIS), sales and marketing (SMIS) and human resources (HRIS) systems. In these ‘classical’ IT systems, both the content (financial, customer and employee data) and the application (financial statements, customer reports and employee records) are relatively standard across a wide variety of industries.

By contrast, knowledge content (like leading practices, industry analyses and methodologies) is particular to each industry and to each department and process within a company. Here are four examples to illustrate this:

Example of Content
Example of Process
R&D process in a pharma company
patent for a new drug
drug development
Sales process in a consultancy
customer analysis
selling an assignment
Production process in a newspaper
article or editorial
intelligence gathering, editing
Distribution process in a newspaper
newspaper edition

A knowledge tool designed for one of the above processes is unlikely to be optimal for another process, any more than a machine designed to make lasers would be optimal for blending fruit.  And there are three additional dimensions to the problem that complicate matters further:

  1. There are four steps in the knowledge/learning cycle:
    1. Capturing: e.g. putting it in a memo, filing cabinet or Windows folder
    2. Manipulating: e.g. repurposing, reapplying, or reorganizing it
    3. Learning: e.g. internalizing someone else’s knowledge, reading, taking courses, OJT
    4. Sharing: e.g. conversations in various media and forums to exchange knowledge
  2. There are three ways to improve the ‘knowledge culture’ of an organization (see diagram at top of this article)
    1. Develop or improve tools which enforce standard processes which, in turn drive effective employee behaviour,
    2. Develop, teach and inculcate processes that drive effective employee behaviour, and
    3. Directly address effective employee behaviour by training, reward systems, communication etc.
  3. There are three distinct types of knowledge in orgaqnizations:
    1. Intelligence, also known as ‘explicit’ knowledge or ‘ know-what‘ (such as the four content examples in the chart above)
    2. Expertise, also known as ‘tacit’ knowledge or ‘know-how
    3. Networks, contacts, relationships and ‘know-who’

In its full complexity, this can lead to knowledge problems as diverse as these:

  • How can a pharmaceutical company capture patent information more effectively to reduce drug development costs
  • How can a consultancy’s intern learn the expert’s know-how about, for example, selling assignments to CIOs
  • How can a newspaper repurpose archived articles and editorials to provide readers with better context for today’s news

It is naive to believe any small group of standardized ‘KM tools’ could optimally solve such diverse knowledge problems right out of the box. So, if a tool is to be part of the solution at all, it must be highly specialized or customized to provide the precise functionality needed to solve the problem, without awkward or extraneous features. That is not to say that commercial tools could not provide a starting point for the design of an appropriate solution.

From the above, a six-step process for knowledge-based performance improvement can be inferred, which might provide the basis for a more flexible and sustainable framework for a new business discipline than KM has proven to be:

  1. Start with the Problem: Identify the specific productivity, revenue generation, customer satisfaction, learning, or decision-making (these being the five value propositions for KM) problems in each business unit of your company, that are caused at least in part by lack of, or barriers to, exchange of knowledge.
  2. Set Targets: Identify specific, realistic ‘success’ metrics, that tie directly to improvements in one or more of the above value propositions; set a current state benchmark and a target benchmark for each metric.
  3. Decide on a Solution Set: Determine which combination of tools, process improvements and behaviour change programs (training etc.) will optimally solve each problem and attain the target benchmarks.
  4. Create Tools: If tools are part of the solution set, decide whether to buy and modify, or build, these tools. Don’t expect tools ‘out of the box’ to work effectively. Don’t buy any tool, no matter how cheap or elegant, that doesn’t solve an acknowledged business problem in your organization.
  5. Invest in the Solutions: Decide whether a one-time project will achieve the target benchmarks, or whether an ongoing investment in infrastructure is needed to sustain improved performance. Make the investment.
  6. Implement: Implement, measure, continuously improve, or, if they no longer work, obsolesce the solutions and start again.

Bottom line: Forget all the jargon and theory of KM. Be skeptical about most ‘KM’ tools on the market. Start with your business problems, not the features and benefits of proffered solutions. Knowledge tools, processes and programs are just means to a business end.

The author has been Chief Knowledge Officer of a large multinational organization since 1994. In Part 2, he will provide a Case Study applying the above methodology.

This entry was posted in Working Smarter. Bookmark the permalink.


  1. The Raven says:

    I see that this is something like the third or fourth “KM” piece you’ve run here. That being the case, it seems reasonable to infer that you have some specific connection to this material beyond a passing interest. Do you train people in this stuff? Have you seen organizations that are running off the rails get bailed out through KM implementation? How about some background here? Regards, – R.

  2. Rayne says:

    The biggest single problem I’ve run into with implemented KM or the lack there of: flexibility of data. Functions operate in silos, build data into silos which can’t be crossed by other functions, instead of find a way to create a megadatabase that’s highly flexible backend. With and without KM.It’s as if everyone forgot Rummler-Brache.

  3. Dave Pollard says:

    Raven: Background is that this is what I have done for a living since ’94, and ours is one of very few organizations (I know this first hand since I’m on consortia with about 30 other CKOs) to have achieved considerable KM success. I have seen some notable, focused KM successes that have solved very specific business problems with very satisfactory results, and they have used the approach I describe in this article. I know of many other organizations that have caused more problems than they’ve solved with KM, and hence the high rate of abandonment.Sorry for tossing this over the wall into my blog without advance warning. This version is meant for those that live with this stuff. I need to develop a longer version for lay readers with a lot more background and examples, but I needed to get my ideas down. I do have a few readers in the KM field that read this blog and will find this comprehensible. Other readers, please step around this large imponderable object and continue with your reading.Or keep asking questions if the subject interests you. -/- Dave

  4. Dave Pollard says:

    Rayne: You’re right, one of the major KM problems is that information (perhaps like this article) isn’t shareable unless there is context around it. Large data warehouses that can be mined for specific trends can be very valuable, but large knowledge warehouses usually just make it harder for ‘shoppers’ to find stuff. Imagine a Wal-Mart that has no signs and organization and has stuff just piled anywhere with cryptic little labels saying what it is, and you get the idea. Actually, I think I’ve been in such a Wal-Mart. Often in KM as in blogging, more is less, and less is more.

  5. Rayne says:

    Dave – part of the problem I’ve seen both in business frontend and IS backend is that people only focus on immediate objectives — they pick a nice GUI software based on achieving those objectives, nevermind what’s behind the GUI. If an organization could at least agree that all GUI’s should reside over a standard database platform underneath, it would make “bridging the white spaces” much easier. Actual example: organization runs its entire global financial books on SAP-mainframe; CRM is Siebel on IBM-owned servers; related customer service software is in a specialized software with Oracle backend on Unix and NT servers; production is in VAX and Unix systems, several different databases. A marketing manager could not reasonably expect to be able to pull any new reporting on forecasting he needed without a sizeable staff and month(s) of time. Silly, when it might have been possible for everyone to agree to port a clone of data to a single data warehouse or at least find GUI’s that sit over a standard database app. All the data is labeled already, but not minable unless reentered as it stands today.

  6. Dave Pollard says:

    Yes, we’re really fortunate that 90% of our knowledge is in Notes or HTML, and the other 10% is old legacy systems that are being replaced this year. The replacement FIS and HRIS are being built specifically to work with the Notes/HTML apps. So data mining is, and will be, easy. Alas, with over a million reports and presentations to search through, the task of finding relevant best practices is still a challenge. But I sure don’t envy the additional technical nightmare in the organization you cite.

  7. Rayne says:

    Dave — the example cited is far more the norm than not in certain industries with highly distributed computing. Mergers & acquisitions are a special hell to IT project managers. I’d have slit my wrists if anyone had suggested, oh yeah, let’s do KM integration during the M&A. Argh!

  8. Dave Pollard says:

    Even companies that don’t have the IT Hell of M&A to contend with, often face one almost as bad– globalization. By comparison, globalization of KM is a cakewalk.

Comments are closed.