A Strategy for Improving Knowledge Work
In search of a strategy that can improve knowledge work just as industrial work was improved in the last century.
The economic gains of the 20th century were driven by the industrial productivity improvement techniques pioneered by Frederick Taylor. Those techniques worked by observing a job, breaking it down into a series of component tasks and skills, and then defining the sequence of tasks and skills that yielded the most output for the least effort. The range of jobs for which this strategy worked was broad, ranging from factory jobs to many office jobs. Refinements to that strategy included simplifying and specializing jobs, identifying and eliminating work steps that didn’t add value to the final output, and substituting automation for people when the numbers said it made sense.
Why wouldn’t this same strategy apply to knowledge work?
The essence of industrial productivity improvement is to reduce or eliminate variation in the output of a job by reducing or eliminating variation in process. As Peter Drucker observes, the defining characteristic of knowledge work is that knowledge workers define their own jobs; moreover, they are expected to define what constitutes quality output. In knowledge work, quality is "Job 1" by definition. The industrial strategy for productivity improvement applies only after the identification of a known job with a known output.
If this strategy won't work, what will?
One strategy for improving knowledge work might be to simply wait until the tasks and quality output are understood and move ahead with conventional productivity improvement. When a knowledge work task can be converted to a standardized information-processing task, this strategy may well work.
Consider inventory management. In the 1960s, managing inventory levels was a complex knowledge-work job. Over time, operations researchers structured and defined inventory management problems so that what were managerial decisions became the outputs of accurate information filtered through algorithms. What was a knowledge-work job morphed into an industrial information-processing job.
When Jobs Aren't Fully Understood
What about knowledge work jobs that have yet to reach that stage?
I’m convinced that there are many jobs in today’s organizations that won’t soon be turned into well-defined information-processing tasks—that is, tasks that will have major impacts on the success or failure of the organization or will be the day-to-day work of more people in the organization.
Although the power and success of conventional productivity-improvement techniques is tempting, applying them to these jobs poses several risks. Chief among them is the risk of locking in on and improving a process that doesn’t actually exist. One of the forgotten lessons of the days of expert systems is that experts (a.k.a. knowledge workers) are simultaneously glib about what they think they are doing and generally only partially able to recognize and articulate how they actually work. Failing to appreciate and account for this discrepancy risks applying improvement strategies to irrelevant and accidental features of a hypothetical process.
Another risk flowing from ill-defined (and ill-definable) processes is that technology will be applied inappropriately. Think of knowledge management systems that sweep up and inventory all final deliverables without supporting better communication among experts. Think of word processing software that is powerful enough to lay out a multi-column employee newsletter, yet offers no tools to help a writer organize and manage research notes and early drafts on the path to a final report.
If the machinery of industrial productivity improvement is inappropriate, is there an improvement strategy that fits the particular characteristics of knowledge work before it can be reduced to conventionally improvable work?
In many respects, knowledge work is craft work. Like craft work, the outputs of good knowledge work are utilitarian and esthetically pleasing. The quality of outputs is a function of the level of mastery of the knowledge worker. The unique features and characteristics of each knowledge work product are what makes them valuable.
The Craft Work Connection
The importance of these parallels between knowledge work and craft work is that the improvement strategy that drove productivity improvement in the industrial economy remains applicable, even if the specific tools are suspect. That strategy begins with observing the work itself. (See my comments about work visibility at http://esj.com/Enterprise/article.aspx?EditorialsID=1327.)
What we should be looking for is an understanding of what contributes to quality in the final product and what detracts from quality. Unlike industrial productivity improvement, we won’t be focused on eliminating variations in final outputs, although we may reduce those variations that don’t link to quality differences. What we can expect to discover, courtesy of Peter Drucker, is that current methods and tools don’t work very well even though today’s knowledge workers will be quite comfortable with those methods and tools.
That will lead us to a series of challenges that are rooted in change management, but the foundations will be observing how knowledge work is done and understanding how to recognize quality in the end result.
Jim McGee is a Director at Huron Consulting Group where he helps clients improve their IT organizations and the practice of knowledge work.