Figuring Out What Productivity Means

According to Peter Drucker, improving knowledge worker productivity is the economic priority in the Western world for the 21st century. In his article “Knowledge-Worker Productivity: The Biggest Challenge,” he discusses how to address that challenge.

He begins by recounting the rise of “Manual-Work Productivity” in the 20th century. From the first efforts by Fredrick Taylor through to Edward Deming, our industrial improvement was based on improving the productivity of manual workers. We did a great job. Drucker claims a 50-fold increase in manual worker productivity during the 20th century alone.

He then contrasts manual work and knowledge work to show how productivity happens in knowledge work. Note that most work falls on a spectrum between pure manual work and pure knowledge work. Manual work has always included some knowledge, and knowledge work includes some manipulation of the world, some manual skill. In order to make the contrast as clear as possible, in this article, I will try to get as close as possible to pure forms of both kinds of work. I’ll also follow Drucker in that I’ll make pretty absolute statements about the two ends of the spectrum.

Productivity for the manual worker

First, let’s define what productivity is for a manual worker. Then we’ll see what Drucker says about knowledge worker productivity.

In simplest terms, productivity is output divided by input. In order to increase productivity, the output must increase while input stays constant or the input must decrease while output stays constant.


In modern manual work, the output is predefined. An example will suffice: it would be difficult in a car factory to make anything but cars. As that output is broken into successively smaller chunks, individual-sized jobs are identified. Through this process, we arrive at a single task: put five lug nuts on the front driver side tire. That represents the output and, leaving quality aside, for now, it is fixed in the short term. Tools are designed and work begins.

In a well-designed factory, the inputs to each job are almost completely predefined also. The car arrives along with the tire and five lug nuts. The tool is there as well. So, the most easily controlled input is the cost of the time of the worker as the costs of the other variables are fixed. Therefore, productivity focuses on decreasing the amount of time the task takes.

The definition of quality is how much of the output meets the definition of output, as stated in advance. It is telling that quality, in manual work, is defined by a negative result, a bad output. This fact highlights our understanding of a pre-defined product. This idea will be important as we discuss quality in knowledge work. In manual work productivity terms, a bad output doesn’t count as output, so those inputs were wasted and overall productivity decreases. Quality is related to productivity, but as a maximum, an asymptote.

A few key differences between manual and knowledge work

A few ideas arise that may be important for us later when we look at knowledge work:

  • In manual work, productivity is approaching infinity. Robots are replacing humans, now that we have the technology and the tasks are completely defined and routine. If we don’t need a worker, the manual work input approaches zero, but the output stays the same. So the productivity of manual work goes to infinity.
  • In manual work, productivity is almost entirely in the hands of the boss. The people who designed the factory defined the task. The boss knows how it should be done, inside this particular factory. The boss tells the worker what to do and the (ethical) worker attempts to perform as close to that description as possible.
  • In manual work, the boss owns the means of production – the tools, the factory, the inputs. In knowledge work, the worker owns the means of production, the worker’s mental skills and set of knowledge. The worker cannot really sell those productive assets. The boss can’t own them. They can only be rented.

Let’s see what these differences do the notion of productivity.

Improving the productivity of knowledge work

What can we do to improve our own productivity in knowledge work?

Drucker gives six ideas:

Define the task

  • In manual work, the task is predefined. It was defined when the factory or assembly line was built. The task only changes when the factory changes, which is why the manual worker cannot, really, be innovative in the short term. The primary productivity question is: how should the work be done, in order that it takes the minimum number of inputs, including time?
  • In knowledge work, the worker defines the task, because only the worker knows the worker’s own knowledge set. It is unlikely that all of the workers in a category agree on what that task is. However, they can probably tell what their work is not.
    • Drucker cites an example of nurses. When asked about what their job was, the nurses were divided between patient care and serving the doctor. When asked what their work was not, they unanimously agreed that many routine tasks could be delegated to a cheaper, non-nurse resource.

Productivity is the responsibility of the individual knowledge worker, not the boss.

  • In manual work, the worker’s maximum productivity is defined by the factory, which is created by the boss. If the worker is diligent and careful, productivity will move forward. It is almost as bad for the worker to finish too quickly as it is for the worker to finish too slowly.
  • In knowledge work, only the worker knows whether the worker has the knowledge and skills to perform the job at all. Similarly, only the worker can estimate how much more productive the worker will be with additional knowledge or other tools.
    • As an example, if the task is to solve a bug in a computer program, only the worker can say whether the work is easier or harder given a particular tool.

Innovation has to be part of the knowledge workers’ job

  • In manual work, process improvements can be forced on the worker. New tools are a good example. If the boss wants the worker to use the new tool, he can simply replace the old tool. In addition, modern factories typically have a team of analysts to study how processes can be improved. Recently, some firms have started asking workers to help with this process, but change is expensive and management must choose to invest.
  • In knowledge work, only the knowledge worker can say which innovations might be useful. Other people can make suggestions, but the worker must have the final say on what makes work easier or better. Also, the worker can’t really be forced to invest in improvement but must care enough about the work to be motivated to do the difficult work of innovation.

Continuous learning and teaching

The fact that only the knowledge worker can do innovation is the basis for constant learning and teaching. Assuming that the knowledge worker wants to become more productive and, thus, more valuable, the knowledge worker must be continually improving. Improving is an internal learning exercise more than an external training exercise. It is in the firm’s best interest that the knowledge workers teach others so that other people can understand what the knowledge worker does and what to expect as output.

Quality of output is at least as important as quantity

Knowledge work has at least two problems when it comes to measuring quality

  • We struggle to define the task, as we saw above. Therefore, we have a difficult time determining how well it was done.
  • We don’t know what is possible. Let’s use this article as an example. My task is to deliver information and knowledge from Drucker’s article. What is the best possible fulfillment of that task? How close did I get to that ideal? How much of the information and knowledge did I need to deliver? These are questions that are impossible to answer in many kinds of knowledge work. The best we can do with knowledge work is to attempt the task and then figure out how well the solution addresses the problem. This is a very iterative approach and the exact opposite of the predefined nature of manual work.

Knowledge workers must want to work for the organization despite having multiple other opportunities

This arises from the fact that the knowledge worker owns the productive assets rather than the boss. Therefore, the market for knowledge workers is pretty efficient. If the knowledge worker identifies a better offer, the worker exercises the portability

In manual work, the worker’s knowledge set concerns how to make the boss’s assets work. That knowledge is likely particular to the specific assets, and, thus, not particularly portable to other contexts. In this sense, the worker needed the job more than the job needed the worker. The boss could threaten to remove the worker’s access to productive assets, rendering the manual worker useless. Thus, “The management of people at work, based on millennia of work being almost totally manual work, still assumes that with few exceptions (e.g., highly skilled people) one manual worker is like any other manual worker.”

From a manual work perspective, worker retention is an exercise in avoiding the cost of:

  1. finding a new worker
  2. training the worker in the predefined task, since a new worker is assumed to know very little about the work task

These costs also exist in knowledge work.

Identifying a knowledge worker with the relevant knowledge is a serious challenge; thus, a significant cost. It is likely more expensive than replacing a manual worker. So firms that need knowledge workers will want to avoid turnover at least as much as those who need manual workers. But, in addition to merely avoiding turnover costs of losing knowledge workers, firms need to recognize that carrots and sticks do not work well as motivators in knowledge work. In fact, in his book Drive, Dan Pink cites research that shows pay for performance (e.g., bonuses) actually decreases the quality produced by people doing complex work. In addition, since defining the output and measuring quality are difficult, knowledge workers need to be motivated to do their best work by participating in the vision of the organization.

The path forward for managing knowledge work

So, knowledge work changes our perception and practice of productivity. Drucker’s six thoughts show us how. As knowledge workers, we should seek opportunities to engage in these practices and improve our productivity. But Drucker’s thoughts also carry two themes about knowledge work that we need to recognize.

Management of knowledge work and workers is very different from the management of manual work and workers. If organizational leaders rely on dated mental models of management that are derived from the industrial age, they will actually interfere with the productivity of their knowledge workers. Many leaders struggle to recognize this problem. Even when they do, they struggle to understand what they should do differently. The ideas in Drucker’s article are a good starting place for any leader to begin to understand how organizational leadership should change in the era of knowledge work.

The difference in knowledge work management affects knowledge workers also. As knowledge workers, we are responsible for our own productivity to a much higher degree than we probably thought we were. However, there is little training in the methods of knowledge work. If, as Drucker says, the most important work is identifying what the work actually is, how do we go about doing that? In manual work, we looked to our organizations to define our work. Given the struggles of organizational leadership I described, we can no longer rely solely on our organizations to help us with productivity. We must be working consistently to understand knowledge work and to develop practices through which we can be more productive.

Check out the original publication of the article referenced here.