They don’t like to be told what to do. They enjoy more autonomy than other workers. Much of their work is invisible and hard to measure, because it goes on inside their heads or outside the office. They are a growing part of the U.S. workforce, and their skills are hard to replace.
They’re knowledge workers, and they are performing well below their potential because companies still don’t know how to manage them, says Thomas Davenport, professor of information technology and management at Babson College, in Wellesley, Mass., and director of research for Babson’s executive education program.
“Knowledge workers are going to be the primary force determining which economies are successful and which aren’t,” he says. “They are the key source of growth in most organizations. New products and services, new approaches to marketing, new business models—all these come from knowledge workers. So if you want your economy to grow, your knowledge workers had better be doing a good job.”
Yet after studying more than 100 companies and 600 individual knowledge workers, Davenport has come to the conclusion that the old dictum of hiring smart people and leaving them alone isn’t the best way to get the most out of knowledge workers. As he writes in his latest book, “Thinking for a Living: How to Get Better Performance and Results from Knowledge Workers” (Harvard Business School Press, July 2005), although knowledge workers “can’t be managed in the traditional sense of the word, you can intervene, but you can’t do it in a heavy-handed, hierarchical way.”
Executive Editor Allan Alter has followed Davenport’s career from his days as a pioneering thinker on business process reengineering and knowledge management. He met with Davenport in his office at Babson College’s School of Executive Education in order to learn how managers, and CIOs in particular, can improve the performance of this critical segment of the workforce. An edited version of their discussion follows.
CIO Insight: How do you define knowledge workers?
DAVENPORT: People whose primary job is to do something with knowledge: to create it, distribute it, apply it.
Most of the time they also have a high degree of education or expertise. They include anywhere from a quarter to a third of the workforce, but not everyone who uses knowledge. If you are digging ditches, you may have some knowledge on the job, but it’s not the primary purpose of what you do.
Are companies doing a good job of managing and improving the performance of knowledge workers?
They’re not. What most organizations do is HSPALTA: Hire smart people and leave them alone. We’ve spent a lot of effort recruiting knowledge workers and assessing how capable they might be before we hire them. But once they’re hired we don’t do a lot to improve their performance. Process improvement has mostly been for other workers: transactional workers, manufacturing workers, people in call centers. All the serious approaches to improving work have largely escaped knowledge work.
We let knowledge workers get away with saying there’s no process to their work, that every day is different. We don’t measure much of anything about knowledge work.
If we don’t measure knowledge work, why do you think there’s room to improve knowledge worker productivity and performance?
It’s a pretty well-informed hunch. People improve processes all the time; they just haven’t done it with knowledge-work processes as much. It’s an extrapolation of the same logic in other work, that processes can be improved.
Here is one number that indicates performance and productivity can be improved: IDC found that 1,000 knowledge workers can lose as much as $6 million a year just searching for nonexistent data, or repeating work that has already been done. Is it possible every knowledge worker is working to his or her potential? It’s possible, but unlikely. We can get a lot better at improving their performance.
Why aren’t more companies getting better performance from knowledge workers?
One of the problems is we treat all knowledge workers alike. Obviously it’s more convenient and efficient to impose the same solution on everybody. Certainly in IT, broadly speaking, we try to. It’s troublesome if everyone wants different software and computing environments, so we create common environments. But people work in different ways.
And politically, we don’t want to admit that some knowledge workers are better than others, and that some might deserve different office environments and technologies. We don’t mind treating the C-suite differently—why not our most productive knowledge workers? These are the people determining the future of your company.
Companies have spent billions on IT to help knowledge workers. Why aren’t our knowledge workers getting more from all these investments?
Most people feel more productive, and in part they are. But we spend a huge amount of time futzing around with stuff. Most organizations have no training or education on how to use these tools effectively in their work. Call somebody in a big organization and say, “Transfer me to your colleague Bruce down in accounting,” and 90 percent of people will say, “Gee, I’m not really sure I remember how to transfer. Here’s Bruce’s number just in case.”
We’ve had call transfer capabilities for 40 years in organizations. Why can we still not use them? The same thing is true of all these other devices—laptops, desktops, PDAs, cell phones.
Even when people are trained on knowledge-oriented applications, such as Excel, PowerPoint, CAD or CRM, the training focuses on how the software package works, not on how it fits into the context of the job. The vast majority of organizations that implemented CRM didn’t really help their salespeople figure out how to use the system effectively to help them sell better. It’s one of the reasons CRM has had the problems it has had. People were not comfortable using it with the customer around. And there weren’t any good examples of how salespeople did their work, so a lot of CRM systems were not effective at all.
Why hasn’t knowledge management helped more in the effort to improve knowledge-worker performance and productivity?
Knowledge management was an early attempt to intervene in knowledge work. For the most part, it wasn’t particularly successful, because we didn’t look closely at how knowledge workers did their work. We tried to be too broad in our focus.
Most organizations simply created one big repository for all knowledge and all workers.
The only way we can get people to use knowledge on the job is to understand how they do their jobs, and then figure out some way to inject knowledge into the course of their day-to-day work, not make it a separate thing you have to consult when you need knowledge.
We have to be much more targeted in approaching knowledge management. We have to target a specific job. And the best way is to use technology to bake the knowledge into the job.
One example I’m fond of—I wish I could find a lot more—is Partners HealthCare System, an organization of Harvard teaching hospitals in the Boston area. As with a lot of jobs, there’s just too much knowledge for a Partners physician to master. About 260,000 articles are added to the biomedical literature every year. So every time a doctor orders a test, a drug or a referral, the order-entry system checks it out and asks if it’s consistent with best medical practice.
The system might come back and say, “Sorry doc, but that’s the wrong drug for this setting, according to the Partners drug therapy committee.”
It takes a key work activity and injects knowledge into the process without the doctor having to go look things up. This works extremely well. It’s reduced adverse drug events, as they call drug mistakes, by 55 percent. It’s also resulted in fewer medical errors, fewer malpractice suits, and much faster response in terms of introducing new knowledge to doctors.
How do you get physicians to prescribe a new drug? You put it into the system, and it says that the Lipitor you had to take twice a day is now available in a once-a-day version. Doctors might not know about it if they had to master that knowledge individually.
How do we improve knowledge worker performance?
There ought to be a lot more experiments. We have experiments now, but we don’t measure anything, so we don’t learn anything.
If we say we’re going to put people in cubicles to improve communication, then we ought to at least measure some subjective aspects of communication before and after. We ought to see if cubicles work well with a small group before we put a whole company into them.
What is the most radical change that’s needed in the management of knowledge workers?
We need to start focusing much more on job-specific knowledge and information environments, at least from an IT perspective. That’s why I’m so fond of the Partners Healthcare example. Another example is Capital One, where they’re creating an information and knowledge environment for people who design credit card offers. Their credit card analysts, who do 30,000 experiments a year on what works in getting responses to credit card solicitations, figure out what rate they should offer for a particular customer segment, and decide what the attributes of the product are going to be, who’s most likely to pay off their bills, and so forth.
They won’t share much, but the idea is that when they design a new offering, the system tells them that blue envelopes don’t work for this audience, or the optimum interest rate for balance transfers for that audience is 1 percent over prime. It allows the analysts to learn from the vast amount of experimentation that Capital One has done.
Capital One applies some of the same experimental approaches to technology. Do their knowledge systems really work? Does it improve productivity? Does it improve communication? All the things that people talk about but never really measure.
Designing these knowledge environments for knowledge workers is expensive and hard to do. But if we’re serious about making knowledge workers more productive, we’re going to have to focus on particular jobs and sometimes even particular individuals.
There are a whole range of possibilities for differentiating knowledge workers so that we don’t treat them all the same. I don’t think you should use any one segmentation approach, but the one I like best is a 2-by-2 matrix that’s based on how much collaboration is involved in the job, and how much expertise is involved in the job.
Intel has five or six categories that differentiate knowledge workers on the basis of mobility and how aggressively they adopt technology. There are “cube captains” who don’t move around much, and “nomads” who move around a lot and need portable technology. Some involve factory workers and others who don’t have a lot of personal technology available.
This sounds as if it’s the company’s job to tell knowledge workers what works. Don’t knowledge workers prefer to learn from one another?
Maybe. I think a good learning program for knowledge workers would combine classroom learning and learning at their workstations. What we all want is just-in-time learning, where when we have a problem we’ll click on a learning program and it’ll tell us how to address that problem. But it’s hard to do and we’re a long way from it.
Knowledge workers have a lot of power, and you don’t want to impose things on them they don’t want to do, because they don’t like to be told what to do. They may put up with it for a while, but eventually they’ll look for a job that gives them the autonomy they think they deserve.
Besides, managers can’t easily enforce an order when work takes place in people’s heads. You have to make it easy for knowledge workers to do what you want them to do.
But some organizations are starting to mandate the use of productivity tools. Historically, pharmaceutical companies have asked scientists to use electronic lab notebooks, but never made it a requirement. Hence, they had no way to ensure that lab results were collected in a similar way across the organization. Infinity Pharmaceuticals, a company in Cambridge, Mass., has made using an electronic lab notebook a condition of employment.
You’d stand out if you didn’t answer your electronic mail or voice mail messages today—it’s semi-mandatory. We might as well make it a little more mandatory and help people use the stuff more effectively.
You’ve been using words like measure, process, mandating.
With a light touch.
Mandating with a light touch sounds like something of a contradiction in terms.
We have a choice here. We can get more productive with our knowledge work or we can lose our jobs. There are other parts of the world where people are very serious about being more productive, and are doing it for a lot less money than we charge. Twice as many Indian software providers are certified at level 5 on the Capability Maturity Model as U.S. companies. People should realize that unless they do knowledge work better, they’re not going to be doing it at all.
I have yet to meet a knowledge worker who isn’t interested in making him or herself better. Knowledge workers take pride in what they do, and they want to be productive.
And no one likes drudgery. Most of the interventions involve getting rid of work no one likes to do anyway. Ask people, “What do you think can be done to the job?” When you redesign knowledge work processes, it’s got to be much more participative than it was in these top-down reengineering efforts. People have to see what the benefit is to them.