• Perspectives on Safety
  • Published February 2017

In Conversation With… Amy C. Edmondson, PhD, AM


Editor's note: Dr. Edmondson is professor of Leadership and Management at Harvard Business School. Her latest research focuses on cross-boundary teaming in and between organizations and on the ways leaders enable the kinds of complex collaborations that such teaming generates. We spoke with her about new thinking about teamwork.

Dr. Robert M. Wachter: What got you interested in teamwork?

Dr. Amy C. Edmondson: What got me interested in teamwork was the discovery that it was how most work had to get done in modern organizations. When I say discovery, I mean insights and observations that guided some of my early work, initially in a consulting firm and then later as a researcher in organizations in several industries. I recall a dawning recognition of how profoundly people were interdependent in accomplishing nearly any significant goal. I, like many people, had an abstract mental model in my head that so long as people did their jobs and did them well, patients or customers would be taken care of. Of course, that's just not the case. It's a matter of people doing their jobs well working interdependently with each other; without recognizing this interdependence, people are unable to manage it well.

RW: Before we turn to health care, let's talk about other industries. Is this a big change? Is teaming and teamwork more important now than it was 50 years ago?

AE: Absolutely. Fifty years ago, it was far truer that the org chart was a good representation of how work got done. Meaning, those at the top of the organization (or the org chart) have a clear line of sight on what needs to be done, they divide that work up into responsibilities to be taken on by different functions or business units or geographies, then specific roles are described and people are hired for those roles. Roles come with task responsibilities, and managers help make sure those roles are doing what they're supposed to be doing by giving them feedback and evaluating their performance. That was a pretty good representation of how work got done in a relatively stable environment—with relatively stable technology, markets, customer preferences, and so on. In stable contexts, work and roles could be well designed and stay stable for a while.

This way of working started to break down over time, and more and more people realized that the most important activities in organizations, for example, any kind of innovation, customer service, or patient care, are necessarily cross-disciplinary and often cross-functional activities. This recognition led to the formation of cross-disciplinary teams, especially for activities where the need for different perspectives is clear. In new product development, for example, you need an engineering perspective, a marketing perspective, a manufacturing perspective, an accounting perspective, and so forth. Putting everyone on a team to figure it out together is far better than asking senior executives to work it out.

But today, increasingly, much of the work cannot be done even by a stable intact team because the tasks are too fluid; they're shifting constantly, and not enough is known at the outset of any given project to compose the "right team" for the project in advance. When responsibilities and situations are shifting, people still need to collaborate and coordinate across disciplinary lines, but they work in much more fluid team-like arrangements.

RW: Interesting. As you were speaking, I was imagining that the simple stuff that could be described through the org chart and is linear either was taken over by technology or sent to other countries. Is that right?

AE: There's a lot of truth to that. Much of the simpler stuff that could be automated got automated, or could be outsourced, did get outsourced. Then we're left with work that has more thinking, more judgment, more coordinating with other people in dynamic reciprocal ways to solve problems that come up on the fly.

RW: Take us through the life of a leader who 30 or 40 years ago was being appropriately autocratic and top-down. What kind of wall did that person hit whenever these trends began to play out?

AE: That's a great question. One way to look at what's happening in that scenario is that the leader's mental model of how things "should be" is no longer in line with reality. That mental model is not bad or good. It's just not useful anymore. If you're operating with a model that says, "I should have the answers" or one that says "When things go wrong, it's probably because someone didn't execute on the answers in the proper way," you are not going to be as effective as will be a leader holding a model in line with the new reality. Today's reality and task environment present customer needs in constant flux, needs that cannot be served by developing and executing on a blueprint. Instead, many people—working alone and together—need to use judgment to respond to new situations that they've never encountered before.

Today's leaders need to stop and challenge themselves when things go wrong. Instead of assuming someone didn't try very hard and that person needs to be reprimanded or better motivated, leaders have to recognize instead the high probability that something went wrong because the situation changed, or was new and challenging, or because a complex set of factors came together in a new way to call for an unprecedented response. This also means that whatever went wrong presented a learning opportunity, not simply a disappointment. In sum, your first, spontaneous causal diagnosis of the problem's cause is likely to be wrong, and doubling down on that diagnosis is almost certainly unhelpful, not to mention ineffective. Your mental model has to be updated.

RW: Stick with leaders for a second. Updating that mental model, what does that imply in terms of what makes a good leader or how one of those old leaders had to evolve, whether they could even evolve. Or do you need a whole new cadre of leaders?

AE: Most people are capable of changing and adopting new behaviors once they recognize the mismatch. Yes, some people have a hard time doing that, but for the moment let's go with the idea that most people can learn and change (rather than be replaced) and see what that might look like. Instead of the mental model "I'm the leader, I should have the answers, I should tell you what to do, you should do it, and then I'll let you know whether you did it well," the new mental model must be: "It's my responsibility to set the direction and to articulate that direction in a way that helps get people excited about contributing to it." Further, leaders need to tell themselves, "I need to be very clear with myself and others that I can set the direction, but I don't have all the answers for how to get there. In fact, in some cases I may have very few of the answers for how to get there. Therefore, I need to very much ask for your help." By you I mean you plural. More often than not, figuring out how to make progress is a team sport.

The new stance held by leaders is something like this: "This is what we're trying to accomplish for our customers (or, in health care, for our patients, or, in a business-to-business environment, for the companies we serve). I have some thoughts about how we might get started, and your expertise is going to be invaluable. There's no way to make progress without effective, frequent, transparent communication across disciplinary lines, or across geographic lines, or hierarchical lines (depending on the nature of the challenge). I'm setting direction, not providing answers. I may be suggesting a starting point, but it's not enough, not correct, not adequate. If we do our jobs right, this starting point will shift as more is learned. We will need to have two-way conversations—rather than one way (boss-to subordinate)."

Leaders need to do as much listening as talking. They need to listen for new ideas. They need to train themselves to become excited about the ideas that people will bring to the project, or to the work, instead of inadvertently discouraging judgment. They need to be explicit in communicating that judgment is needed by all.

RW: Is there some longing for the day of the strong leader who can be autocratic and just make stuff happen?

AE: Yes, because then the employee is off the hook and it's the boss's job to figure out what's needed. The employee's job is then just to do what he or she is told. People know how to do that. And, it's great, because when things go wrong, they can say, "It's not my fault." When employees believe that they should just show up at work, do their 8 hours, and then leave—because they believe the organization is asking them to leave their brain at the door—in fact, they will leave their brains at the door. But most adults really do want to bring their full adult selves to work. We want to contribute something important to a meaningful goal, and we want to be respected and valued as an adult.

RW: So if this all implies a different kind of leader who has different sensibilities and maybe is trained and selected differently, it sounds like it implies a different kind of worker too.

AE: Absolutely.

RW: So how is that playing out?

AE: Well, as you can readily imagine, it's playing out differently in different places. Some of the more wonderful success stories in health care and beyond are stories of workplaces where people are both invited and empowered to bring their full selves to work, to bring their brain, their voice, their hearts to work, to work with their colleagues to make things happen day in and day out in service of creating a better world.

RW: We've been talking generally about organizations. Let's shift to health care. It sounds like health care was not your first gig. There's a long history of people entering health care and saying it's probably just like GE or Alcoa or some other industry, and then shaking their heads and saying, not really, it's pretty different. What did you think was pretty similar and what did you think was quite different?

AE: What was similar was the ways in which people cared. Most places I've visited, most people really do care deeply about the work, the company, and the customers. In health care, it's usually even more the case. People know what's at stake, and they want to do the right thing. That was largely similar. There were two big differences: one, the incredible complexity of health care. I categorize health care delivery as a complex service operation, in contrast to a simple service operation where there is a high degree of commonality from one event or one customer to the next. Health care is necessarily profoundly customized. There are some similarities from patient to patient, but there are many more differences. Each patient is unique. And so, each care episode is to a certain degree unique. At the same time, there are also many processes that can be (and increasingly are being) standardized. Health care has made a lot of progress in recognizing the implications of this reality, but it was stunning to me to discover say in the early 1990s how much unnecessary variety I found in work processes, say from one floor of a hospital to another in a leading academic medical center, drugs would be labeled differently, carts would be organized differently, procedures would be managed differently. There was no apparent standardization, even around things that could be standardized.

RW: Right, I assume you saw that as a lesion and a flaw rather than the nature of the business.

AE: A huge lesion. I understood that it came from a professional craft mentality. Professionals are trained in their craft, which means they master a body of knowledge; they master expertise that requires judgement in its application. Thus, it's easy to start to believe everything's unique. I'll do what it takes for each and every one of my unique patients. Just leave it to me; I'll do what it takes (whether I'm a nurse or a physician). What I saw, however, was these wonderful well-trained experts were swimming upstream against an incessant tide of challenges, not all of which had to be there. Some of the challenges absolutely had to be there—but if you could standardize the routine (repeating) parts of the work, then the clinician's judgement was freed up to be more available for the unique and puzzling parts. If a critically ill patient shows up, that is a challenge no matter what. But, the ability to react according to a well-learned protocol that is invariant from one clinician to the next, is valuable for the organization, the patient, and the caregiver alike. One thing they shouldn't have to contend with is variation in how the organization labels or does things on this floor versus that floor.

RW: Are there any other industries that live in that box on your chart of complex service industries?

AE: There are other complex service operations. Health care does epitomize the category, however. Other examples include contract innovation services, like at IDEO, where different clients will come to the firm with vastly different needs. Those engagements are longer term encounters, compared with most patient-care encounters. Those projects will extend over months, not over minutes.

RW: And nobody dies if you don't get it right.

AE: And nobody dies if you don't get it right, absolutely. Perhaps this is a bad example, but think about Starbucks. In the early days of Starbucks, the value proposition was a better cup of coffee in a familiar (that is, standardized), pleasant environment. As the firm grew and as time went on, Starbucks found itself offering thousands of different combinations of drink and food orders, and so it actually did become a complex service operation. That created long queues and the company had to figure out how to re-standardize some elements of it. Because if every order is completely unique, it slows down the cycle time and makes it hard to deliver high quality service.

RW: Talk a little bit about the overarching org chart and management leadership culture you found in health care, because I'm sure that issue of standardization plays out in what the CEO does or doesn't do for a living.

AE: In the health care organizations I've studied, there are multiple org charts occupying the same physical space. So physicians have one org chart. They have department chiefs and attendings and house staff or interns and residents and fellows and so forth. They have their own hierarchy. It's meaningful. It sits there right next to the nursing hierarchy, and they both sit right next to the administrative hierarchy. Sometimes people in CEO roles are physicians or nurses; sometimes they're not. It's always been remarkable to me, compared with other or business organizations how hospitals have three coexisting, but sometimes not even coacting, hierarchies all trying to run the same place.

RW: Amazing, isn't it? You probably have heard the line a hundred times that they say about medical staff: What do you call a 99-to-1 vote in the medical staff? The answer is a tie.

AE: Right. So hierarchy matters. A lot!

RW: Yes, it's the idea that every physician has veto power over everything.

AE: Everything, exactly. It's good that expertise is in charge; that is not the problem. The problem is when that power inhibits others from speaking up with what they see, or with questions or concerns.

RW: Over time, organizations have generally progressed to a deeper understanding of the importance of teams and the fallacy of the org chart. In health care you have a unique overlay, the culture of physicians, the craftsman piece. Over the last 15 years, you have another overlay—the patient safety and quality movements. It says to health care: not only are you bankrupting the country but you produce a product that's not very good and sometimes harms people. As we've thought about other industries, a big part of the solution is thinking about systems rather than individual performance. How does that movement relate to the things we've been talking about?

AE: You're absolutely right. Part of the answer lies in your question. First, the challenges we were just discussing were bad enough, when we thought that we did our best and yes, sometimes our best wasn't good enough because of the vagaries of disease and so on. But to suddenly discover that even though we're doing our best, sometimes what went wrong wasn't the nature of the disease itself; instead we discover that we sometimes produced harm through poor coordination, poor decisions, or other errors such as mislabeling. Second, the recognition that when things go wrong, it's often a system breakdown, not individual human error or negligence, but rather a series of small things lining up in just the wrong way to create or let harm pass through. This is both daunting and empowering at the same time. It's daunting because systems are necessarily complex and if we have to worry about systems not just clinical knowledge, that's an extra challenge. But of course it's empowering, because we can get ourselves dislodged from that habit of just looking for whom to blame and to start thinking and working together to redesign the system or make small improvements to make it better.

The system lens is very important. Meanwhile, one concurrent trend is equally important to recognize. In addition to discovering the reality of harm and system breakdowns in health care, patients have become less likely to just show up in the clinic or hospital with an acute condition that could be treated episodically, sending them back to normal good health. Instead, more and more patients were coming in with chronic conditions that often brought multiple comorbidities and challenges. Ergo, their care was complicated and long lived. The medical professions had grown up learning how to treat acute events. But now they had to figure out how to work together across diverse skills and areas of expertise over longer periods of time to manage complex cases.

RW: Talk a little bit about technology. In health care over the last 5 to 7 years, we have become primarily a digital industry. As you've seen the technology wave flow through other industries, how does it change the nature of teams, the org chart, leadership, and all that?

AE: Technology can be the great leveler. In the old days, only the higher ups had roles that allowed them to know the important information, and that information was limited to some number of pieces of paper in a file drawer. In those days in health care, the medical chart was a file, a set of pieces of paper, and it lived in a certain place. Now, technology basically frees information, makes it almost infinitely shareable within the constraints of the law and ethics. But it means—both in and outside of health care—we cannot use as an excuse anymore that those at the top have the knowledge and they're going to tell others what to do. The knowledge can travel freely, and we must do that in thoughtful ways. It does change everything. Technology can make what people do far more transparent. Most people recognize that everything they do—online for example—is essentially living online forever. So it's tricky. On the one hand, technology is scary because it has such a long life, and on the other hand, it's empowering as a way of quickly transferring information where it needs to go to the point of care.

RW: Yes, but I could see it as moving back to the primary issue about teams versus hierarchy, if that's a real tension. At some level, technology could empower a team. It gives them the information. They can move it around more seamlessly and use it to improve. At another level, the boss has it and therefore it can be an enabler of more central control. More snooping might enable a return to the 1950s where I'm the boss and I know what's going on the floor in a way I didn't have access to before.

AE: When the employee was somewhat invisible.

RW: Right. So it sounds like both of those can happen.

AE: Yes, both of those can happen. So we need technology strategies. We need technology design. We need to be thinking carefully about technology. And one thing that we must avoid is having technology come between the clinician and the patients, and then we have to serve the technology rather than serve the patient.

RW: And that adds another layer because the team now more explicitly is going to include the patient, not as a passive recipient of care, and technology is an enabler for that as well.

AE: Right.

RW: As you've seen health care evolve and you've seen it up close and personal, are you optimistic or pessimistic? What do the trend lines tell you as we think about where this is going to go over the next 5 or 10 years?

AE: I'm optimistic. It is clearly challenging, but many thoughtful people are grappling with these issues, and trying to see the big picture in new ways and thereby to tackle pieces of it. A lot of good experiments are happening out there. Everything from medical homes to Accountable Care Organizations to even experimenting with different uses of patient portals. Researchers and clinicians alike are trying to find ways to engage patients in their own care. We also need to engage caregivers in new ways with each other and with the community to fundamentally change the game in population health. Not just sick care, but health care—focused on keeping people healthy to lower the care burden and improve the quality of people's lives. You can point to pockets of stunning success here and there, and these experiments make me optimistic. Also, the recognition that it will be an overwhelming crisis if we don't find new ways, is motivating. The cost trends cannot keep going the way they've been going. Health care cannot take over the entire economy!

RW: Are you optimistic about teams in health care? Are you seeing example of teams interacting in new ways?

AE: Yes. The industry culture is changing. And small changes can make a big difference. Melissa Valentine, a professor at Stanford, and I did a study of the implementation of "pods" in a major urban hospital emergency department. It was a simple intervention to improve the teamwork among different role groups, (nurses, attendings, residents, and others) with a reasonably simple structure that assigned patients and caregivers to pods. The waiting time went down by about 4 hours and the experience of the clinicians also improved markedly. So this was an example of a small intervention that made a substantial difference. I don't want to make it sound too easy, but I do want to emphasize that most people enjoy and feel good about healthy collaboration and coordination across role groups. They know that it's better for patients and that it's a better, more meaningful way to work. Part of this is coming up with good designs, and part of it is removing barriers to good collaboration.

Previous interviews can be heard by subscribing to the Podcast

Subscribe to the Perspectives on Safety Podcast now

View Related Interview

Back to Top