We often hear that companies must learn to embrace change. This is particularly true of companies that are applying advanced technologies to improve their competitive position. The full advantages of such technologies cannot simply be purchased off the shelf; they are won by patiently and carefully tailoring the technology to fit a given firm’s organizational and strategic context. At the same time, organizational skills, procedures, and assumptions within the firm need to be adapted to fit the new technology.1
Little is known, however, about how organizations actually go about modifying new process technologies, or how they adapt their own practices in response to technological change. Most of the research on this topic has assumed that users learn about and modify new technologies gradually. These assumptions have been built into our theories and images about technological adaptation — such as the familiar learning curve, which implies a highly regular accretion of improvements over time. The same assumptions are built into the prescriptions many researchers offer to management. These researchers exhort managers to “allow plenty of time” to digest new process technologies and to strive for “continuous improvement” (see Figure 1).
Yet most of the research on which these assumptions are based was performed at the aggregate level. Certainly, an entire firm or factory must strive for continuous improvement. But, at the level of a particular new technology, the process of learning about and modifying a new process may not be continuous at all. Indeed, our research suggests that the pattern of adaptation for an individual new technology is often a decidedly “lumpy” or episodic one (see Figure 2). In general, it appears that the introduction of a new technology into an operating environment triggers an initial burst of adaptive activity, as users explore the new technology and resolve unexpected problems. However, this activity is often short-lived, with effort and attention declining dramatically after the first few months of use. In effect, the technology, as well as the habits and assumptions surrounding it, tend to become “taken for granted” and built into standard operating procedures. This initiates a period of stability in which users focus attention more on regular production tasks than on further adaptation. Later on, users often refocus their attention on unresolved problems or new challenges, creating additional spurts of adaptive activity. In many cases, this episodic pattern continues over time, with brief periods of adaptation followed by longer periods of relatively stable use.
In this paper, we discuss the evidence for such an episodic process of adaptation. We draw on our own research in U.S. and European companies, as well as existing research on the practices of successful Japanese companies. After presenting this evidence, we also discuss why such an episodic pattern — provided it is understood and managed — may serve as an effective and powerful way to pursue ongoing improvement of new process technologies.
Explicating the Pattern of Adaptation over Time
In a recent research study, we investigated how three manufacturing and service organizations in the United States and Europe adapted new process technologies.2 The first site was BBA (names have been disguised), a leading manufacturer of precision metal components, where we studied forty-one projects involving the introduction of new capital equipment in European and U.S. factories. The second site was SCC, a multinational software developer of custom-built computer applications, where we followed the introduction of computer-aided software engineering (CASE) tools in three U.S. offices. The third site was Tech, a research university in the United States, where we examined modifications made to user-customizable computer tools such as text editors and electronic mail utilities.
Our main findings are consistent with the pattern described in Figure 2. First, we found that the installation of a new process technology was followed by an immediate and relatively brief burst of adaptation efforts. Thereafter, such efforts fell off precipitously. Thus, experimentation was more likely to occur and significant changes more apt to be implemented immediately following introduction than at any later time. This rapid fall-off of adaptive activity was apparently not a simple “learning curve” phenomenon, because it occurred even when outstanding problems had not been fully addressed.
However, the initial period was not the only time when important modifications were made. In each company, events sometimes triggered new episodes of intensive adaptation effort. These later episodes were also short-lived, but they were critical because they enabled users to tackle outstanding problems and to apply the additional insights gained through use over time. Thus, the cycle of intensive improvement followed by relatively stable operations tended to repeat itself.
The timing of adaptation at BBA illustrates this pattern. As shown in Figure 3, we found that most of the adaptations made were accomplished within a very short time after implementation — on average, 54 percent of all adaptive activity was completed in the first three months, or only 12 percent of the average total time to full integration. This pattern was remarkably consistent across all of the projects analyzed; the episode of adaptation that seemed to accompany initial implementation lasted about the same time (approximately three months) whether the project involved five people or fifty, and whether the technology was familiar or a departure from current procedures. Further, it was clear that adaptation efforts were not falling off simply because the users had resolved all problems within this period; on average, respondents reported five significant problems still outstanding at the time when initial adaptation efforts were curtailed. Indeed, most of the new technologies were not considered “production worthy” for many months.
Following the initial burst of activity, most of the technologies entered a phase of regular use as a part of the overall production process. On the other hand, participants did not completely ignore possible improvements to new technologies after the initial period of adaptation. In most projects, they regrouped and refo-cused attention on modifications some time later, again in a concentrated manner and for a short period (two to three months). Three-quarters of all projects at BBA showed a second spurt of adaptive activity. On average, this episode began about eleven months after the initial installation, and it accounted for an average of 23 percent of all reported adaptive activities. Further, in several of the projects, there was a third such spurt of adaptive activity about six to twelve months after the second episode.
Similar patterns emerged at the other two firms studied (see Table 1). At SCC, a large amount of adjustment and modification took place directly following initial installation of CASE tools into a new project site. In each project, the tools had to be fitted to the particular client organization. However, once application programmers (i.e., the users responsible for the actual production of new application software) began work on the project and started to use the CASE tools as process technology, further changes to the tools halted. These tool users required that their process technology be stable and reliable to facilitate production work. Thus, further refinement of the tools declined very sharply after the initial spurt of adaptive activity.
As at BBA, however, a significant surprise or major breakdown later in the project could turn users’ attention back to the need for ongoing improvements. At these times, technical support personnel were reassigned to undertake a new round of adaptations.
At Tech, too, users’ adaptation of their computer tools fell off abruptly soon after initial implementation. In particular, exploring or experimenting to learn about the technology virtually ceased after the first few weeks of use. Instead, users quickly settled on a computing environment and tried to maintain its stability. As one Tech employee explained, few people even thought about making changes once they had become comfortable with the software: “It’s just the way I do it. . . . It’s not that [further changes would be] hard, it’s just that it’s not worth the effort.” Yet most users at Tech (forty-nine of fifty-one) noted that specific events did occasionally refocus their attention on the software and trigger further customizations; thus, further adaptations occurred, clustered in relatively brief spurts that were interspersed with periods of routine use of the technology.
In short, all three of these very different organizations displayed a distinctly discontinuous pattern in the way they adapted new process technologies. Significantly, this did not seem to be a conscious management policy in any of the companies. To the contrary, managers (and users) frequently stated that they recognized the need for continuous ongoing changes to new technologies, but that it was difficult to keep people focused on this sort of modification activity for more than a short time. Thus, once users became familiar with a new technology, it tended to become a “taken for granted” part of normal operations.
The forces for stability and routinization, however, were occasionally disrupted by events that forced — or allowed — technology users to ask new questions and to reexamine old problems. Typically, the events that created additional opportunities for adaptation were new developments that somehow interrupted routine operations. At BBA, for instance, the reported new episodes of adaptation were generally associated with events that placed new demands on existing operations and also created a pause in the normal production schedule. For example, when new machines were added to the production line where the technology was in use, they often created increased demands for high-precision or high-speed processing that had not yet been achieved. At the same time, the installation of the new machines imposed a temporary line shutdown. Users in our study often took advantage of this time to address old problems and to initiate new adaptations to their technology. Similarly, the introduction of new products or product requirements, the imposition of new production procedures, or occasional breakdowns of the new technology were also times when the need for improvement became apparent, while providing a brief and sanctioned stop in the action. At Tech, the release of new versions of computer software forced users to interrupt their normal routines; these events accounted for almost one-third (28 percent) of all later episodes of adaptation observed there. Users at Tech also reported that they occasionally returned their attention to making software modifications when existing procedures became too frustrating, or when they were exposed to new ideas for making their routines more efficient.
Regularity or Pathology?
Several aspects of our findings are notable when compared to the conventional wisdom about technological adaptation. First, improvement is episodic, not continuous. That is, the initial burst of adaptation as well as later episodes are limited in duration, quickly giving way to longer periods of relatively routine operation. Second, some unusual event that interrupts normal productive operations typically triggers these periods of adaptation or at least triggers users to ask new questions. The contrary nature of these results raises a number of questions. Do these results suggest a new way of understanding the adaptation process, or are they simply the result of mismanagement at the companies we studied?
To answer these questions, we examined several detailed accounts of how some successful firms in other industries and nations absorb and modify new technologies. In particular, we were interested in Japanese firms that embrace “continuous improvement” in their overall operation. We asked whether the pattern of adaptations around a specific new technology in such firms is in fact gradual and continuous over time, or whether it reflects the episodic pattern that we observed in our data.
We discovered that the successful Japanese operations do not invite or expect continuous adaptations to specific new technologies. Instead we found a discontinuous model of adaptation that basically resembled our findings. An important difference, however, was that these Japanese firms appeared to consciously and carefully manage the timing of adaptations. Managers in these organizations apparently create and exploit the very episodic pattern that we have described. Specifically, the managers in these companies appear to do three things. First, they aggressively utilize the introduction period to adapt new technologies. That is, they identify and make the maximum number of modifications as early as possible. Then, following this period, they impose routine on the use of new technologies, and they exploit that routine for what it can teach them. Third, they consciously and periodically create new opportunities for further adaptations.
Next we describe these three aspects of technological management in more detail. We then present reasons why exploiting the episodic pattern of technological adaptation can be a particularly effective (and attainable) approach to learning and improving over time.
Aggressively Adapting the New Technology
The Japanese operations we examined truly exploit the initial period of technology introduction. They do not build in a great deal of extra time for debugging a new technology before moving into full production schedules. Rather, they develop very demanding, early production commitments for new technologies — and then they take steps to ensure that the new process technology will be ready. The ability to do this stems partly from the careful early design of new products and processes, well documented by Wheelwright and Clark.3 What is less widely understood is that meeting tight production deadlines in Japanese operations also stems from “the intensity of revisions on the spot during startup.”4 At Toyota, for instance, there is “a direct engineering assault to correct [problems] in the beginning, [which] prevents the need to dribble a constant stream of engineering changes through the formal system over a long period of time.”5
This early spurt of activity is often not apparent to outsiders. For example, U.S. observers who toured a successful Japanese electronics operation reported that “the Japanese can simply ‘flick a switch’ ” to start up new process technologies at high yield. In reality, successful introduction required three months of intensive, exhausting effort by a team of engineers. These engineers knew that there would be problems, but they also knew that they had to resolve the problems during the brief ramp-up period. Production commitments were absolutely firm, and the company’s profitability depended on meeting them.6
In the automobile industry, for example, Clark and Fujimoto describe the Japanese approach to managing initial ramp-up of new technologies as far more intense than normal problem-solving activities — in fact, they label it “the Japanese ‘war time’ approach.”7 They find that, while U.S. and European automobile manufacturers typically allow almost one year after the start of production to meet their target quality level on a new line, the Japanese allow only one to two months. The idea behind this Japanese strategy, explains Hall, is to “work through as many engineering changes as possible when a new model starts into production so that, at the close of start-up, it is ready to run smoothly.”8 This strategy is backed up with significant resource commitments; for example, at Toyota, “design engineers, manufacturing engineers, production control personnel, quality managers, and anyone else who is necessary live on the production floor during start-up. [Time losses are minimized because] engineering changes are made on the spot.”9
Imposing and Exploiting Routine
While many Japanese firms pursue continuous improvement, this does not mean that a newly implemented technology is subjected to a constant stream of changes. Instead, before any improvement is undertaken, “it is essential that the current standards be stabilized” and institutionalized.10 In the kaizen philosophy, “there can be no improvement where there are no standards.”11 This means that standards and measures must be imposed, “and it is management’s job to see that everyone works in accordance with the established standards.”12
Specifically, in the case of new process technologies, once the initial episode of debugging and modifying the technology is completed, standard operating routines are developed and enforced. Managers do not expect or allow operators to introduce modifications on an informal, everyday basis.
Jaikumar provides a vivid example; in his sample of flexible manufacturing system (FMS) users in Japan, operations were so smooth that production planning took only one hour per week and unexpected downtime was virtually nil. With such predictability in operations, there was no need and “no allowances for in-the-line, people-intensive adjustments.”13 Even the appropriate recovery routine for each identifiable failure mode had been codified and provided to operators.14
This does not mean that technology users in such firms neglect ongoing improvements — only that the changes they actually implement are tightly controlled. Operators in these firms perform considerable off-line experimentation, but they do not make unauthorized changes to production technology.15 Except for urgent corrections, they are permitted to make adaptations to the technology or production practices only at designated times.16
This emphasis on routinization is echoed in non-manufacturing operations. In Japan’s new software factories, there is rigid adherence to existing standards and little allowance for individuals to make changes during the production of new software systems.17 Managers in Hitachi’s software works, for example, realized that creating standardized procedures not only helped them to improve programming efficiency and productivity but also “helped them identify best practices within the company and within the industry for dissemination to all projects and departments.”18
Periodically Creating New Opportunities for Adaptation
Descriptions of technological adaptation in these Jap-anese firms reveal that, at the level of a given technology, continuous change really involves repeated cycles of change and stability. Imposing the discipline of routine procedures ensures that the timing of further changes is carefully managed. Whenever possible, the managers batch modifications together systematically and implement them in one intensive episode of adaptation.
Very often these episodes are timed to coincide with other major changes, such as product-model changeovers, releases of new software versions, or yearly factory shutdowns. Hall explains that the most efficient manufacturing companies “generally plan engineering changes one to two months before the effective date. The effective dates of most changes are timed to occur at schedule-change times.”19 Similarly, in Japanese software factories, development methods, tools, and products are generally held steady during a given project but are periodically and systematically upgraded or revised. A review of Hitachi’s software factory shows that Hitachi managers knew that the procedures they initially created would not be perfect; they “clearly recognized that these procedures and standards would have to evolve as personnel and technology changed, and they made provisions to revise performance standards annually.”20
In short, by creating distinct episodes of adaptation, managers provide the operation with both the benefits of routine and the vitality of ongoing change.
Why a Discontinuous Pattern of Change Can Be Effective
One way of interpreting the evidence from our research and the Japanese studies is to suggest that effective companies take the naturally “lumpy” pattern of adaptation and exploit it. That is, managers in the firms cited achieve maximum benefit from their adaptation efforts by carefully managing both spurts of adaptation and periods of routine operation. Indeed, there is evidence that such a discontinuous pattern of modification can yield important benefits. First, there appears to be a natural surge of energy at the start of projects, which smart managers exploit fully. Second, managers can enhance both learning and efficiency goals by imposing (and using) periods of routine operation in between periods of rapid change. And third, by revisiting the adaptation agenda at intervals, managers can make problems more tractable and can render change more attractive and more manageable. Next we discuss each of these issues.
Utilizing the Initial Window of Opportunity
The period immediately following initial introduction of a new technology provides a special “window of opportunity” for adaptation. At the start of the project, the level of energy is high. The novelty of the situation helps people to focus on the new technology and to see it as a distinct and malleable tool. As Weick argues, “The point at which technology is introduced is the point at which it is most susceptible to influence. Beginnings are of special importance because they constrain what is learned about the technology and how fast it is learned.”21
In the three organizations we studied, we noticed powerful organizational forces that underscored the importance of beginnings. In many cases, the motivation to change and to achieve targets was highest at the start of the introduction period. Challenging project objectives had a catalyzing effect at the start that often faded over time. For example, when one of the BBA plants began using an advanced precision grinding cell, the plant manager explained that “grinding all five faces [is] the key objective in this project.” Productivity improvements were viewed as less challenging and were simply assumed. Yet eighteen months later, when users still had not achieved five-face grinding, the motivating power of these objectives had faded. One engineer simply dismissed the earlier objective, commenting, “We only tried doing all five faces on this machine as an experiment. It was sort of an add-on that did not work.” Thus, in this case, as in others we observed, people slowly lost sight of their aggressive original objectives as they became accustomed to the level of performance actually achieved.
Another powerful factor is that modifications and improvements are easiest to implement at the beginning of the introduction effort. Successful implementation means that, over time, the technology becomes increasingly integrated into the production process. The new technology gets physically interconnected with the rest of the production process, and users learn to rely on it for production needs. Thus, after the initial period, further adaptation threatens to disrupt the physical flow of goods or services. Later modifications also threaten to destroy the routines and procedures for using the technology that users establish over time. For example, one engineer at BBA explained that it was hard to get production people to agree to further adaptation because “now the operators depend on the machine — it’s built in, they don’t want to change.” Even at Tech, where there were few system constraints on the changes that individuals made to their personal computing environments, users admitted that their own routines or habits tended to constrain further change. One user stated, “I got a set of custom [settings] from [a colleague] about four years ago. Now they’re ingrained.”
Another reason why beginnings are so potent is that, as time goes on, key project personnel often move to other assignments and are not available to help with continued fine-tuning and modification of the new technology. Even when team members are not reassigned, teams tend to lose enthusiasm. At a BBA plant in Germany, an engineering manager commented, “It’s easy to get plant engineers to start working on large projects, but it’s extremely difficult to keep attention focused on the details over time.”
Research on human behavior suggests that these tendencies are not just a function of mismanagement or short-term thinking in Western companies but rather are normal aspects of human task performance. Psychological studies have shown that people’s motivation to engage in effortful problem solving is partly a function of time: with extended exposure to a given phenomenon, people tend to become less alert and to notice fewer details than they do initially.22 Familiarity also makes people less willing to invest time and effort in difficult problem solving.23 Sustaining such motivation over long periods of time is difficult.
Finally, identifying and resolving problems during the initial introduction period is often easier because there are fewer competing demands on people’s time. Later on, production issues tend to dominate the list of priorities, simply because urgent problems have the power to remove attention from more important, but less urgent, issues. One project manager we interviewed stated, “The basic operating fact is that you need to produce good parts every day. [Even if] these people [the project team members] see a problem or get an idea and want to try it, some days you just can’t.”
Some project managers in our study imagined that they could continue to modify the technology gradually, “as we went along,” but found that production and adaptation did not mix well. One project engineer explained:
On this project, we tried to mix production and engineering work. But once we really got into production, time to do important engineering work was squeezed out by everyday work with the machine and operators. The sheer volume of work made it impossible to search very far for new solutions, or to examine and test ideas before they went on line. There was plenty of money, but no time — we only had Saturdays for testing new solutions. The result? Lots of grey hair!
The difficulty of simultaneously tackling production and adaptation was not unique to Western firms. One Japanese engineer faced with this problem is quoted as saying:
The production quota was high at that time. Even when team members gathered at the operation room on time to do an experiment, we were often kept waiting until the production quota was filled. Experiments often started at midnight. These circumstances deteriorated the efficiency of the work significantly. Consequently, it took longer to fix the problems than we had expected. This experience taught us the importance of finding the potential bottlenecks as soon as possible. The later the problems are found, the more it becomes difficult to solve them.24
Learning from Routine Use
Once an initial set of modifications has been identified and implemented, there are several reasons to hold the technology relatively constant for a period of time. First, there are obvious efficiency benefits in allowing the operation to run for some time in a stable fashion without frequent, disruptive changes. Hayes and Clark have shown that plants not subject to high levels of “confusion” from frequent changes in their tasks and technology have superior growth in productivity relative to “higher confusion” environments.25 In his study of Toyota, Hall argues that “dribbling a constant stream of changes” into the operating system can seriously compromise operating effectiveness.26
Second, routine operations are an important test bed; if constant and uncontrolled changes are being made, it will be very difficult to assess the effectiveness of previous modifications.27 Especially in “noisy,” complex environments, where signals are misleading or difficult to interpret, problem solvers need to observe the system over relatively long periods of time before defining further adaptations.28
Finally, periods of routine operation allow users to explore a new technology and to learn from their own reactions.29 Through extended use, they can identify features that are inconvenient or that cannot meet the evolving daily demands of a particular operating environment.30 Thus, as Imai explains, progress can occur only when one “institutionalizes [a given] improvement as a new practice to improve on.”31
Reopening the Window of Opportunity
There are several reasons why it makes sense to return to the adaptation agenda in short but intensive spurts, rather than to try for a more gradual pattern of change. First, short but intensive episodes of adaptive activity exploit “economies of scale” in problem solving. Many of the problems that affect a new technology require diverse resources. Managers, engineers, and users must create a cross-functional team, call in outside experts, set up experimental apparatus, develop prototypes, and so on. Frequently, regular operations must come to a halt. Gathering these resources only once to attack a number of issues is obviously more efficient than gathering them repeatedly as issues arise.
Changes are also less disruptive to ongoing operations if they are bundled together than if they are trickled out piecemeal to operations.32 Indeed, some problems simply cannot be solved — or cannot be solved effectively — unless they are examined as part of a complex set of interdependent issues. When problems are dealt with en masse, instead of one by one, these interdependencies can be identified and utilized.
There are also motivational economies of scale associated with brief, intensive spurts of adaptive activity. A team (or even an individual) may be more easily motivated to devote attention and energy when the goal is large and obviously significant (resolving many problems) than when it is small and apparently unimportant (solving just one issue). Similarly, the rewards and satisfaction that come from resolving a significant set of issues can motivate team and individual efforts.
Finally, episodic cycling between short spurts of change and longer periods of regular use makes it possible to revisit problems or issues as knowledge is gained with experience. Multiple cycles make it possible to respond to changes in the operating environment that occur long after initial implementation of a new system — whether these are exogenous developments or shifts in users’ preferences and expectations.33 When users move occasionally into an adaptation mode, problems that eluded their understanding during one phase can be reframed and perhaps approached more successfully during another phase.
Managing Attention and Effort over Time
These arguments suggest that a “lumpy” pattern of adaptation around a specific new technology may not be ineffective. Instead, relatively long periods of routine use, coupled with occasional but intensive episodes of adaptation, may be a powerful combination. Reaping the joint benefits of short episodes of adaptation and longer periods of regular use, however, is not automatic. It requires explicit management of the attention and effort applied in both phases.
Managing cycles of change and routine use of technologies demands a way of thinking that is unfamiliar to many managers. Users in all three of our research sites noted that managers very seldom took explicit action to create episodes of adaptation. For example, managers seldom intervened in a specific project to raise performance expectations, nor did they require regular project audits that might have focused attention on persistent performance shortfalls. Indeed, our interviews with managers showed that, while they were concerned with the way their organizations were introducing and using new process technologies, they were not sure how to manage the process. None of the people we interviewed seemed aware of the need to create or take advantage of discrete windows of opportunity for technological adaptation.
Our findings suggest that managers must consider how to create opportunities for adaptation, how to utilize those opportunities, and how to exploit periods of regular use of technologies for generating new insights and ideas. We cannot offer a recipe for doing this. However, we can suggest some ideas for moving toward a more conscious management of opportunities for technological improvement.
Creating Opportunities for Adaptation
Sometimes events outside of managers’ direct control create new challenges or expose users to new ideas. At BBA, the imposition of new products and additional equipment often created renewed opportunities to focus on problems with process technology. However, we found few cases where managers actively created new opportunities, even though they could have. For example, requesting post-project audits of new process technology is one way of helping users to refocus on original project objectives and to compare them with current operations. Managers can also precipitate “minicrises” within the operating environment that push users to stretch existing capabilities — for example, managers can declare a target of zero defects during a given week. Taking a different approach, managers can inject new resources for problem solving on a temporary basis. For example, during a period when orders are low, plant managers can call a “problem-solving day” when no production takes place, and each person suggests and works at implementing improvements.
Rotating people between assignments is another way to generate opportunities for change. We know that new project team members tend to bring fresh ideas and pose questions that existing team members have ceased asking.34 Additional team members also represent resources that are often needed to undertake modification activities. For example, we noticed that at BBA the addition of a new engineer into the factory environment sometimes triggered new episodes of adaptation. Likewise, existing team members who spend time in another setting or in a different function may return with fresh ideas to put into practice or a different perspective for assessing performance.
In creating windows of opportunity, it is important to realize that simply knowing that there are problems (or that better alternatives exist) is not enough. Somehow management must provide the incentive to undertake improvements, the resources necessary to accomplish this, and the chance to halt normal production rhythms for at least a short time. New product introductions, new quality requirements, or special assignments are events that often provide these — but not always. If operations are too chaotic, or if additional requirements place too great a burden on existing resources, it will be difficult to focus on potential new improvements at the same time.
Besides the specific actions taken, the way these actions are framed is also important. Employees may interpret a new or unexpected event as a threat (which produces insecurity and rigidity), when they could see it as a welcome opportunity for change.35 The words, rewards, and actions managers use all affect such framing. To recognize a new development as an opportunity, employees must believe that the situation holds the potential for meaningful gain, and that they will have access to the competencies and resources necessary to develop that potential.36 Consider, for example, the post-project audit. This can easily become just one more onerous reporting requirement, in which employees spend considerable effort justifying actions already taken. Yet audits can present opportunities for real exploration if managers frame them differently. Managers can ask users to skip the economic justification, asking them instead to identify three unfilled objectives and to outline plans for improvement.
An interesting question is whether new windows of opportunity must be surprising or unexpected events. Some organizations seem to turn very regular events, such as yearly plant shutdowns, new product introductions, scheduled technology upgrades, or maintenance reviews into successful episodes of adaptation.37 Yet there are also many cases in which the element of surprise is important for jolting people out of entrenched routines and assumptions. In a well-known study, Meyer showed how a potentially crippling doctors’ strike led some hospitals to undertake new organizational experiments, to restructure operations, and to redefine internal power relationships.38 Other examples are also common; an extreme case was the San Francisco earthquake in 1989. Following the quake, some computer users discovered they had lost the information they had saved on disks. Although experiencing a loss, they were also released from past constraints, and some used this opportunity to reconfigure their workspaces both electronically and physically. Whether such a disruptive event is necessary to create new opportunities for adaptation may depend on how deeply entrenched existing routines have become.
Exploiting Windows of Opportunity for Adaptation
At least three factors influence how an organization exploits opportunities for adaptation: its capability to act rapidly during a limited time window; the knowledge to select and undertake useful adaptations; and the choice of objectives to guide activities during this period.
The ability to act rapidly following a surprise, disruption, or halt in normal operations is critical if windows of opportunity are inherently brief. Yet this ability is rare; it requires that the operation be organized for fast response. One important aspect of fast response is what Bohn has called “information turn-around time” that is, the time required to collect data (by observing regular operations or by running special tests), analyze them, and make decisions about the next steps.39 If test results are obsolete before action is taken, then modifications will be both slow and often misdirected. One plant engineer we interviewed explained: “We could work on this [mold] to improve it, but we’d have to send it out to the lab [for evaluation after each trial], and the lab lead time is one week. You can’t develop a process like that! So, we just decided to consider this part ‘done’.”
Similarly, how support functions respond is critical to maximizing the limited time for adaptation. Very often, modifications require input from technicians, development engineers, systems analysts, programmers, maintenance personnel, and so on. If users find it difficult and time consuming to get the attention of support personnel and if weeks elapse before specialists respond to users’ requests, then the opportunities to undertake modifications will dissolve rapidly.
One of the Italian project teams we studied created an innovative solution to this problem. Plant managers recognized the need to respond quickly to the problems arising with a new process technology. They created a special workstation near the new process, called pronto intervento, where dedicated maintenance personnel and other resources (such as extra tooling and machine parts) were available to deal with problems immediately. Similarly, rapid response from external part or tool vendors, combined with smooth functioning of internal purchasing and receiving units, can be critical.
It is important to note that the organization’s capability to respond rapidly is highly systemic. If project deadlines are habitually allowed to lengthen because of outstanding problems, then key personnel will be perpetually busy and unable to assist with modifications. Worse, these commitments will be unpredictable, and the whole system will suffer increasingly from lags and unresponsiveness.40
This illustrates the self-reinforcing quality of rigid project deadlines and limited windows of opportunity: when project deadlines are respected, key personnel are free to make improvements in other parts of the operation. When episodes of adaptation are short but intense, such personnel can devote their full attention to their work, yet still be able to return to other projects in a short time.
Similarly, strict deadlines can help focus energy on needed adaptations. When tasks are bounded by definite deadlines, it often helps focus people’s attention and increases the chance that they will work actively on the problems. Especially when people are working in groups, deadlines help to keep individual efforts aligned and to maintain motivation.41 Awareness of tight time limits helps groups to assess their progress and to develop new approaches when existing ones are not working.42
Of course, deadlines can also become dysfunctional if they are unrealistic or too rigid to accommodate unanticipated contingencies. Remembering that new windows of opportunity can be opened in the future for further work may help managers set more reasonable deadlines and expectations for the period immediately following installation of a new technology.
Perhaps the most serious problem that distracts personnel from attending to technological adaptation is the need for continuous fire fighting just to maintain normal operations. When operations are out of control to begin with, introducing new (and advanced) technology only creates confusion and chaos.43 Similarly, introducing technology that is itself very immature can swamp the operation with crises and divert attention from genuine improvement efforts. One of the engineers involved in a disappointing project at BBA explained: “There were so many machine and quality problems at first that we had to change so many things. . . . Once we got into these problems, I couldn’t do any more [significant improvements], only try to attend to the little problems. The result was a lot of frustration.” The more that problems can be resolved before introducing the new technology, the more users can exploit windows of opportunity for making improvements, rather than simply for keeping their heads above water.44
Besides availability of physical and support resources, users need considerable technical capability if they are to take advantage of opportunities to improve operating technologies. Thus, technical training for users of new process technologies is critical. However, many conventional training programs ignore the fact that opportunities for learning, like technological adaptation, may also be episodic or cyclical over time. Traditionally, user training occurs before installation of a new process technology. Most systems development processes schedule user training as one of the last phases before handing off the system to the users. Certainly, some basic-level skills are needed at installation. However, just as all problems with a new technology do not show up immediately, so users cannot absorb all that they need to understand at the outset. As they gain experience with the new technology, they gain insights and increase their “absorptive capacity” for further formal training.45 Increased comfort with the new technology and a greater understanding of their own operating requirements allow users to better exploit educational opportunities on more advanced topics.
Moving away from the “one-shot” training approach would be particularly valuable in the case of software. Frequently, users simply cannot explore or appreciate the complex features and functionality without significant experience with the technology. One company we know has specifically implemented a two-tier training program for its personal computer users. The company does not regard the second tier as optional training, but as the continuation of the training course begun six months earlier.
Other users are another source of insight into how to adapt new process technologies. For instance, at Tech, other computer tool users suggested a number of software modifications.46 However, physical and organizational boundaries often prevent users from visiting other sites or from borrowing ideas from other users. This blocks new opportunities for adaptation and prevents informed choices.
Finally, the objectives organizations apply are perhaps the most critical determinant of whether organizations exploit opportunities for technological adaptation. Too often, managers view the period immediately following initial installation of a new process technology as the time to get the system up and running smoothly. Thus, the objective is to resolve problems that will interfere with full assimilation of the new technology. An alternative and more aggressive approach is to view this period as a time to surface as many problems as possible. The reason is that if problems are not identified early, it may be difficult to focus on (or even to recognize) them later. Yet such unresolved issues may compromise performance of the technology over the long term. With such an alternative approach, the objective of the initial startup period is not just a working technology, but also a new understanding of how the technology can be most fully exploited in a given operating environment.
Ogawa, who studied new technology introductions in the Japanese steel industry, argues that many managers hold a mistaken view of the test period.47 They assume that the purpose of this period is to enable operators to get accustomed to the new equipment, to set operating parameters, and to get product samples approved by customers. Ogawa suggests that the best managers also see the test period as a time to surface all major problems with the new technology. They reason that the new equipment should be placed under unusually severe demands as early as possible. Ogawa calls this the “rapid max” strategy. Under “rapid max,” the new system is quickly but temporarily brought up to maximum operating rates during the start-up and test period. The point with “rapid max” is not to reach stabilization rapidly, since this can lead to stabilization of suboptimal routines, or even to “false stabilization.” Rather, the objective is to experience rapidly a full range of issues and to uncover problems that would be difficult to deal with later.
Employing a “rapid max” strategy in new process start-ups would be a bold move for many managers. The approach employs the same counterintuitive logic as the kanban system, which tells managers to decrease inventory in order to make their problems more obvious. Indeed, if we recognize that experience brings with it inevitable forces for routinization, the “rapid max” strategy appears to be a powerful learning tool that can contribute to long-term effectiveness.
Exploiting Periods of Regular Use of Process Technology
We have argued that periods of sustained regular use can be complementary to episodes of adaptation. Periods of regular use provide data on how the technology is working, on whether previous changes are yielding positive results, and on what new problems or opportunities need to be addressed. Thus the key to utilizing periods of regular use is careful observation of operations and collection of data about them.
This is not a simple task. In many companies, users of technology perceive a conflict between production goals and data-gathering activities. One user at Tech remarked that she often runs into a certain problem with her software, but is typically “too busy” to carefully log the circumstances surrounding the error as a base for future analysis.
This indicates that managers should make data collection a part of users’ regular responsibilities. For example, Jaikumar reports that, in Japanese FMS installations, operators spend almost one-third of their time observing system behavior, examining statistics on system performance, or running tests to generate new data about the system.48
Managers could also make data gathering easier with a number of automated methods. For example, if software users are too busy to log errors and the circumstances surrounding them, an automatic “log-it” function could capture relevant data when problems occur. Or users could employ an electronic mail message system to send ideas and remarks about system problems to a central person, who could then collate and categorize comments. Additionally, automatic tracking of manufacturing operations (e.g., as part of statistical process control systems) gathers performance variations over time. Users could cull existing records of repairs, service calls, or engineering change orders for data on problems and opportunities. At the very least, users could keep individual electronic journals of observations or just a file of ideas (including things they like or do not like). Whatever the mechanism, useful data should include both well-structured information on predetermined topics (e.g., level of defects, nature of defects), as well as unstructured observations and ideas from users as they interact with the new technology.
Nonelectronic innovations can be equally useful. For example, one of our colleagues keeps a file of ideas from students or others on ways to improve his course. Instead of ignoring a suggestion because he is too busy to deal with it, he simply jots it down and files it. Once a year (when he revises his syllabus over the summer), he reviews the contents and acts on any ideas that have come up repeatedly.
Framing the data-gathering effort by setting appropriate objectives is also important. During periods of regular operation, managers need to help users focus on testing existing solutions, discovering unresolved problems, and increasing their understanding of the technology — not just on getting product out the door. This ensures that the next time users find an opportunity to focus on adaptations, they will be able to investigate problems at a deeper level and to tackle more challenging types of change. In this way, the repeated cycle of adaptation could offer the chance for something better than just ongoing modification. It could create opportunities to identify increasingly subtle problems or to set increasingly challenging objectives for the technology (and its users) over time.
We have described a central paradox that affects the implementation and later use of new process technologies. On the one hand, ongoing adaptation is an important success factor for implementing and using many new process technologies — and such adaptation takes time and experience. On the other hand, the more experience that users gain with a new technology, the more they rely on established routines and habits. Over time, their sharp focus on the technology as a separate and malleable object fades; thus, both the technology and the way it is used are eventually taken for granted.
Given these tendencies, is it possible to pursue ongoing improvement while enjoying the benefits of stable, routine operations? We suggest that it is, but that achieving both objectives requires careful management of time and attention. At the level of a specific technology, there are important benefits to applying adaptation efforts in an uneven, episodic manner, rather than on a gradual, continuous basis. The episodic pattern described here allows users to rely on stable production routines most of the time, but also provides discrete “windows of opportunity” to reexamine and change those routines. Short but intensive episodes of adaptation enable team members to devote their attention to adaptation efforts without undue distraction from — or interference with — ongoing operations. Such brief, intensive periods of adaptation also make it possible to exploit economies of scale by attending to many small problems at once.
Unfortunately, many companies neither recognize nor manage these episodic cycles of change and stability. Managers typically exhort employees to seek out problems and to pursue improvements on a continuous basis. Yet we suggest that managers could exploit a more discontinuous pattern to good effect by: (1) aggressively exploiting the opportunities for change that accompany the initial introduction of a new technology into the organization; (2) mining subsequent periods of regular, routine use for new data and new insights into technological problems and opportunities; and (3) periodically creating and utilizing new opportunities for further adaptation. This last point may be the most challenging. It involves both focusing users’ attention on the need for change and providing the resources and capabilities needed to act quickly — before the window of opportunity closes once again.