Structuring and managing R&D work processes—Why bother?

Figure 1. On average, it takes about 3000 raw ideas...

The overall innovation process is not only risky but also complex (Figure 2). Managing this effort is neither easy nor simple. Attention must be paid to the organization, work processes, and activities of individuals while concentrating on the responsibilities of management. The key questions to be answered by managers approaching this task are summarized in Table 1 (3). At the center is the question related to structuring work processes: Are the work processes defined, disciplined, and structured?

Figure 2. The innovation process contains several steps...

Table 1. Some key questions to answer when managing innovation

Deschamps and Nayak (4) offer three general managing principles that must be kept in mind as this question is answered:

• Business performance tends to be determined by a few high-level, critical processes on which management attention should focus.
• A process cannot be executed well until every participant understands the entire process and is motivated to make it work well.
• A process that cannot be measured cannot be improved. These are important guidelines, but to which processes do they apply?

When processes are totally unstructured, they cannot be repeated. In a free-for-all atmosphere, researchers may call for greater clarity in direction and understanding. But where there is a high degree of structure, bureaucracy can restrict freedom of action, and lack of flexibility is introduced. One individual can stifle change, and policies and procedures may be ignored. Many may learn to work around the system, or the system may work against itself.

Should these principles be applied to collective and individual work processes equally? Some people develop highly organized, detailed, repeatable routines in their individual work; others do not. Debates arise when someone else dictates how an individual is to structure his work because it needs to be linked or coordinated with that of others, because someone has found a "best practice," or because experience has taught others some winning tricks or techniques.

Figure 3. The structure continuum.

Where does structure belong?
At DuPont, we have found that many of the steps in the innovation process can indeed be managed, but different degrees of structure or discipline are appropriate for each one. Figure 3 (below) shows where we place a number of processes on the structure continuum, but we are continuing to question and experiment with some additional structure for several of these processes. Here are a few to consider.

Technology planning. This process is interactive, iterative, and incremental. Imposing a detailed structure is ill advised, if not impossible. Mapping the process that really takes place is difficult, but we have found that certain aspects of the process as well as its outcome--the technology plan--can be characterized and the quality of the effort assessed. Several aspects of the process can be scrutinized.

• Involvement: What are the roles of those who do the planning?
• Timing: How often is the planning is done?
• Communications: How is the plan communicated?
• Auditing: How often is the implementation compared with the plan?
• Improvement: How is the planning process improved?

We have developed a guide to technology planning that can help manage and partially structure the process. Currently, planning is in the middle of the continuum, but we continue to develop requirements for longer-range technology planning that can introduce more structure.

Product development and commercialization. Introducing a high degree of structure into the product development process using "Product and Cycle Time Excellence" (PACE) (5) has been effective for DuPont. The structuring of PACE involves a product approval committee, a phase review process, a core development team linked to resources, special development tools and techniques, and a process "driver" (i.e., a PACE engineer).

Structured development under PACE consists of four hierarchical levels called phases, steps, tasks, and activities. Each level is carefully defined. A given project may have 3-6 phases, 15-20 steps, 200-500 tasks, and more than 1000 activities, all of which must be well understood by the direct participants.

Using this structured process, we give each development project all of the resources required. As a result, at any one time we have been able to work on only half the projects on our schedule. Cycle time, however, was reduced 40-60% for the average project, giving significant overall benefits. For example, in one business, the percent of total revenue from products developed over the past five years jumped from 15% to 75% after the structured process was implemented.

More than 100 firms are using this structured process developed by Pittiglio, Rabin, Todd, and McGrath, and many other firms are adopting similar stage-gate processes (6).

Discovery research. DuPont uses a much less structured approach in discovery research: the "Stine Model," named for DuPont's research head of 70 years ago. Charles Stine created a process in which broad areas for exploration were defined and high objectives were set, but researchers were left to develop their own detailed work plans. Back in 1926, when the company's growth began to level off, du Pont family members called for aggressive diversification into new businesses. Dr. Stine wrote to the company leaders (7), seeking funds for "pure science or fundamental research work." He argued that "the volume of fundamental work is rapidly losing ground as compared with the volume of applied research. In other words, applied research is facing a shortage of its principal raw materials." Stine fought for the funds and eventually got them.

Later he pointed out that "[f]undamental research is bound to result in the discovery of new, highly useful, and in many cases indispensable knowledge" (8), and he went on to propose general areas for such work: "Fundamental work in organic chemistry and in physical chemistry such as catalysis, colloids, and polymerization ... also certain physical-chemical work which will involve considerable physics." The latter area was to lead to major efforts in chemical engineering.

It is quite exciting to read Stine's original memo, which essentially launched the entire polymer industry. In addressing polymerization, Stine wrote, "In most cases very little is known about the actual mechanism of the changes that take place .... It is believed that a more thorough understanding of polymerization would permit us to develop oil and resin products and similar materials which would possess properties markedly superior to the properties of materials of this sort at present available" (8).

These insightful words were to lead to the hiring of Wallace Carothers, the invention of nylon, the creation of an entirely new industry, and DuPont's dedication to discovery research that continues to appear and reassert itself through the years (9).

We have now started to experiment with a bit more structure in this process, adapting some of the elements of PACE to include a review conducted by senior scientists and an external scientific review panel instead of a business team. Each stage in the development of the technology is studied and linked with an examination of the potential business implications if the technology development is successful. The final stage of this experimental process leads into the PACE process.

Technology networking. We cannot overlook the critical process of networking, which sustains product innovation. At Bell Labs, a study found that the "star" performers were differentiated from the rest by several key attributes. An important one was the ability to network (10). The extent of structure that is imposed on the human networking process will depend on the purpose of the networking and the kinds of networks involved (11). Structure will increase as one moves through the six purposes of networks: personal networking (discussing short-term, ad hoc topics); information sharing (sharing common subject matter); supportive interaction (providing assistance and a sense of belonging); collaborating (sharing information and work between pairs or small groups); synergistic networking (collaborating on broad issues, extensive learning); and implementing (doing work, getting results).

Structure also will increase as one moves through the three kinds of networks: content (interest in or focus on a common subject); situational (common role or common conditions); and business or organization process (work flow, functional, or capability-based, which may form the "virtual" organization, run business processes, and sometimes act as situational and content networks).

The more than 300 networks in DuPont have saved more than $200 million annually while serving as important technology transfer vehicles, responding to critical customer needs, and providing valuable sources of training and learning. We have found that results from the content or situational networks have grown in the absence of attempts to manage the networking process.

Competitive and technology intelligence, discontinuous innovation, and idea generation. We currently are reviewing these processes to determine whether added structure can be beneficial. But the question remains: Can further structuring and closely managing work processes improve R&D productivity? How we answer this question will depend on three things: how the business process is viewed, the nature and skills of individual participants, and the extent to which information technology is used. We can examine each of these elements.

Business processes can be viewed in different ways. The industrial engineering approach sees a business process as a work flow--a collection of activities that takes various inputs to create a valuable output. The inputs and outputs can be well defined, and the steps can be mapped. Relying too much on this view, however, can cause us to overlook processes with less well defined inputs and outputs (e.g., networking). A more general approach sees a process as a coordination of work--exploiting sets of skills and practices to create a special capability (12). Structuring work processes, therefore, is possible to the extent to which we can define inputs, outputs, steps and skills, practices, and desired capabilities.

The nature and skills of the participants in the work process can be a factor in determining whether adding structure will be effective. Psychologists have identified two contrasting styles of creative problem solving (13). They have measured the extent to which individuals seek to "make things better" or to "make things different" in their work.

Those who want to make things better have ideas that are seen by most as reasonable, practical, and inside the system. These ideas ensure the daily survival of the business and are readily accepted and given great weight. Such researchers handle detail work efficiently, making the best use of resources. They like the comfort and security of structure and guidelines that help to get things done. They promote stability and can foster strong effective teams.

Those who want to make things different have the potential to change the very nature of the business by questioning existing thought. Because they are seen as radical and outside the system, they have to work hard to get ideas accepted. They often prefer to leave the detail work to others while they move on to new initiatives. Such researchers will challenge the existing rules and guidelines, perhaps even ignoring those that get in the way. If the purpose of the work is to make things different, and individuals whose creative style is to make things different must be relied on, then the R&D manager will find that a high degree of structure can be counterproductive.

Using information technology often requires a disciplined work process. Work may be done in concert with the aid of computers, groupware, intranet Web sites, common databases, and related information technologies. Here the activities, inputs, outputs, and timing must be defined carefully for all involved. A greater degree of structure than otherwise might have been necessary is forced.

Figure 4. The structure hierarchy.

Next steps
The challenge is to examine each process or collection of processes that is important and consciously decide where it should reside in the structure hierarchy (Figure 4). In 1950, Crawford Greenewalt, then-president of DuPont, gave the prescription for managing R&D: "Provide a broad commercial base ...leverage research across diverse business lines. Pick your problems with great care and judgment. Be patient; don't evaluate results after one year, but wait as long as five or six. Know when to quit a problem. This is done largely by intuition, but we prefer to call it intelligence" (14).

Our efforts at structuring are simply attempts to bring together judgment, leverage, evaluation, and intelligence with our intuition.

REFERENCES