1.5 Plan for Continuous Improvement
Improvement should be a continuous effort. It is easy to imagine the end of an SPI project, when you count your successes and failures. It's equally obvious that improvements are in some sense endless: As you alleviate some problems, others become visible. Both views make sense.
Organizing your SPI efforts as projects that have an end point lets you stop, step back, and evaluate your SPI initiative and its progress. When an SPI project ends, you can move responsibility for the new processes into the organization itself. Nonetheless, as Figure 1.5 shows, SPI initiatives are necessarily ongoing because there are always new problems and challenges, and solutions to old problems must be maintained and further developed.
Figure 1.5 SPI should be continuous
Examples from Practice
Although our experiences span only a few years, it is clear to us that a continuous approach is important. At Danske Data (Chapter 5), recent discussions have focused on how to organize the SPI effort in the coming years. Danske Data has long had a large methodology department. Initially, the SPI group attempted to dissociate itself from this department because it was not well regarded by software practitioners. However, the SPI effort has been gradually reorganized to include the methodology department so that it can fill the growing demand for maintaining new software processes. In some ways, the methodology department and the SPI initiative have merged. In other respects they are distinct initiatives. In any case, the SPI effort has now become a stable part of the organizational structure.
Another event at Danske Data supports this view. After two and a half years of SPI effort, a Bootstrap assessment showed that the company had reached its goal of maturity level 2. The CEO then decided to go after level 3. Evidently, the SPI effort will be ongoing for some time. A question remains, however, as to how far Danske Data will or should go on the conventional maturity ladder. Metaphorically speaking, it is not important to reach the summit. What matters is that you stay on an ascending path that is appropriate for your organization. Reaching one goal thus creates another, higher goal that has meaning for your processes and their context.
The key factors in a continuous approach are stepwise improvement, top-management commitment, and a sustainable improvement organization. Maturity models such as CMM, Bootstrap, and SPICE (Software Process Improvement and Capability Determination) embody the essence of stepwise improvement. All experiences suggest that such ladders take years to climb.
A continuous approach thus requires commitment from top management. A local or bottom-up initiative can be effective for a while, but it easily loses momentum. As time goes on, the focus on SPI initiatives will decrease unless they succeed in becoming part of the organization's strategic thrust.
To successfully climb the maturity ladder, you must install and maintain impeccable improvement processes. To do this, you need a sustainable improvement organization that is adaptable to changing circumstances. Management must directly allocate resources to such an organizational unit and must help it create and maintain organizationwide SPI awareness.
Factors that undermine the continuous approach are inadequate results and a marginal SPI effort. If results are not documented or the return on investment is poor, your SPI effort is unlikely to survive past the first project. In most cases, you cannot simply document improved maturity on a normative scale. Practical, convincing results are needed, and it is typically best to establish a simple metrics program that can help you argue for continuing your efforts (see Chapter 17).
Even when an SPI effort is showing results, organizational politics or other organizational issues can still marginalize it. Brüel & Kjær's SPI effort went from marginal to central and widely known and back to marginal (see Chapter 6). At this point, it looks as if Brüel & Kjær will most likely discontinue its six-year SPI effort.
The IDEAL model clearly expresses SPI's continuous nature (McFeeley 1996). The main theory underlying the continuous approach is Humphrey's CMM (Humphrey 1989). The CMM is specific to software development, but it shares a basic structure with Crosby's five-level model of quality systems (Crosby 1979). The idea of continuous, stepwise improvement is also common to many other quality models.
In his work on Soft Systems Methodology, Checkland has been particularly clear on how alleviating some problems makes way for others (Checkland and Scholes 1990). Effective problem-solving approaches are therefore continuous. Checkland and Scholes also distinguish between two types of problem solving: intervention and interaction. The interventionist mode is external and uses problem solving to structure an internal inquiry. The interaction mode is internal and uses problem solving to make sense of experience. The latter mode is similar to the continuous approach, whereas the interventionist mode corresponds more with a focused collaboration with SPI consultants over a limited time.
Theories of organizational learning view learning as continuously present and as a significant factor in organized work and behavior (see, for example, Argyris and Schön 1996, Senge 1990).