A Quality Advisor Book Review by Richard E. Biehl.
Copyright 1993, Data-Oriented Quality Solutions. All rights reserved.

PROCESS INNOVATION
Reengineering Work Through Information Technology
by Thomas H. Davenport, Harvard Business School Press, 1993

DEVELOPMENT EFFECTIVENESS
Strategies For IS Organizational Transition
by Vaughan Merlyn and John Parkinson, John Wiley & Sons, 1994


The mid-1990's finds the information technology function being expected to take on expanding roles in driving radical process improvement throughout the enterprise. The traditional role of automator and systems builder is giving way to an expectation that IT can somehow enable the enterprise to rethink, and reengineer, the major processes that drive the business. In Process Innovation, Thomas Davenport describes the enabling role that IT can play in making radical change possible throughout the organization. In Development Effectiveness, Vaughan Merlyn and John Parkinson describe the changes needed within the IT function in order to ensure that the necessary capabilities are available when called upon by the radical improvement movement.

Referring to Davenport's writing, Merlyn and Parkinson concur that "much of the thinking behind the drive to innovate or redesign business processes assumes that information technology will be a key form of leverage in the redesigned business process." They conclude that "there is little doubt that information systems will be asked to play an increasingly central role in the successful enterprises of the twenty-first century." [pg. 7]

Davenport builds his discussion around a presumed dichotomy between continuous incremental improvement and radical process innovation. His "experience suggests that companies can institutionalize incremental improvement through organizational and cultural change programs, with those doing the work identifying and implementing small changes in product and process." [23] He "believe(s) that the project or special initiative structure is the only way to accomplish radical innovation." [24] Rather than seeing incremental improvement and radical innovation as different quantities of the same basic activity, Davenport has defined them as fundamentally different activities. By stating that "only process innovation is intended to achieve radical business improvement," he simply assumes his conclusion: that major positive change throughout the enterprise will only be achieved through top-down specially organized innovation initiatives.

By relegating continuous worker-empowered improvement initiatives to the background, Davenport would presumably have workers sit idly by while management dictates changes from above. "Although the notion of conducting top-down innovation and bottom-up improvement at the same time might seem appealing, much of the process improvement activity is likely to be for naught, inasmuch as teams might strive to improve narrow processes that do not even exist in the new process design." [35]

By suggesting improvement-vs.-innovation as a fundamental dichotomy, Davenport is forced to acknowledge problems caused by such a logical split. "Ultimately, a major challenge in process innovation is making a successful transition to a continuous improvement environment. A company that does not institute continuous improvement after implementing process innovation is likely to revert to old ways of doing business." [25] The transition is only made necessary because a certain level of success is supposedly intended before-the-fact. While describing process maturity as a characteristic of high performance organizations, Merlyn and Parkinson view the order of improvement and innovation as reversed from Davenport. They state that "after a time, the cost of continuous improvement is not offset by the benefits of the improved process. The process is now a candidate for innovation before a new round of improvements can begin." [94] They see a continuous cycle of processes "undergoing continuous improvement", then "awaiting innovation", and finally "recently innovated and still stabilizing before improvement starts" again. [94]

If instead, one simply decides to make some process in the business better, the level of success achieved will dictate how we describe the effort after-the-fact. Those efforts that affect the process incrementally will be labeled improvements. Those efforts that achieve radical change will be labeled innovations. The intent will not be assigned based on the result.

Putting aside intention for the moment, Davenport states that "if the objective is incremental improvement, it is sufficient to work with many narrowly defined processes, as the risk of failure is relatively low, particularly if those responsible for improving a process are also responsible for managing and executing it. But when the objective is radical process change, a process must be defined as broadly as possible." [28] What Davenport lacks is a model that explains why the breadth of the target processes has a direct influence on the outcomes of the efforts to affect the processes.

When working to improve business processes, analysis must be conducted on the characteristics and attributes of the target processes. Davenport, Merlyn, and Parkinson all cite Information Engineering (IE) as a key discipline in this area. Under IE, one develops data flow diagrams to effectively describe the inputs to, and outputs from, key processes. Merlyn and Parkinson define a process as "the set of logically related activities that transform a defined input into a defined output. Processes have process owners and customers." [92] Lacking from this definition is that fact that processes also must have suppliers. One can integrate the principles of Total Quality Management (TQM) with IE such that each input or output becomes either a requirements or conformance flow as the web of processes becomes a chain of internal and external customers and suppliers communicating.

Every process plays dual roles as both customer and supplier in this chain. As customers, every process communicates requirements to some other process in order to receive the necessary conformance. A billing process sends an invoice (a requirement) to a customer in order to receive payment (the conformance). An order fulfillment process sends aggregate demand (a requirement) to manufacturing in order to receive available inventory (the conformance).

Much of the complexity of every process rests in this customer role. The fact is, when suppliers fail to conform to requirements, it is the customer process that carries the burden of managing the nonconformance. Billing processes must send late notices and produce accounts receivable aging reports. Order fulfillment processes must create and manage backorders when inventory isn't available. For decades we've seen in IT that these exception subprocesses consume almost half of our development resources, require the most complex code, involve the most testing, and produce the most ongoing change and enhancement requests over the life of software systems. Improvement efforts often focus on these areas. Without mentioning the customer-supplier relationship, Davenport agrees that "a key source of process benefit is improving hand-offs between functions." [28]

Using cost-of-quality as a model for understanding these subprocesses, the exception subprocesses under analysis can be viewed as nonconformance, or failure, costs. They represent the cost of managing a defect that has already occurred in the business, such as the invoice is past due, or the order quantities are unavailable. Investing in better invoice aging or more elaborate backorder processing is only making further investments in doing things we'd rather not be doing at all.

Cost-of-quality tells us that for each of these nonconformance costs, there exists a conformance option at a fraction of the cost to the enterprise of the current nonconformance processes. Instead of a new aging subsystem for invoices, look into better customer credit checking so that customers who don't pay don't get goods. Instead of better backorder management, invest in better materials requirements planning so that the right inventory is available when and where needed.

Quality professionals know that the conformance option is always a more positive option than the nonconformance alternative. The resources invested will always be less, and the investment is in doing things the enterprise wants to be doing. The catch is that, when analyzing a process, the nonconformance option is always within the process scope and the nonconformance option is never in the process scope. The path of least resistance will always shift resources toward enhancing the nonconformance option unless management adds the commitment and resources necessary to achieve the conformance alternative.

Cost of quality is the root cause of Davenport's apparent dichotomy. Those efforts that attack the conformance alternatives will involve cross-functional issues, need high-level sponsorship and commitment to cross organizational barriers, will achieve the most business impact, and will be labeled innovations after-the-fact. Those efforts that accept the status quo and attempt changes in the nonconformance alternatives will be workable by smaller isolated teams, rarely cross organizational boundaries, will achieve sporadic internalized benefits, and will be labeled incremental improvements after-the-fact.

The choice of the types of benefits to be achieved by an improvement effort rests in a process Davenport refers to as qualification, or "gaug(ing) the cultural and political climate of a target process." [33] Organizations simply can't handle ongoing continuous efforts to challenge the status quo. The cultural need to rest causes subsequent efforts to limit exposure to improving the nonconformance activities within processes resulting in only minimal or incremental improvements. After a year or so, organizations are again ready to tackle improvements based on conformance alternatives. This action-rest-action iteration drives the process maturity cycle described by Merlyn and Parkinson.

Beyond offering his five-stage approach to process innovation, Davenport's key contribution in Process Innovation is his insistence that organizational enablers always be combined with technology enablers in achieving radical process change. Merlyn and Parkinson concur early on with their belief "that it is the process and people implications of new technology that are typically short changed, both by IS management, and by the literature." [xxi] In Development Effectiveness, they offer a broad and thorough survey of issues facing IT management and staff in the late-1990's.

Davenport short-changes the quality profession by continually citing poorly implemented quality programs as reasons why IT needs to move beyond quality to more robust solutions in the future. Those of us in the quality profession recognize that more robust solutions have always been the point, and the poorly implemented programs along the way are part of the learning process. Merlyn and Parkinson concede that "some of the talk about quality is, admittedly, little more than hype. ... There is, however, overwhelming evidence that quality management can pay handsome dividends, and there is now a growing research into the business and social impact of TQM." [107-108] Because Davenport accepts a definition of quality that includes a "systematic, step-by-step approach to streamlining processes," [145] quality will never provide a path to innovation. "A systematic approach that never completely challenges the initial process structure presents little possibility of creative breakthrough or radical change." [146]

Notwithstanding these mostly semantic issues, Process Innovation and Development Effectiveness both make significant contributions to our current knowledge of how to leverage IT resources for the benefit of our entire enterprise. More research is needed to clarify all of the reasons why innovation is so difficult to achieve and how information technology can help or hinder the process. Until then, both of these books will serve as useful reference sources for the quality professional.