In part one of this two-part series, quality by design was discussed as a business problem involving successive gaps in new product introduction. A five-phase architecture was introduced. Part two looks at the last two elements of this architecture, customer-focused optimization and dominance over variability.
ADVERTISEMENT |
Customer-focused optimization
Quality by design starts and ends with the customer. Every new product introduction always has some amount of trade-off involved. If there are multiple customers, they may have conflicting needs. Even the same customer may have needs that compete with each other. Capacity and speed compete with cost of operation. Capacity can compete with speed. Flexibility and feature-rich offerings may have reduced ease of use, and so on.
Quality by design offers a range of tools and methods to make these trade-offs explicit and optimal for the customer. Some tools are highly mathematical, and others relate more to customer behavior. All focus on hitting the “sweet spot” that makes the right trade-offs from the customer view.
The role of creativity in product development is sometimes hotly debated. Creativity and innovation must be highly valued, and quality by design sets strong expectations for creative approaches to functional design, product features and goals, and production design. Because quality by design provides strong and systematic assurances that the final design will create delighted, loyal customers, creativity will have significant payoff. The risks of innovations that will not sell, or creative designs that will not work, are much lower when companies observe the principles of quality by design. The chances of a truly magnificent innovation succeeding are greatly increased within this structured environment that ensures the defect-free delivery of that great design.
Dominance over variation
Variation is everywhere. The relative priority of needs among customers varies. The performance of the final product will vary. The production process and materials have variation in them. Through most of modern economic history, both producers and consumers have simply endured and worked to correct the consequences of variation after it occurs. Even today, when we have a fuller understanding of the nature and consequences of variation, much new product introduction is still afflicted with it. Instead, designers should work to eliminate the seemingly inevitable variation that tends to plague development efforts.
Quality by design incorporates the most advanced modern tools to dominate variation rather than merely suffer and recover from its consequences. These tools and methods always begin by measuring and understanding the variation that exists. For new products, of course, there is an element of newness that may seem to limit our ability to measure variation from historical data. We can fill that void, however, from a number of sources:
Historical data on those elements that are carried over. Only rarely is something so new that there are no identical or at least substantially similar elements from which we can measure useful historical variation. If the component was not used in the immediate predecessor for the product being developed, it, or a similar part, it may have been used in a different product line, or even an entirely different application. Of course, care must be exercised in extrapolating from one application to another, but we still must consult such data as do exist to help set limits on our ignorance.
Testing and modeling. Most new products have some level of testing before their release—although the number that do not, especially in services, is truly staggering. Unfortunately, most testing uses very small samples, does not design its experiments efficiently, or covers such small ranges of use that variation could not possibly be measured adequately. While modeling has moved from static to dynamic in recent years, much of it still is not stochastic, and the impact of variation in the parameters is at best tested by sensitivity analysis without full multivariate stochastic understanding.
Historical data, testing, and modeling can be combined. Doing this helps to forecast, analyze, and eliminate the deleterious effects of variation using standard statistical techniques based on Poisson or other appropriate distributions.
Once we have measured and characterized the variation we face, quality by design offers a number of strategies to dominate that variation, each one supported by a number of tools and methods. These strategies include:
• Elimination of the sources of variation
• Robust designs that are insensitive to the input variations so are very stable on the output they deliver
• Tolerances that are derived statistically from actual process variation and the customer’s numeric critical-to-quality (CTQ) requirements
• Interventions or barriers that detect and defeat the variation when it occurs
• Mistake-proofing methods that make it impossible, or at least highly unlikely, that the variation can, in fact, occur
• Controls that monitor variation early in the process and intervene when it exceeds levels that will adversely affect the final outcome
A final word
Unfortunately, the discipline and methods required to ensure the highest quality and most reliable new-product introduction has received less attention than the methods and tools for removing defects after production begins. The initial attention to quality improvement of existing products and processes is the perfectly appropriate place to start. There is broad consensus on the outlines for good quality improvement, and the results are relatively easy to come by, although continued results require significant effort and dedication.
Quality by design, however, has suffered from a blizzard of different terms, acronyms, and labels. The application of these methods is a good deal more demanding, and implementing them is sometimes viewed as a frontal attack on the existing product development bureaucracy. As a result, its adoption has been much slower.
We can dispose of these barriers and move ahead. First, although we have offered a specific set of terms and definitions in this two-part article, one should not be wedded to any set. But one must insist on applying the methods and tools, no matter what the name.
Second, once an enterprise has mastered lean Six Sigma methods for improvement, it has a good solid base for the additional tools and methods. Black Belts in DMAIC with a modest amount of support and additional training can become highly effective facilitators or leaders for major new product introductions that follow the quality by design principles.
Finally, one need not completely replace an existing new product introduction process with an entirely new set of quality by design terms, methods, and tools. One of the most successful implementation strategies has been to lay the enterprise’s current new product introduction process alongside the full set of quality by design methods and tools, and then add to the current process those steps, methods, and tools that are missing, even changing names to fit the current culture as necessary. This should not be seen as diluting the quality by design principles, but rather as strengthening the current process. In this strategy, the biggest hurdle may well be the change in the way the product development department is staffed and led, but at least one is able to keep familiar terms and reports so long as they fit with the overall objective.
Add new comment