How do companies pull away from the pack once their performance already matches best practice? In automotive supplies, for example, productivity in terms of parts per employee is much the same be it in Japan, the US, or Thailand. The same is true in automobile manufacturing, where it takes roughly 16 man-hours to build a small car whether it is assembled in Japan, the US, or a new factory in eastern Germany. Similar progress is being made in product development and sales processes, where benchmarking is leading to the convergence of standards at a single, high level.

The obvious way in which companies can still strive to differentiate their performance is by lowering factor costs, to which end many have already shifted production to low-cost countries. But another, neglected area is the improvement potential that still lies within existing technologies. To investigate just how much scope there is to improve the performance/price ratio of technological products and processes, McKinsey embarked on a research project with the Institute of Production Engineering and Machine Tools at the Technical University of Darmstadt (see the boxed insert, "About the research").

The results have been encouraging. Everyday technologies are frequently located at a fairly low point on their S-curve, when measured as performance relative to price.¹ In other words, there is ample evidence of huge improvement potential. Research projects and emerging best practice have shown, for example, that it is indeed possible to:

• manufacture air conditioners for cars using classic compressor technology with a failure rate of less than 50 parts per million, compared with the usual 3,000 to 5,000.
• design experimental combustion engines for cars capable of running 30 kilometers to the liter, instead of the usual 13 to 16.
• redesign and deliver overnight injection-molded plastic parts for car prototypes.

We also conducted some experiments of our own. In one such experiment, engineering students pulled apart two complex machine tools and started to redesign them from scratch in a simpler, less costly manner. Within a matter of hours, it was determined that the same functionality could theoretically be achieved using only 50 or so different parts, compared with approximately 485 different parts in the original design of one of the tools (and a total parts count of 1,900.) That’s potential, even when a technically and economically feasible target would be approximately 100 to 150 parts (see Exhibit 1). The number of parts in the second machine tool was reduced by the same proportion.

The research is not yet complete, but all results to date are essentially the same. "Core" costs (that is, the absolute minimum a product or process would require) are usually 15 to 30 percent of current cost levels; core times range from 10 to 20 percent of current process times; core quality levels for modules in terms of failure rates can be reduced to 50 parts per million; and core complexity can be reduced to approximately 10 percent in part counts. So, how can companies tap this sizable improvement potential? There are four steps:

Step 1. The most critical leverage point has to be identified in terms of either cost, time, quality, or complexity.

Step 2. A core level for the chosen parameter has to be calculated. The core time of a process, for example, would exclude all delays due to retooling, transportation, repairs, and so forth and focus simply on the necessary machining time. The core complexity level would be calculated by simply counting the number of parts that have to move against each other. Theoretically, everything that does not move relative to another part can be manufactured as one.

Step 3. An optimum target, derived from this core level, must be set. Optimum cost, time, or quality levels are likely to be fairly close to the core level. However, the parts count may be higher than the core level (usually by a factor of 3 to 5). This is due to the value customers give to differentiation, and the fact that it might only be possible to reduce complexity further by incurring extra engineering and manufacturing costs.

Step 4. A blueprint of how the product or process can be designed to reach this optimum has to be drawn up. It is not a question of rejigging the old design, but building up from the core level—a core-based redesign.

The first three steps can usually be accomplished in a matter of weeks. The last step before implementation—drawing up the blueprint—requires more time and imagination, but it can generally be accomplished in a month or two. Core-based redesign is, in itself, a quite straightforward process. The real difficulty lies in convincing staff that improvements of this magnitude are achievable.

When it is first mooted that core costs are typically only 15 to 30 percent of current levels, managers often express disbelief or are indignant at the suggestion that they are managing costs that badly! Providing examples of companies that have accomplished the "impossible," and encouraging senior managers to conduct their own core analysis will help convince the skeptical. Lower level managers, however, are still likely to show some resistance, anxious that such targets cannot be reached, and reluctant to be judged by them.

Here, a staged implementation plan can help. If, for example, a company that manufactures heavy equipment redesigns a processing line to reduce throughput time by 160 days (from 180 to 20), it might set an initial target of 80 days. If the design is any good, reaching the halfway mark should be manageable, and early success will erase fears and build confidence. Once the process is operating smoothly at this target level, the next one (usually another 50 percent cut) can be set.

Another way of breaking down implementation is to uncouple product from process redesign. Though no one designs a new product without considering how it can be produced, it sometimes makes sense to postpone the planning and implementation of the optimal production process. If, for example, the cost of failure is deemed too high because of the double risk of a new product and process, or if simultaneously implementing the two would demand too much from employees, simply realigning the existing process and concentrating on the new product design may turn out to be the better course of action for the time being.

The purpose of core-based redesign is not to downgrade the value of either benchmarking or continuous improvement; both have their place in the management toolbox. But in today’s environment, neither tool will help companies set new standards. Something more powerful is needed. The core-based redesign of products and processes can serve the purpose, for there is still a lot of room left for improvement in our basic technologies. The trick is believing in its power.