9 1 9 0 3

GM Report Executive Summary

This is the executive summary of a report on the cultivation and application of tacit knowledge I submitted to General Motors on November 6, 2001. The report was based on research for a Ph.D. dissertation that took place from February 1996 through March 1999. The GM organization has evolved in the years since. No doubt some of these points no longer apply.

I observed more than 400 meetings and interviewed more than three dozen people within design, marketing, and engineering over a three-year period to investigate two issues:

  • How, and how well, does GM's product development organization understand customers?
  • How, and how well, does it apply this understanding when developing new products?

Basic findings

1. Data will not get you there. While market research effectively gathers information on many topics, it cannot provide "the answer" concerning:

  • Customers' expressive needs (what the vehicle should "mean")
  • How these needs translate into product attributes
  • How to design the parts so they form a coherent whole

2. To address these issues, tacit knowledge (knowledge that cannot be proven or even verbally expressed; a.k.a. judgment, expertise) must be effectively cultivated and applied. No amount of market data, design criteria, or common process specificity can eliminate this imperative.

3. The frequency of job rotation (especially in VLE and VAPIR team positions) and the answer-seeking nature of most market research limit the cultivation of tacit knowledge.

4. The frequency and timing of job rotation, hand-offs, parochial incentives, and the involvement of isolated functional executives prevent the development of trust within program teams, between program teams and the functional organization, and among the functional organizations. Because trust is (justifiably) low, requests for data are common. This not only slows the process to a crawl but discards tacit knowledge, since by definition proposals based on this knowledge cannot be proven. Trust tends to be lowest when it is most necessary, during concept development, because teams tend to be formed at this stage.

5. The current level of "creative tension" makes nearly every product detail a potential battle. Only so many battles can be fought (they must be "picked"), so many details are not attended to from the perspective of the whole product. Product integrity suffers.

6. Most program teams do not sufficiently communicate the product concept and the story behind it to the functional organizations. Functional representative and "All People's Meetings" cannot adequately perform this task. Deprived of a context, programs' requests concerning this part and that part often do not make sense. Also, functional people need the context of the whole product to effectively apply their tacit knowledge. Finally, functional people cannot be expected to suboptimize their imperatives to benefit the "total car" when they have little or no idea what the product is supposed to be "when it grows up."

7. A bright spot: Extremely high levels of frustration indicate that most GM people care a great deal about the company and the success of its products. They fervently desire to hit home runs, and would hit them if the organization did not hinder them in so many ways.

8. Bottom line: The GM product development organization does not effectively cultivate and apply the tacit knowledge of its members. As a result, it cannot consistently hit home runs.


1. While conventional market research often seeks to replace judgment, alternative forms can and should be used to cultivate and inform judgment. Progress has recently been made here.

2. Reduced and more systematic job rotation, i.e. rotation that leverages and continues to develop an individual's knowledge and the relationships required to apply this knowledge. This can be supported by a "culture of expertise." If people were no longer prevented from developing their personal knowledge and using this knowledge to create home run products, they could derive satisfaction from these experiences rather than from "winning" against other functions and advancing up the hierarchy (where few can ultimately be satisfied).

3. Teams including knowledgeable people who trust one another must be in place at or soon after the beginning of concept development, or time must be allotted to allow trust to form.

4. Eliminate hand-offs such as those between planning and program teams and between advanced and production design studios to maintain knowledge of and commitment to the product concept and stabilize interpersonal relationships.

5. Commonize incentives across functions. This should make changes to the functionally-biased matrix reporting structure, which aids the development of expertise, unnecessary.

6. Program team members should directly communicate the product concept and the story behind it to working-level people within the functional organizations. This communication should take the form of dialogue with small groups. When everyone is working towards a shared, jointly owned vision of the "total car," they can do more work on their own, and far more details can be appropriately attended to and blended to form a coherent whole.

7. Place the authority to make decisions with the people who have the knowledge and relationships to best make and execute them. Develop trustworthy teams and then trust them.

8. Bottom line: Only through a product development organization and process fundamentally based on personal knowledge and interpersonal trust can GM create and apply knowledge on the scale necessary to consistently hit home runs.