The head of a record label sets up structures, but he also defines the sound of the label, which is to describe what is desirable, what fits and what is quality for that label and then to create an environment where that sound can thrive.

You need judgment, you need to utilize conventional resource-allocation analysis, you have to work backward from estimations of the market to the current investments and you have to do some benchmarking of your product and its potential against your competition.

The manager's job - the impresario's job - is to preside over the company's efforts to jam so the business runs really well.

In a large pharmaceutical company, where it's a big bet, you're going to need finance people to be involved in the decision-making because the investment can run into the hundreds of millions of dollars. You're going to have to run scenarios. You might even need agreement from the C.E.O. to make that type of decision. If it's an incremental, low-cost decision in a marketing-oriented company, it may be a very different set of stakeholders a lot further down in the organization.

Ultimately, the impresario must know when to simply get out of the way.

There are some pretty obvious ways of benchmarking creativity. One way is to perform what I call a creativity audit, which is to look at your capabilities and look at your performance and examine the percentage of revenue that comes from products that are less than five years old, less than three years old and that are current with the present accounting period. You can then compare those figures to those of your competition along the same axes.

In a corporate context, companies have to try very hard to oppose the enticements of conventional wisdom. They must aim for the leaps, which means that companies have to do more than simply manage their knowledge, which is composed of the insights and understandings they already know. They also have to manage the knowledge-generation process. It's not just about, "Oh, we're going to create a data warehouse and we are going to invent a computerized filing system to get at all the stuff we know."