Innovation is the heart and soul of the new economy. The question isn't how can you do it once, or even a few times, but how can you do it continuously? In the IT field there is nothing more valuable than a continuous source of technical invention. And there is no more potent a symbol of continuous, groundbreaking innovation than the idea of a skunk works. If you can figure out how to create your own, you have a recipe for competitive and technological advantage.

The original Skunk Works (www.skunkworks.net) still exists, as a division of Lockheed Martin Corporation, where it began more than 50 years ago. The engineers and scientists there gained fame for their inventions that pushed the limits of aviation technology. And as they searched for an identity, an inspiration came from the Lil' Abner cartoon, where there was a "skonk works" in a secluded hollow with a still that always had something interesting cooking in it. The "skunk works" was born, and over time, it has evolved into a widely emulated business concept.

A skunk works is much more than an R&D lab. Its high degrees of secrecy, autonomy, talent and resources focus on breakthrough changes in how business is done. A lot of people talk about creating one, but doing it is a real challenge. Several years ago, I became a member of the leadership team in a large organisation with the understanding that I would help reinvent the way it worked from the ground up. It became clear to me that one part of my job would be to build a skunk works. There was no such existing capability, the expectations for change were very high and the organisational culture was too constraining for doing work out in the open. Here's my recipe for leading the creation of a skunk works.

Step 1: Sell it.

Before there is a skunk works, nobody thinks they need one. Everybody's got a different story: "We're already doing well enough with what we've got," or the well-worn, "We just can't afford to spend any more on R&D." So your first job is to sell it to someone who counts, near the top, who can help you grow and protect it. To do that, I made sure mine addressed a strategic problem or opportunity. The purpose of a skunk works is to solve real problems - not just dream up new ideas. In this case, we needed entirely new sources of knowledge and a reinvented value chain to keep up.

Step 2: Start it.

This is pure entrepreneurship. Find the office space. Get the supplies. Rough out a plan of attack. Get a few good people to commit. Start right away with a few seed ideas. Scrounge some seed funding or just steal the time. And above all, don't tell anyone. I started my skunk works with a project facade that hid not only an entirely new R&D effort but also a completely new way of working. I looked for a couple of early wins that would show my sponsor this could really work. The most important was when senior executives from our clients and business partners were introduced to our products and sent word that we were on a worthwhile new path. Then we started to get the breathing room we needed.

Step 3:Fund it.

Paradoxically, funding usually comes after you start. Funding creates visibility, which leads to constraints and expectations. Both can kill the skunk works in its infancy. The secret for me was to create a success that no one could argue with (my facade project - which, by the way, took two years to complete). That way, when I argued for more resources, anyone arguing for constraints would be arguing against a visible accomplishment. It's the only way to keep the strategic degrees of freedom you need.

Step 4: Staff it.

With an expanded resource base, you can start to go after high-quality people you couldn't previously attract or afford. You don't need many, just the right combination of competence, character and chemistry. I stole a couple of good people from around the company, attracted a few from the outside, and within six months we had tripled in size and were pursuing five projects rather than one.

Step 5: Protect it.

By the time you're taking on more people, getting more resources and showing some results, you'll also be attracting attention. Then the attacks will start.

Some will misunderstand and criticise an operation without an easily visible impact on the bottom line. Some will be jealous, fiercely competitive or merely threatened. Others will question your strategy. And the conservative ones will say the risks are too high. I faced all of these. Some things I could still keep secret. Some I could defend on my own. But most required the deft, consistent and intelligent support of my sponsor and the others who had been won over on the leadership team.

Step 6: Lead it and laugh about it.

Once the skunk works grows past its initial core, the culture will start to change. It will naturally become less freewheeling, less fun, less innovative.

It takes leadership to bring a healthy discipline to a growing organisation and at the same time keep its soul intact. Creative freedom is the essence of a skunk works. Don't let it get too large. Most important, make it an adventure.

Make it a crusade. Name the forces of evil you are fighting against. Revel in your victories. Laugh about your defeats and mishaps.

Step 7: Profit from it.

There is nothing like success, and every skunk works has to produce it.

Whenever you succeed, take the opportunity to tell the story of how this would not have been possible without the organisation behind it. Equally important, build the body of evidence that shows a return to the institution on the investment it has made. Once you're able to show that equation, you're almost there. In my case, by the time we entered our third year, even the most sceptical bystanders and adversaries admitted that our results were worth the effort.

Step 8: Sustain it.

People move on. Sponsors desert you. Leadership changes. Strategies shift. For a skunk works to live on and continue its contribution, it has to adapt and evolve. In most cases, this is not a function of process but one of people and identity. If enough of your people have the fire of innovation lit inside them, they will attract others and the organisation will stay alive. And if your group can create an identity - a brand, if you will - that is irrevocably matched in people's minds with innovation, then it is much harder to kill.

This recipe takes a bit of work, but it's worth it when you're finished. If you've never done it before, why not whip up one of your own? vChristopher Hoenig has been an entrepreneur, government executive (director for information management and technology issues at the GAO), consultant (McKinsey & Company) and inventor, and is author of a forthcoming book on problem-solving and leadership techniques. He is now chairman and CEO of Exolve in Washington, DC, focusing on next-generation Web-based problem solving