Why experiment? To improve your process, that's why. Reduce production costs, speed up manufacturing schedules, improve product quality, raise productivity and reduce scrap and rework -- you name it.

In a recent discussion about a furnace he was designing, a client said his customer had specified a custom furnace atmosphere to protect the product from oxidation. The atmosphere was mostly inert gas, which we both agreed would do the job, but the end-user also specified the addition of a highly flammable enriching gas, "just in case." Both the furnace builder and I hope the end-user will take the time to experiment with that atmosphere mixture to see if he can get by without the flammable component. It will make the furnace a lot easier and safer to operate. Once the line is in and running, though, it will probably be a hard sell. Any suggestions about manipulating the atmosphere will probably be met with one of these objections:

  • "It's working fine. Why mess with it?"

  • "I know all the experts and books say it can be done, but I just don't want to take a chance on it."

  • "We're too busy trying to meet ship ping schedules to do development with our production equipment."

  • Or, the flip side of that coin: "Things are pretty slow now. We're trying to avoid any unnecessary costs."

  • And then there's the unspoken one: "I got my company to spend a lot of money on that system. Do you think I'm crazy enough to tell themtheymay not really need it?" The last objection will be tough to counter -- the basic issue is job security, not process technology -- but the others should be easy to poke full of holes.

Why experiment? To improve your process, that's why.

And what do I mean by improve? Reduce production costs, speed up manufacturing schedules, improve product quality, raise productivity and reduce scrap and rework -- you name it. Too many people look at experimentation as a risk to customer goodwill. Actually, it can be the route to increased customer satisfaction, higher profitability for your company, or both.

I'll be honest -- I've seen some really ill-conceived (well, downright stupid) experiments carried out by some companies, and they turned into complete disasters. They don't invalidate the need to experiment; they just make a good case for thinking things out before you start.

When I first got out of college, I worked for a company making a lot of highly stressed aerospace components. The quality specifications stipulated that once product samples had been accepted by the customer, the manufacturing process parameters were frozen: We were forbidden from changing anything unless the customer tested and okayed a set of samples made by the modified process. You'd think such a strict, bureaucratic procedure would stifle process changes, yet we were constantly experimenting with new and (we hoped) better ways to make the products. Many -- well, okay, most -- of them worked, and we offered competitive pricing, reliable deliveries and a stunningly low scrap rate.

So, if you don't have a history of experimentation and want to give it a try, how do you get started? Follow these steps.

Assess Your Company's Tolerance for Risk. If they're paralyzed by the fear of screwing up, tread softly. Begin with things that have a high probability of success and little downside, so you can build confidence before going on to riskier ventures.

Identify a Valid Reason for the Experiment. For example, set lowering excessive costs at a certain process step, reducing the percentage of production lost to scrap, or breaking a production bottleneck as the goal. "Gee, wouldn't it be nice to try . . . ?" is not a good reason.

Estimate the Benefits of the Change the Experiment Is Supposed to Validate. In today's industry, nearly everything, even experiments, has costs assigned to it. You've got a better shot at selling the idea if you can attach a tangible benefit to it.

Plan Your Experiment Carefully. Expect the unexpected and have contingency plans in place.

Keep It Simple. Don't get distracted by side issues, and whatever you do, don't let your project get hijacked by people who want to tack on their pet ideas. The more complex and far-reaching the experiment becomes, the poorer its chances of success and the less likely you'll have a clear validation of your original idea. Anyone who has conducted laboratory experiments can tell you that the number of variables must be limited.

Enlist -- and Acknowledge -- the Help of Others. Want to torpedo your own experiment? Just try to hog all the glory. People who feel betrayed have ways of getting even.

If You're Successful, Toot Your Horn -- With Tact. Unfortunately, good works don't always draw attention to themselves -- they need a little publicity. And don't ever forget to give credit to all the others who helped make it a success.

Now, do you want a suggestion for a really risky experiment with your products and business? Just sit there and do nothing while your competitors are stealing the march on you in quality, productivity and customer satisfaction. Go for it!