THE CORPORATE CURMUDGEON

A measure of leadership: Keep trying

"The best bosses don't care if the glass is half full or half empty. They are asking, 'Why glass? What else could we make it out of? Can we sell ad space on the side? And what can we do with the empty half -- can we lease it out?' "

-- From a previous column arguing that great bosses are neither pessimists nor optimists, but "experimentists."

The trouble isn't being illogical; the trouble is in choosing which logic to follow. That was one of the great revelations of my college days, and it came, unlike most of the knowledge I picked up in college, in a classroom.

It was a sociology class, and the instructor asked us which socioeconomic group has the greatest likelihood of mental illness. Then he made the case for each. I'm summarizing here, but basically it was that the lower class had the poorest health care and nutrition, the upper class had the most health care and thus the greatest likelihood of being diagnosed, and the middle had a bit of both. The real answer was of less interest than showing us three logical solutions and pointing out that any of the answers looked obvious in retrospect, thus making the research seem slightly silly. If we'd merely been told the conclusion -- the lower class -- we'd have scoffed, "Of course!" Instead, we were left to ponder the limits of logic.

What got me thinking about the limits of logic was a new book, "Breakthrough Business Results With MVT," by Charles Holland and David Cochran. If you think that sounds a bit dry, then wait till I tell you that MVT stands for "multivariable testing." But, hey, the book is profound and lively reading. Really.

What multivariable testing does is allow a number of variables to be tested at once, rather than the dreary business of having a single-test variable and holding all others constant. For instance, when the folks at Lowe's home-improvement stores asked Holland's team to assess possible improvements in Lowe's advertising circulars, they came up with 29 items to test. By testing combinations of various ideas, they could figure out just what worked and what didn't.

For instance, the research concluded that a 48-page circular had just as much influence on sales as the 72-page circular, thus saving millions a year in printing. Also, while there were internal debates about the cartoon characters used in the advertising -- should they be rounder, male or female, dressed up or down? -- the research showed that using people worked better.

Finally, and this is the point in the book that brought to mind my sociology prof's lecture , which do you suppose worked better: Four products on the cover or eight?

You could argue that eight doubled the odds of customers' seeing something they wanted to buy, or that a clean, simple set of four was more likely to invite the reader to look inside. The research found for the latter logic.

Speaking of logic, we all know that testing and experimenting are good ideas -- no surprise there. But what makes Holland's work revolutionary is not the statistical techniques, but how they get to the point of applying those techniques. First, they invite everyone who "comes near the processes and wants to participate" to be included in the idea generation, and then they pare the idea list by finding the possibilities that are -- get this -- "practical, fast, and cost-free."

Holland's work fits neatly into the "experimentist" mind-set -- everything is up for testing. His firm has evaluated more than 150,000 ideas considered good enough to test, and the results show that about one-quarter make a positive contribution, half do nothing, and the other quarter have a negative impact.

The optimist doesn't want to test, preferring the gut feeling that it will work if we will it to. But the pessimist is correct -- the idea probably won't work. It's the experimentist who likes the odds -- the odds of learning something useful -- knowing that experiments never fail.

Dale Dauten is a syndicated columnist. He can be reached at dale@dauten.com.