May 23 2005, Page 89
As a materials science engineer, James Kowalick spent 25 years
assembling bullets, bombs, rockets and nuclear warhead detonators
for the U.S. government and Aerojet, a defense contractor. Now he is
trying to use his expertise to add a little brisance to advertising.
The shattering power he has in mind comes not from shock ads but
from the sedate realm of multivariable testing.
Multivariable testing is a basic element of the engineer's tool bag.
You want to find out, say, what combination of catalyst, temperature
and pressure produces the highest yield in a reactor vessel. You
could find out by trying thousands of combinations, but what if each
experiment costs a lot of money?
The situation calls for a science known as experimental design, a
way to harvest the most data from a small crop of experiments. The
technique favored by Kowalick is called the Taguchi Method, created
just after World War II by Japanese electronics engineer Genichi
Taguchi. The method has been used in product design by many
companies, including Aerojet, Eastman Kodak and Ford Motor.
Kowalick, 66, thinks it's time to apply Taguchi to ads. What
combination of headline, copy and price information delivers the
most response from an ad? Experiments will tell you. Try different
direct-mail pieces or e-mail pitches to see which ones get results.
Madison Avenue types do not, by their nature, think like engineers.
Isn't copywriting more art than science? But they may come around.
"Creating ads to win awards will soon take a second seat to
profitability," Kowalick says.
Nordax Finans, an unsecured-loan lender in Sweden, is among the 22
companies that are shelling out up to $150,000 per project to
Kowalick Inc. in Sacramento. Nordax, one of the few Kowalick clients
that agreed to be identified, credits his system with boosting the
response rate to its direct-mail campaigns from 0.9% to 5.4%,
thereby adding $50 million since last year to its $150 million loan
portfolio. "This is really good stuff," says Mats Lagerqvist, Nordax
Dell Inc. tapped Kowalick last summer for an e-mail pitch. There
were 11 different ad elements, including product models, warranty
offers, discounts and prices. From the 10,368 possible combinations,
Taguchi's algorithm selected 18 that could best tease out the effect
of the different variables. Every customer in 18 test groups, each
with 2,000 people, got one of the randomly selected pitches.
Within a week Kowalick was able to analyze the effect of the
different variables on the response rate. One ad incorporating the
most compelling features - including a discount of 10% and a
hurry-offer-ends-soon call to action - was then created and sent to
150,000 consumers. Kowalick's data show the fraction of recipients
willing to open up the e-mail was 16%, three times the rate for a
similar e-mail barrage Dell sent to a control group.
Dell won't comment, but Kowalick says sales per ad from the
promotion - a total of $400,000 - were seven times better than in
the control group.
Results like these aren't making Kowalick rich - yet. Kowalick Inc.,
which the engineer founded in 2003 with Mario Fantoni, a former
management consultant to Oracle and EDS, is still a tiny enterprise.
Their company's 2004 revenue was just $1.5 million.
But Kowalick is so optimistic about the potential to sell the
Taguchi method to desperate marketers that he gave up a teaching
post at the California Institute of Technology, where he lectured
about the methodology. He also stepped down as president of the
Renaissance Leadership Institute, where he taught Taguchi to
Explosives, junk mail - Taguchi can be applied to just about any
optimization problem amenable to experimentation. At home Kowalick
applies the rules of multivariable testing to optimize the amounts
of garlic and tomatoes he puts in salsa.
Kowalick is attracting some supporters. "A lot of the ads that win
awards don't tend to move the business forward or make money," says
Kevin Wassong, president of Minyanville Publishing & Multimedia. He
met Kowalick when Wassong headed an online advertising group at New
York ad agency JWT (part of WPP Group).
But there are also challenges. Kowalick admits some agency
executives, namely creative types who are skeptical of ad tests,
have turned him away. Ad agencies "generally go with the most
creative ad of the moment," he sighs. "They see any additional
testing time, even if it is very short, as an unnecessary delay to
getting their ads out."
Some of the ads that have come out of the Kowalick process are
certainly creative, but they aren't award-show candidates. Kowalick
is a big fan of attention-getting images and headlines, or
"interrupters," that he can test, often bringing in another
optimization system, called TRIZ, an acronym for a Russian
expression meaning "theory of the solution of inventive problems."
TRIZ numerically rates consumer responses to images and headlines
based on a database of 24,000 ads. Kowalick uses those responses to
help clients pick advertising centerpieces that command attention.
For client TEMO Sunrooms of Clinton Township, Mich., Kowalick's
system helped winnow down 67 images and 282 headlines to the best 6.
A resulting four-page newspaper insert for the company in the
Detroit Free Press last year included the 6 combinations. One image
featured a "100% Carb Free!" headline over a picture of a sunroom.
In another a boy flies through the air in his family's sunroom. That
ad was duly ridiculed on NBC's Tonight Show With Jay Leno with the
line: "How long does this boy got to live?"
Denise Wahl, senior writer in the marketing department at TEMO, says
it received twice as many responses as it did from comparable
pre-Kowalick inserts. But she says she expected a threefold
increase. "This method didn't necessarily provide the extreme
results we had hoped for," she says.
Like many marketers, Kowalick tends to overpromise. He likes to
boast that some clients have had tenfold increases in response
rates, but he admits such claims make doubters even more skeptical.
To reassure them, Kowalick now says marketers pay only if they at
least double their current response rate. No doubt he would be
willing to put that ploy to the test. Let's see if it gets him any