After the launch of our new automated split-testing software, controversy soon followed.

Most of it was caused by the math behind it (which is complex and understandably hard to grasp for most people), or by competitors spreading misinformation.

So people were either confused by it or misunderstood it.

For example, some people posted in popular forums that my copy for MutiTrackGenerator.com is misleading because it’s not pure Taguchi.

(Well, technically, they’re right.)

Here’s why.

I cannot speak for Marc or Armand. But since I’ve had lengthy discussions with both, since the software was based on what I wanted (and I wanted Taguchi), and since I wrote the copy, I was compelled to explain a little further.

First of all, MultiTrack Generator is based on Taguchi and does use Orthagonal Arrays (i.e., the math that makes Taguchi work) in its algorithm. (A better term may be “inspired” by Taguchi, because it is indeed not entirely pure. But I will return to this.)

There is a reason why I used the term “better than Taguchi” in the copy. We don’t just throw around the word “Taguchi” for the mere sake of riding the buzz bandwagon.

(Believe me, it took over a year for Marc Quarles, the programmer, to learn, understand and program the math correctly into the software, as Taguchi is indeed very complex and advanced.)

Let’s clear up some misconceptions, shall we?

There’s a huge difference between “multi-variable” and “multi-variate” (or “multi-variant”). The former is sequential. It’s running multiple split-tests simultaneously — yet independently, because combinations are not observed.

Sure, it’s a step up from traditional A/B split-testing. But each test still requires a certain result to determine its accuracy and importance.

This is time-consuming.

What I prefer, and what MTG does, is the latter: multi-variant (or multi-variate) testing. And NOT multi-variable.

Multi-variate is either Taguchi-based or Taguchi-inspired. I say “inspired” because, while the concept is fundamentally the same as Taguchi, the application is different as it has been customized for the web.

Here’s why.

Mutlti-variate (or Taguchi-based) split-testing, in essence, uses Orthagonal Arrays to closely predict the best possible combination of variables with a smaller number of actions/results.

(In the case of a website copywriter, for example, it may be the best possible combination of, say, headline, background color, opening paragraph, bullet list, offer, price, guarantee, call to action, PSes at the end, etc, with the least number of visitors and the least number of sales each variable generates.)

Taguchi thus allows you to make a judgment call based on far less results/actions by statistically “guessing” which combination will provide the best conversion with only a handful of tests.

But Taguchi, by itself, has its flaws — particularly when applied to the Internet.

I’m not a mathematician, and I’m sure Marc can explain this better than I can. But here’s my “shotgun” explanation:

(By the way, this does not reflect MTG, or the views of Marc or Armand. They are solely mine and my opinions alone. Note that I have NO interest in this software. In fact, I didn’t even get paid for writing the copy. I did it for a friend.)

Taguchi is powerful math designed for the car manufacturing industry. Normal Taguchi methods only apply the calculations after the testing is completed. Which is really the only way you can do it — because you must produce an actual car that needs to be tested only once the product is physically devised.

(You can’t test “an idea” of a car.)

Taguchi thus allows you, using its complex math, to statistically determine the probability of success with a combination of car components using fewer test models (or, in this case, cars.)

If you’re using 12 different car components and you want to determine which combination of these components work best together, then you normally would have to produce, say, 12 x 12 (i.e., 144) test cars to actually test which combination is the best.

(Again, I’m no mathematician. But I think you understand my point.)

During WWII, Japan’s resources were scarce. So Genichi Taguchi came up with a way to save money by creating less tests models (i.e., less cars) by determining which outside influences exist (called “noise”), and to which degree they affect the end result.

That way, using Taguchi we can determine, in advance, which influences could be isolated and eliminated, and which ones would have an impact, and above all, to what degree.

Consequently, you can estimate the best combination of car components with far fewer cars, thereby saving you loads of time and manufacturing costs. Using my earlier example, rather than creating 144 cars, you only need to create, say, 10 or 8 or whatever.

And for the same reason, Taguchi works well in the case of direct mailings because the results are only determined once the mailings have been conducted and the test performed.

Similar to manufacturing, you can do a few tests, crunch your data, “Taguchinize” them and make an estimation on the best possible combination with a greater degree of accuracy.

But it’s not so with online salesletters.

The Internet is dynamic. Sure, Taguchi is great for ads, banners and small web pages. The difficulty lies in determining Orthagonal arrays for unknown numbers of variables per element.

And with online salesletters, for instance, the potential permutations are vastly greater than, say, a simple Google AdWords ad or a small web page.

With normal Taguchi, there is a distinct cycle of create-test-evaluate.

You have to do it this way because you must first conduct a limited number of tests and run your test results through Taguchi ONLY once those results are achieved.

With the Internet and especially with long copy salesletters, you have the ability to optimize your ad dynamically. That’s not all. You also have a far greater number of potential combinations.

And more important, outside influences can also change quite rapidly — sometimes dramatically.

That’s why the problem with current Taguchi-based software in their current state is that it’s not flexible enough to work with long copy salesletters as it can only compute a specific number of tests, and do so only after the tests are conducted and results tabulated.

Once this step is done, it gives you an importance level attributed with each factor so you can make a judgment call on the dependability and accuracy of the resulting favored factors. Then you move on to the next set of tests, if you wish. And so on.

So it’s quasi-sequential to a degree — not because of Taguchi but because of the rigor, demands, speed and complexity of long copy salesletters.

That’s why, even though Taguchi reduces the number of tests required and the time it takes to reach a certain result, current Taguchi-based software is simply not enough.

What MTG can do is determine the importance that the Taguchi function gives you while testing is still going on, and apply it to the elements as they are still being tested.

So for example, if the headline is way more important than the background color, then the aim is to favor the better headline for the remainder of the testing because it is the most important element.

If the system can estimate which one is best in combination with all the other variables and potential combinations, it should use that exclusively while the remainder of elements get tested.

So ultimately, the overall conversion ratio may be improved during the testing process. Not after.

That’s the beauty and the challenge of the Internet, and its ability (and also, I might add, the necessesity) to conduct real-time, dynamic testing — which Taguchi is not really meant to work with.

That’s why I said it’s “better than Taguchi.”

Or better said, it’s better than the way Taguchi is currently being applied. Because normal Taguchi is static and limiting. It’s NOT dynamic.

It was never meant to be.

About the Author

Michel Fortin is a direct response copywriter, author, speaker, consultant, and CEO of The Success Doctor, Inc. Visit his blog and signup free to get tested conversion strategies and response-boosting tips by email, along with blog updates, news, and more! Go now to http://www.michelfortin.com.

Last 5 Posts by Michel Fortin

• Why Some Marketers Are Psychics
• How I Used My Forum As a Marketing Tool
• Author of The Famous Seven Dirty Words Dies
• Apply The Law of Contrast to Build Desire
• This Plugin Increases Membership Conversions

Share This Post

Share this post with a friend by clicking "share this" below. You may freely reprint or redistribute this article, provided the content and links are left intact, and the "about the author" section is included. Get notified of new posts by RSS or email, below.

Other Related Posts

• Secrets of a 10% Conversion Rate (9)
• There’s More to Tracking than Conversion (13)
• Validate Your Profitable Headlines (3)
• The Real Sinister Side of Forced Continuity (74)
• Taguchi Explained… For Dummies (4)