Taguchi Explained… For Dummies

This is a reprint from my last newsletter issue. But so many people loved it and appreciated it, I decided to reprint it here.

First of all, I’m not a mathematician. And Dr. Genichi Taguchi’s work specifically relates to the manufacturing industry, particularly cars. Not direct marketing. His formulas were only recently extrapolated to the direct marketing and advertising industries, in the area of split-testing.

So my answer will be limited to split-testing alone. I’m going to attempt to explain this in layman’s terms and as best as I can. I hope not to lose you along the way… Fingers crossed.

What it is, is this…

Split-tests are normally based on two different versions of one ad. That’s why they’re often called “A/B Split Tests” (or “A/B Split Runs”). The object is to determine which ad — either version “A” or “B” of a salesletter, for example — pulls the greatest response.

Normally, direct marketers test one variable at a time: whether it’s the headline, lead, deck copy, price, premium, color, guarantee, or whatever.

Each one of these is called a “variable.”

In direct mail, most people send a small run of version “A” to a specific number of people, and version “B” (often, this is done simultaneously) to another group.

You then take the winning ad (i.e., the one with the highest conversion of leads to sales), which becomes the “control,” and use it to mail to the rest of the list.

Depending on how big your list is, you can do a variety of these split-tests before you determine the control and launch your winning ad. (As you can see, that can be limiting.)

Online, it’s a little more effective, because you can have a program that randomly pulls either one of these 2 versions with each visitor, as they see your ad, and have it calculate the results after reaching a specific sample size (i.e., a certain number of visitors and/or sales).

But here’s the dilemma — or the limitation, in other words.

Split-tests are linear in fashion.

With a salesletter (like mine, which are long copy, 5,000+ words salesletters), you can test one variable at a time to determine the winning version. And once you’re done, test a new variable in a subsequent test, and so on and so forth.

The object is to constantly beat your “control.”

With a long salesletter, you have hundreds of these possible “variables,” too. But that’s not all. You also have multiple variations of each variable — such as different versions of a headline. (Variations of one variable is called a “factor.”)

So in these cases, you can run a 3-, 5- or even 10-way split test.

An “A/B/C/D/E, etc” split-test, in other words.

But that’s not the problem. In traditional split-tests, you typically test one factor at a time (different versions of one variable). Each test is done individually. And sequentially.

The dilemma is, it’s possible that a variable in a subsequent test could have produced a higher response with the LOSER of a previous test.

Let me give you an example to illustrate.

For instance, in one typical split-test, you test 2 different versions on a headline. After so many hits, you determine that headline #1 outpulled headline #2. So naturally, you decide to use headline #1 in a subsequent split-test, which may be, say, testing 2 different prices.

Now, let’s say price #2 is the winner in that test. A question remains: what would happen if you tested prices with the loser from the previous split-test (that is, headline #2)? The response could have been greater. Maybe. Maybe not.

But the problem is, you don’t know.

Sure, you can go back and test prices with the loser. But what happens if you’re running multiple split-tests? What happens if you’re testing several variations in just one single split-test, let alone multiple ones?

If you run multiple split-tests, like 5, 10, even 100 tests, with factors as large as 5, 10, even 100 variations per variable, you can see how limiting such tests can be.

The potential permutations and combinations are vast.

That’s the flaw with traditional split-testing. It’s linear, and doesn’t test “combinations.”

This is where Taguchi optimization (and software based on it) comes in. The goal is to discover the best possible combination of factors that outpulls all possible combinations.

(I hope you’re still following me.)

Of course, it won’t determine this based on testing everything. (Not a lot of people can drive that number of traffic, anyway.) But with software nowadays, it can make fairly good predictions based on the science of probabilities.

The software can scientifically “guess” an optimal combination of variables (such as “A1/B2/C3” or “A2/B1/C1,” and so on)…

… Simultaneously.

This goes beyond a simple 2-way split test. What marketers like myself want to know is what combination of different variables will have the strongest probability of success.

(Of course, Taguchi is a little more scientific than this. There are many software in existence, like Optimost for example, which are robust and pretty expensive since they’re mostly geared for large corporations.)

But one such software “for the little guy” is Perry Marshall’s AdEvaluator.com.

As you know, I’m a fanatical tester. And I’m already running quite a few split-tests with it. So I can’t wait to see the results using this new software. I’ll be sharing my impressions and revealing some of my results to my members at The Copy Doctor site.

Now, if you’re not testing at all, please don’t fret with all this stuff.

For now, you’re best bet it at least start testing.

Last 5 Posts by Michel Fortin

• Social Media Isn’t Dead, But It Can Be Deadly
• How to Extract Doubt From Your Sales Copy
• It’s Not About Facebook, It’s About Integrity
• Did Larry Winget Go Overboard?
• The Oft-Confused Features And Benefits

About the Author