Live Taguchi Optimization... part two: The levels
A couple days ago I posted the first installment of a live full service optimization of a client landing page. Well we are almost ready to go, but first I wanted to provide a little more information on the test design. Last time I mentioned the factor (variable) selection.
For this test, we are using an optimization type which we only make available for Full Service optimizations. In this case, it provides for up to 6 headlines, 3 images, 3 headers, 3 form placements, 3 greetings, 3 body copies, and 3 urgency appeals. This means there are over 4000+ combinations we will be able to look at!
One of the things I try to emphasize is that no matter what optimization system you are using, if you put garbage in, you are going to get garbage out. This means that a truly great creative writer can probably do more with A/B testing than an ordinary person could do with any of these advanced methodologies. So much of the ultimate success of an optimization campaign is controlled by the quality of the creative inputs.
Factor selections and level selections can be broadly divided into two categories- "creative" and "mechanical." Mechanical variables are things like font size, color, background color, image size, etc. Creative variables are those like headline copy, ugency type, or call to action. In this case, improved copy will be a key to the optimization. Looking over the factor list, most of the factor selections are of the "creative" type, rather than the "mechanics" type. There is one notable exception- The form position.
Here is our test design, flushed out with the level selections:
Headline
Image
Logo Header
Form Position
Lead In
Body Copy
Urgency Appeal
For this test, we are using an optimization type which we only make available for Full Service optimizations. In this case, it provides for up to 6 headlines, 3 images, 3 headers, 3 form placements, 3 greetings, 3 body copies, and 3 urgency appeals. This means there are over 4000+ combinations we will be able to look at!
One of the things I try to emphasize is that no matter what optimization system you are using, if you put garbage in, you are going to get garbage out. This means that a truly great creative writer can probably do more with A/B testing than an ordinary person could do with any of these advanced methodologies. So much of the ultimate success of an optimization campaign is controlled by the quality of the creative inputs.
Factor selections and level selections can be broadly divided into two categories- "creative" and "mechanical." Mechanical variables are things like font size, color, background color, image size, etc. Creative variables are those like headline copy, ugency type, or call to action. In this case, improved copy will be a key to the optimization. Looking over the factor list, most of the factor selections are of the "creative" type, rather than the "mechanics" type. There is one notable exception- The form position.
Here is our test design, flushed out with the level selections:
Headline
- Why watching this DVD could be worth an extra $10,000 a year
- Before your next performance review, make sure you've watched this
- How watching this DVD could keep you off the pink slip list... for good.
- How to get a raise on your own terms
- Starting -date- get paid what you're worth
- At last a surefire tactic to get a raise on -date-
Image
- DVD only
- Lifestyle- Woman
- Lifestyle- Couple
Logo Header
- UAC Logo only
- PB Logo only
- Both Logos
Form Position
- Right Side
- Left Side
- Bottom
Lead In
- Here is what...
- Dear Bookkeeping Professional...
- None
Body Copy
- CONTROL
- Show you first
- What you will learn first
Urgency Appeal
- Limited Time Offer
- Free for a limited time
- None
posted by Scott Miller at 4:51 PM
3 Comments:
Scott: What is the origin of the L18 variation you're using for this "full service" optimization? I haven't seen any mention of it in the usual Taguchi orthogonal array literature. Besides the obvious (you can test more headlines), what other advantages does it bring to the test?
Burke,
The L18 is simply another array. There are two main variations, one with 8 factors, and one with 7. It's main advantage is that it looks at over 4000 variations of end creative, instead of 2000 with the L12.
Scott
Scott: Thanks. I did some more googling, and it seems that the L18 variations probably originated from Dr. Kowalick's work with Dr. Taguchi as he applied DOE to advertising? Anyway, it's interesting to see how many ways the testing can be set up.
Post a Comment
Links to this post:
Create a Link
<< Home