-
-
- Is Taguchi testing a software
or a consulting
service?
It
has to be a consulting service. No software or product can optimize your marketing campaigns,
just as no calculator can design a car,
or no electric knife
can cook a dinner.
- How do you pick the factors for a Taguchi test?
We use the 80/20 rule:
we do a multi-variate testing of the 20% percent of the factors that represent
80% of the success. For the less influential factors we use split-testing.
- Is the list considered a variable?
Yes, but a different kind of variable than the components being optimized. Lists must be tested separately from the components.
- How viable is your system to markets under 15,000 prospects?
Depends on the medium. For example, an email optimization needs a minimum of 10,000 names, a direct mail optimization needs a minimum of 50,000 names,
a landing page
optimization
requires a minimum
of 1,500 visitors
per day.
- What if I do not want to test so many variables? How can I test fewer variables or use fewer tests?
We only test the most critical variables, applying the 80/20 principle described in Question # 2. For the non-critical variables (20% of the influence) we use split testing (A/B, A/B/C, etc).
- Can the Taguchi Method be used for Pay-Per-Click (PPC) advertising?
Absolutely. A PPC ad has 4 variables (headline, description 1, description 2 and URL) so we test 3 variations of each with 9 test ads to find the optimum ad.
- Is the method tied to the use of Taguchi arrays, or to fractional arrays, i.e. could other arrays (Central Composite, Box-Benkhen, etc) be used instead of the Taguchi arrays to create the ads?
Taguchi arrays are a subset of a group of testing arrays called "fractional". They are all "Latin Squares" (a
name given by Leonhard Euler, a Swiss mathematician that lived in the 18th century and discovered them) and are very particularly
arranged so that a number appears only once on each row and on each column, for example:
1 2 3
2 3 1
3 1 2
Note
that a number
doesn't repeat on
any single row or
column.
The
widely popular Sudoku game is based on
this property of all Latin Squares called "orthogonality" -
which is why Taguchi arrays are also called "orthogonal arrays". Dr. Genichi Taguchi discovered in 1946 that using orthogonal arrays for Design of Experiments reduces dramatically the number of tests. The rest is history.
Now the
answer to the
question: can you
use other arrays
beside the Taguchi
ones for marketing
optimization? Sure.
But we prefer Taguchi
arrays because of
their proven success.
- How do you measure success -
actions or sales?
Both. In our
projects we track
responses and conversions.
- How much variance is there among media format? For example, if you test 9 variations of a direct mail piece, how close would the results match if you tested the same 9 creatives in e-mail, or landing pages?
We found that there
is a 60-70%
similarity.
- Is there a
Taguchi software that you could recommend for someone to use to get more acquainted with the numerical portion of the process?
See
the answer to Question No. 1.
- How well does the Taguchi Method work for landing page optimization?
Very well. In a way landing pages are the easiest of all optimizations because there are no campaigns to be sent - just test pages to be rotated. That's why all our competitors do ONLY landing page optimizations (we are the only ones to
optimize email
campaigns, direct mail,
and printed ads).
-
Where do you place
yourself against
your competitors?
They do only landing pages, we do other media (see Question No. 11).
They charge monthly
fees and require
six months or one year
contracts; we charge
a
one-time fee.
|
