Issue #23 — Tuesday, September 11, 2018
Posted by Denny Hatch
DOWN-‘N’-DIRTY, QUICK-‘N’-CHEAP:
Bob Hacker’s Revolutionary New Testing Technique
By Robert C. Hacker
"A good plan
violently executed now is better
than a perfect plan executed next week."
—Gen. George S. Patton Jr.
You have a control package—one that has been
profitable for months or years. Results
from that package are in serious decline. Rising costs per order are
unsustainable.
You need a replacement—fast!
If you cannot come up with one or more new controls—quickly—you
will be out of business and frantically updating your résumé.
“There are two rules and two rules only for
successful direct marketing. Rule #1: Test everything. Rule #2: See Rule #1.”
—Malcolm Decker
Ed Mayer’s Corollary to Mal Decker’s Rules: “Don’t
test whispers.”
For example, DO NOT TEST $69.95 vs. $69.99.
Test serious differences: $49 vs. $69 vs. $89.
Traditional
testing methodology tells us:
“The Holy Grail of direct marketing is the single
variable test.”
—Don Nicholas
This means:
1) Only test one thing at a time.
E.g., Price, Offer, envelope teaser, or Letter copy
(with identical price and offer on all tests).
2) Always test head-to-head.
Identical mail packages with just one element changed
per panel.
3) Always test statistically valid samples.
This works fine when you mail in high volume and can
afford to allocate 5,000 to 10,000—or more—names to each test cell.
For example, a mailer dropping 1,000,000 direct mail
packages at a time (costing 70¢ each or $700,000 total) can afford to test 20
packages of 10,000 each and still mail the control package to 80% of the file.
But what if your entire mailing universe is
constrained to 25,000 to 50,000 names? This is a situation many
business-to-business—or geographically constrained—direct marketers find
themselves in.
An old rule of direct mail testing dictated:
The minimum for an individual test cell is 5,000 mail
pieces. The reason: back then the average response was 2%. Two percent of 5,000
is 100 inquiries or orders. This is the minimum for a statistically valid
sample to be used in follow-up efforts. This technique is:
Expensive: A series of 5,000-piece mailing
tests are expensive ($1 per piece or more in these minuscule quantities).
Time
Consuming: Writing, designing, printing and mailing takes three weeks.
Bulk snail mail takes at least an additional week-and-a-half to be delivered.
Plus an additional two weeks to receive all orders. Even if many responses come
immediately via 800 telephone and email, you’ve blown two months. Turnaround time for getting confirming tests
printed and mailed is two weeks with another week-and-a-half for all replies to
be in. That’s two months minimum to see if you still have a business!
You may
remember the famous scene from the movie. [In the winter of 1944] the Germans
launched Operation Wacht am Rhein, better known as the Battle of the Bulge.
Elements of the United States First Army are pinned down in Bastogne. British
Field Marshal Bernard Montgomery will need weeks before his troops will be able
to relieve the beleaguered Americans.
George C. Scott—second from left in the film, Patton—spars with
dubious allied officers over his promise to save save Bastogne
with two divisions in 48 hours.
Allied
supreme commander General Dwight Eisenhower asks Patton, “How long will it take
you to get Third Army moving north?”
“I can
attack with two divisions in 48 hours,” Patton replies, to a round of
snickering from the other generals present.
Patton’s
boss, General Omar Bradley, is not amused. “Ike wants a realistic estimate,
George. You’re in the middle of a fight now. It’s over a hundred miles to
Bastogne.”
In the next
scene, troops are marching north.
—SIDEWISE
THINKING:
Project Management and Business Creativity
from a Sidewise Perspective
Biographer
Martin Blumenson called this the penultimate moment of Patton’s career.
The
Penultimate Rule Breaker: Multivariate Tests
With
50,000 names to mail, conventional test methodology would limit the number of
test cells to 5-10, even if you tested the entire file. Often, that’s just not enough testing,
particularly if there is no control package and there is little or no time to
find one.
Or what if you have to create a new control and
you’ve only got one chance to do it? This is often the case for direct
marketing agencies, vendors and freelancers.
They get one shot. And month later they’re either heroes or zeroes.
When either of these situations presents itself, I
always run multivariate tests.
Often the cell sizes are not statistically
valid.
But, it’s still the best way I know to find a winner
fast.
Let me give you an example.
One of our prospective clients gave us the following
opportunity one September. Their current
direct mail cost-per-sale was about $700.
The target was $95 and no agency or vendor could hit it. If she couldn’t
hit the $95 target, her department was going away – and her with it.
If we could test in October and find a control
package that hit their cost-per-sale target, they would allocate $2,500,000 to
a rollout in January. Due to a very
tight time constraint, we had to find a winner on the first drop. We recommended a multivariate test, as it was
the only sure way to find a winner in time.
In this case,
the test matrix contained:
• 32 different copy/offer/package combinations
• 288 discrete test cells distributed over about
200,000 pieces of mail.
Did it work?
Yup, we hit the target on the first test.
We never stopped testing like this for them and drove
the cost-per-sale even lower. A few
years later we were mailing 3,000,000 pieces of mail per-day for this
client. It took six control packages to
get through the universe and we found them all with multivariate testing.
How to
Do a Multivariate Test
The objective of a multivariate test is to develop a
control package as fast as possible.
With a properly run test you can usually find a winner -- or several
winners -- in the first test, almost always by the second test event. Here’s how it works.
Phase 1
The objective of Phase 1 is to determine the key
success drivers as soon as possible.
Always test big ideas, not whispers to . . .
• Discover which lists will work.
• Determine which offers will work.
• Find a winning copy platform or structure.
• Identify a winning package format.
To show you how a multivariate test might work, let’s
look at a simple 25,000-piece test.
(This would be a very typical B2B test scenario.) The
key recommendations are to test…
• Three different offers
• Two package formats, Package X and Package Y
• Five lists of 5,000 each.
Based on the above test recommendations, the test
matrix would look like this:
In this matrix, there are 30 test cells, each cell
with 833 names. The results from each
cell are not statistically valid. To
generate a statistically valid sample, this test would have to be about 150,000
pieces, which is three times larger than the entire mailing universe of 50,000.
However, multivariate test results are still
highly indicative. We let the market tell us what they will respond to
instead of sitting around the conference table guessing. We move from the world of personal opinion to
professional judgment ASAP! You will see
strong trends and “hot zones” that tell you what to do next. Let’s assume that the test generated the
following response rates:
In this example, it is safe to conclude the
following, assuming we need a response rate of 3% or more to hit the economic
target.
• Package X is a clear winner over Package Y.
• Package Y was pretty good. Can it be tweaked to pull up performance and
then put it in rotation with Package X to slow down list fatigue? Typically,
we’d do that after Package X in in rollout.
• Offer B worked best with Lists 1, 2 and 3.
• Offer C worked best with List 4.
• Nothing worked with List 5.
Phase 2
In Phase 2 we check-test the results of Phase 1, this
time using cells that are large enough to generate a statistically valid sample
size. The matrix might look like this, assuming we found anew list (List 6)
that matches the profile of List 4.
With a response rate of 3% or higher and test cell
counts of 5,000, the results from this test would be statistically
projectable. And we’ve been able to find
a winning control package with only 50,000 names instead of the 150,000 that
would have been required for a test of this magnitude with traditional testing
methodology.
In this
world, speed is a strategy. When you are
in trouble, traditional test methodologies are too slow. If you don’t find the answer, somebody
else will.
So, when you have to find a winner fast, use
multivariate testing. If you have small
list universes, use multivariate testing.
The key to making a multivariate test work is to only test big ideas.
By big ideas we mean testing things that are very obvious to the
recipient.
If you look at a package for three seconds and can’t
see the big idea, you don’t have one. Typically, big ideas are:
• Offers
• Copy platforms
• Major headlines, Johnson boxes, eyebrows and
callouts
• Package design
• Package formats
• Outgoing envelope formats
• Outgoing envelope teaser copy
Many direct marketers today test whispers. It’s a waste of time and money. Don’t do it.
###
Word count: 1523
I have only one thing to add: memorize this superb lesson.
ReplyDeleteOh; and another: thanks.
Brilliant! Thank you!
ReplyDeleteGreat information. Only one thing: Although the principles of direct response advertising remain the same, their implementation in the world of social media, email marketing, etc. is different. I'd like to see this and other articles focus more on the contemporary marketing scene. Thanks...
ReplyDeleteThank you for writing. I agree! The great challenge to direct marketers today is know the the old rules and adapt them to the world of social media. Example: The 7 basic copy drivers—the emotional hot buttons that cause human beings to act are: fear - greed - guilt - anger - exclusivity - salvation - flattery. Whatever new or old media used, these copy drivers must be employed to be successful. Thank you for provocative comment.
DeleteExcellent article and worth readying. And I think this could apply easily to email.
ReplyDelete