That would give you a total of 1 + 2 + 4 + 28 = 35 responses, yielding an overall response rate of 3.5%.

Now, let’s see what would happen if you send any of these letters randomly to all of your list without testing it first.

If you

Testing, Testing, and Testing…

But how do you test a direct sales or a fundraiser letter?

This is how I’d do it:

Have at least three different versions of the same letter and mail each to 10% of your total sample.

Then select the letter that has the maximum response rate and mail it to the remaining 70% of your list.

Doing so would give you a 66% chance of increasing your response rate.

How? Let me explain.

Let’s assume that you have A, B, and C versions of your basic letter.

You mail each to 10% of your mailing list and you receive 1%, 2% and 4% response, respectively.

If we assume that your total “sample population” consists of 1,000 names, this means you receive 1, 2 and 4 responses from each of your 10% mailings.

I’d take the letter which rang up a 4% return rate (Letter C) and mail it to the remaining 700 addresses, for 28 responses.

That would give you a total of 1 + 2 + 4 + 28 = 35 responses, yielding an overall response rate of 3.5%.

Now, let’s see what would happen if you send any of these letters randomly to all of your list without testing it first.

If you

This is how I’d do it:

Have at least three different versions of the same letter and mail each to 10% of your total sample.

Then select the letter that has the maximum response rate and mail it to the remaining 70% of your list.

Doing so would give you a 66% chance of increasing your response rate.

How? Let me explain.

Let’s assume that you have A, B, and C versions of your basic letter.

You mail each to 10% of your mailing list and you receive 1%, 2% and 4% response, respectively.

If we assume that your total “sample population” consists of 1,000 names, this means you receive 1, 2 and 4 responses from each of your 10% mailings.

I’d take the letter which rang up a 4% return rate (Letter C) and mail it to the remaining 700 addresses, for 28 responses.

That would give you a total of 1 + 2 + 4 + 28 = 35 responses, yielding an overall response rate of 3.5%.

Now, let’s see what would happen if you send any of these letters randomly to all of your list without testing it first.

If you

How? Let me explain.

Let’s assume that you have A, B, and C versions of your basic letter.

You mail each to 10% of your mailing list and you receive 1%, 2% and 4% response, respectively.

If we assume that your total “sample population” consists of 1,000 names, this means you receive 1, 2 and 4 responses from each of your 10% mailings.

I’d take the letter which rang up a 4% return rate (Letter C) and mail it to the remaining 700 addresses, for 28 responses.

That would give you a total of 1 + 2 + 4 + 28 = 35 responses, yielding an overall response rate of 3.5%.

Now, let’s see what would happen if you send any of these letters randomly to all of your list without testing it first.

If you

If we assume that your total “sample population” consists of 1,000 names, this means you receive 1, 2 and 4 responses from each of your 10% mailings.

I’d take the letter which rang up a 4% return rate (Letter C) and mail it to the remaining 700 addresses, for 28 responses.

That would give you a total of 1 + 2 + 4 + 28 = 35 responses, yielding an overall response rate of 3.5%.

Now, let’s see what would happen if you send any of these letters randomly to all of your list without testing it first.

If you

That would give you a total of 1 + 2 + 4 + 28 = 35 responses, yielding an overall response rate of 3.5%.

Now, let’s see what would happen if you send any of these letters randomly to all of your list without testing it first.

If you send letter A, you’ll receive 10 responses.

If you send letter B, you’ll receive 20 responses.

And if you get real lucky and send letter C, you’ll receive 40 responses.

Thus for 2 out 3, or 66% of the cases, your response rate will be LOWER than what you'll achieve by first testing all three letters on separate 10% samples.

And that’s why I claim that testing the three different versions of your sales copy on three 10% samples would mean a 66% chance of increasing your response rate over the no-test case.

The testing imperative becomes all the more important as more companies are marketing ever more specialized products to smaller niche segments of the population.

Test, re-test, and prosper.

----------------------------------------------------

(385 words, Copyright May 2006 Ugur Akinci)

by Ugur Akinci, Ph.D.
Creative Copywriter and Technical Communicator
writer111@gmail.com
www.writer111.com