Small Business Marketing Guide - DIY (do-it-yourself) BRAND MARKETING STRATEGIES to HELP YOU GROW YOUR BUSINESS

Unlike nearly all other forms of marketing, direct marketing allows you the unique opportunity to test the success of not just the complete marketing effort, but discrete pieces of the marketing program individually. This is called split testing.

A split test is a head-to-head test between two different versions (or variables) of a specific element of a direct marketing program. For example, when split testing a direct mail piece, you would send out two versions of the same sales letter.

Both letters are almost identical, except that one includes a bold headline that asks the customer to order. Both include different website URLs to visit to place an order. Each of the website order pages look and act exactly the same. The purpose is to record which sales letter the customer responded to. This type of split test would allow you to discover which of the two sales letters was most effective, the one with the headline or the one without.

Once you have settled on the best sales letter, you could also split test other parts of same marketing program. You could send everyone the same sales letter, but in one letter ask that they call a telephone number to order, while in the other direct them to a website for ordering. You could also send everyone the same letter, but send some of them in red envelopes and others in white envelopes, or offer free shipping on some and not on others, or offer the product for $9.95 in one letter and $19.95 in another.

Common variables to test include:

• The written content of the letter
• The layout of the letter
• Headlines
• Calls to action (”Order Today!” or “Don’t Delay!”)
• Graphic elements
• Prices
• Different ways of selling the product (”buy one get one free” or “three for the price of two”)
• Shipping costs (free shipping or flat-rate shipping)
• Shipping methods (UPS or postal mail)
• Ways to order (web, telephone or postal mail)
• Different ways to pay (credit card or bill me later)

The larger the sample size, the more accurate the results. So if you only send out 10 test letters, your ability to predict the response to a million direct mail pieces is limited. Be sure test a large enough subset of your total list.

To make your tests even more accurate you should: randomly select your recipients, because new members of your list might respond differently from old members, and do your tests at the same time, because marketing received at the end of the month might have a different impact from something received in the middle of the month.

When split testing, not only is it very important to only test one variable at a time, it is equally important to have a method for accurately determining which version of the marketing program the customer responded to. For each version you need a separate response mechanism. These can be unique URLs, coupon codes, separate telephone numbers, or different post office boxes. As long as you’re able to accurately determine which version the customer is responding to, your split test will be a success.

When you have the results from your split test, carefully track and record as much data as possible. Be sure to to include the number of responses, the number of new customers, the average size of each order, and the average amount spent per order. It is important to track all of this data because you will sometimes be surprised when you compare the results. For instance, one test might result in fewer responses than the another, but with a much higher amount spent per order. Depending on your circumstances, this might be a very good thing. From your split tests, you can also determine which variables result in the lowest customer acquisition costs, which when doing a big expensive mailing could make or break you.

Split testing isn’t just for direct mail, either. You can split test other types of direct marketing including email, text messages and telemarketing calls. The opportunity for split testing in direct marketing is enormous, and very simple, as long as you remember to never change more than one variable during any one specific test.