When testing your site, it's vital to test just one thing at a time so you know exactly which change influenced the results. A method of testing known as "split testing" does just that. It allows you to split your audience into two groups and test, for example, one headline with one group and another headline with the second group--both at exactly the same time.
It's an ideal testing strategy for any online business, but especially for newer websites with less traffic, as it provides an excellent method for generating accurate test results no matter your level of traffic. Let's look at an example to illustrate how this works.
If you want to test elements of your sales letter, you need to use your top performing sales letter as a "control"--a benchmark against which all tests will be measured.
To test a new headline, for example, you need to create another web page that's identical to your top performing one but which includes the new headline. Using special testing software, you then redirect half your customers to the old page and the other half to the new one--over exactly the same time period. For example, if you have two customers named Bob and Jane, testing software ensures that Bob sees sales letter "A" at the same time Jane sees sales letter "B"--both on the same URL. Your other customers are likewise split evenly between the two different sales letters.
If you discover--through tracking how many visitors go to each page and how many sales each page makes--that the sales letter containing the new headline converts 15 percent more visitors into sales, you know that the new headline is more effective. Then you can test other elements of the sales letter one at a time and gradually fine-tune your sales process.
The software gives you the great advantage of testing the two different letters over the same time period. If you instead test one headline over a few weeks, then the other headline over the next few weeks, what if one of those test periods includes a major holiday like July 4th, while the other doesn't? Or what if a major TV event happens during one test period but not the other? Both of these factors could significantly skew your results if they resulted in fewer people being online during one of your tests.
It's easy to set up a few basic tests using software like the following:
- WebTrends 7 offers a range of testing services for a one-time cost or monthly fee, and you can road-test their services for free. WebTrends 7 also allows you to go beyond split-testing and start testing a few different things at one time. You can analyze exactly which combination of elements (for example, a particular headline and opt-in offer placement combination) works best.
- Like WebTrends 7, Offermatica allows you to conduct split testing and multivariable testing. Pricing is based on a per-visitor fee, and you can try their services out first through their free demo.
- ClickTracks allows you to track the response to your tests and understand where your visitors come from, as well as how they travel through your site. You can try their products for free, then pay a one-time or monthly fee depending on your needs.