Split Testing a Web Site

So you've decided to do some split testing on your website? What now? Here are some ideas:

1. Adding An Audio Greeting:

Audio greetings are like video. They can help or hurt depending on the quality of the audio. Audio greetings aren't as popular anymore, now that video has become so prevalent, but it's still often a nice touch. Something to test is whether to have the audio start immediately, or wait until the user presses "play". You can get some users angry if you start the message automatically, so you might use that approach only if it converts significantly better. If it only wins by a little bit, I'd advise you to wait for the visitor to start the audio. (important)

2. The Stylesheet:

If you use a style sheet to set the font styles, sizes, and colors of all your headlines, subheadlines, body text and more, you can test different schemes very easily. This is often worthwhile. If you link to your stylesheet, it can be tough to run the test, depending on the software you're using. So for the test you might want to pull the stylesheet into the head section of your test page.

3. The Layout:

Do you have one column or three? Do you have a long sales letter, or a short fact sheet. The general approach you take on your page can make a big difference. This might even be the most important choice to make. Unfortunately, when using multivariate testing software, it's often difficult to test this factor with other factors, because it pervades so much of the page. You can either test this factor separately, or run your multivariate test from a spreadsheet. Or, with a server-side PHP-based software program, you can test it with other factors with some clever work, though it will probably be much easier just to run a simple split test for this factor alone. (important).

4. Color Of Headline:

When Taguchi testing of landing pages first hit the scene hard in 2004, one of the examples people used most was of testing the color of the headline. Red was the best, it was said. But then there was a debate about whether a bright red (like hex #ff0000) was best, or a deeper red (like hex #cc0000) was better. Since then, I've tested red, blue, black, mixed red and black, green, and others, and I've found that the headline color isn't all that important. You can test it, but I wouldn't expect much. (not so important)

5. Text Of Your Order Button:

Should your order button say "submit"? Probably not. How about "click here?" Probably not. Well, what about "Yes! Let me Get Started Now!" Probably better. This is an often over-looked place to test. I would recommend testing it, though it's probably not quite as important as the headline or the offer. (medium importance)

6. How You Frame Your Price:

Do you apologize for charging $20, or do you sell it for $1,000, and make them think it's worth $10,000? If you frame too high, you might not be believable. Too low, and your product won't have much perceived value. Should you use a coupon code? When should you give it to them? At the top of the page? Should you let your affiliates pass on the coupon code? Price framing is a very important art, that you can develop through testing. (important)

About the Author:

Jim Stone, Ph.D. Created the Split Test Accelerator, and is an expert at using Taguchi testing to improve landing pages. Visit STA to learn more, or visit this page for more split testing ideas.