Split-testing, sometimes also called A/B testing, is the process of testing two or more variations of a webpage against each other in order to determine which one performs better. The goal of split testing is simply to find out how to structure your page so that you can get more clicks, more conversions, more sign-ups or whatever else may be the purpose of the site.

How Split Testing Works

Lets begin with a very simple example: We take a hypothetical campaign where we try to get people to sign up to a mailing-list. To run an A/B test, we create two different versions of our squeeze page, one with a blue colour-scheme and the other with a pink colour-scheme.

We will now send our traffic either to a page that serves up one version of the page or the other via some type of script or we will send the to a script that will redirect to one variation of the page or the other. There are different ways to accomplish this on a technical level, but whatever the solution, the result is that 50% of our visitors get to see the pink-themed page and 50% get to see the blue-themed page.

Now, we want to find out which version of our squeeze page gets more sign-ups. To do this, there are again several possibilities, but it always comes down to tracking the source of all our sign-ups or conversions.

Your auto-responder system will almost certainly feature some kind of tracking for this purpose. If you’re measuring click-throughs (to a sales-page or offer), you can track the conversion rates with a tracking system like Tracking202 or similar. You can also set up two separate redirects, one for each of the page-variations, but both pointing to the same sales/offer-page, to see which one of the redirects is used more often.

Again, there’s no shortage of tracking options, but it all comes down to this: Split your traffic evenly between two variations of a page and keep track of which one gets more clicks, sign-ups, sales or whatever it is you’re after.

--

What to Test

There are three basic rules to follow when doing split-testing:

1. Test only one variable at a time
   If you change several things on a page and get better conversions as a result, you won’t know what exactly caused the increase in conversions. Was it the new font? The different headline? The new pictures? The only way to know for sure is to only change one thing at a time.
   --
2. Test big differences first
   Don’t test tiny details as they will rarely make a difference (e.g. exclamation mark vs. period in the headline). Test large, significant variations first.
   --
3. Keep testing
   Test two variations of a page against each other. When you’ve found the winner, make another variation of that page and test it again. Repeat this process until no change you can think of still produces better conversions. The worst thing you can do in online marketing is not to test. The second worst is to stop testing to early.

Here are some examples of what you can test:

• Text vs. Video
  Have a text-based squeeze-page and test that against one where you read the text and record a slide-show or some other type of video. You could also test video vs. video and text.
  --
• Long vs. Short
  This goes for text as well as video: Test a long, extensive sales-page against a short one. Test a two-minute video against a 30-minute video.
  --
• Bullet-Points vs. Paragraphs
  Write out a few paragraphs highlighting the main benefits that come with your product or offer. Then test that against a brief description and bullet-points listing the main benefits.
  --
• Different Incentives
  Most squeeze-pages offer some type of incentive for signing up, like a free report or something along those lines. If you have more than one incentive at your disposal, test them against each other as well as one against several (remember that too many incentives can make you seem desperate). Also try giving your incentive a more compelling name and/or description.
  --
• Colour-Themes and Design
  Test a “naked” squeeze-page against one with more graphical elements. Test different colour-schemes against each other. Even fonts and background-colours can make a difference. At the end of the day, you have to make sure that your page is visually appealing to your target demographic otherwise they’ll leave before taking a closer look at it.

split-testing, testing, analysis

Last updated 808 days ago by ShaneRQR