Everyone's talking about A/B testing; why you need it, how much it boosts profits by, and who does it best. It is a definite 'buzz' term in many internet marketing related industries, and none more so than web analytics.
The base concept behind A/B testing is that when conducting an improvement plan, you change only one aspect of a website at a time. This allows you to see exactly which changes made a positive difference, which did nothing, and which were actually bad for your customers. Today we give you a pocket-sized guide to A/B testing for website design and search engine optimization, the way that professional online marketing agencies do it.
The Principles of A/B Testing We explained the main principle behind the concept in the introduction. However, here are some other guidelines to keep I mind as you are setting up tests, examining results and deciding where to go from your results.
Do a null test Before changing your page variable, it is good practice to test your diversion mechanism by diverting a 50/50 flow of your traffic through the exact same pages - this is called a null test. It allows you to make sure that the technical aspect is set up correctly.
Change only one variable at a time A variable could be a color, a graphic, the copy for a headline, or a single positioning change.
Divert traffic for accurate numbers The traffic diversion process seems like a cut-and-dried part of A/B testing, but is actually one that can sabotage accurate numbers and therefore useful results. Use a high quality website analytics program, and get professional internet consultancy service advice to ensure that the numbers for each test are accurate.
Ensure statistical significance Check in with a web analytics expert to make sure that the volume of traffic directed through a changed page, and the percentage of different actions for that traffic, are statistically significant.
Use your segmentation tools to run long term tests Your web analytics provider can show you how to use the segmentation tools that are included with any good web analytics package. These tools will let you follow up on visitors and see what long term effect your changes had.
Setting up the test There are two main ways to set up an A/B test through any decent web analytics package. These include:
Replacing a single element before a page loads (done with simple elements like copy, small graphics, etc)
Redirecting your visitor to another page (done with larger changes like different colored themes)
Most of the rest of the process will be program-centric. It's best to consult with a web analytics provider that also has some expertise in conversion optimization and internet marketing, to ensure that the tools are being used in a market-friendly way.
There are a couple of caveats to be aware of when running A/B tests with web analytics, however. One of the most important is not to surprise regular visitors - use identifiers to ensure that only new visitors get the new version of a page. You should also run every test simultaneously - don't get your control figures while you are setting up the alternate version, and then test the changes afterwards - you need the traffic to be from the same period to be comparing apples with apples.
![[Valid RSS feed]](http://www.home-business-source.com/articles/images/icon_Rss.png "XML Feed For RSS"){kind=icon}
Category Rss Feed
Print This Article
Add To Favorites
