AB testing likened to apples or pearsA/B testing is an established procedure to increase traffic, though it’s not as simple as it may seem. It’s more than just testing two or more versions of websites or ads.  While there is almost an endless amount of A/B tests for any given project, it also has some limits.  A/B testing should, by all means, be considered an actual experiment. It’s not just throwing two versions of a test at the wall to see which one sticks.

Starting up

When starting an A/B test, establish your control and your baseline for whatever you’re testing.  It is important also to only make one change to your baseline test versus your control.  More than one change makes it impossible to conclude which adjustment actually makes a difference.  Never assume a seemingly ambiguous change will bring ambiguous results. A simple change of fonts, colors, layout, images and their positions, borders, or design (to name just a few) can noticeably sway test results. A simple change to any one aspect can bring significant change.  This is the limit of A/B testing though; only one slight improvement can be made incrementally, one at a time, per testing period.

Getting a new website against A/B testing your current website


The fruit one can garner with ab testingWhen a business wants to overhaul their website, it would change the entire site all at the same time for that “fresh, new look.”  However I would argue that A/B testing is a more fruitful means of changing a website.  By changing everything simultaneously we cannot garner any information about visitors’ preferences.  On the other hand, by testing one variable at a time, we can find out what is most effective and have a tried and true site instead of just something new-looking.  While there is value to a quick new look and design, there is more value in actual results.

A/B testing is endless

Remember too, that A/B testing is vulnerable to recent trends.  Testing results can be affected much in the same way popular movie characters affect the most popular baby names.  Anything can set off a trend which means anything can affect testing results.  A good example is if a vacuum company has a tornado as its logo.  They would want to hide their logo if a destructive tornado makes a major news story.  The image would most likely have a negative effect on the website.  This is why it’s important to always be testing.  One must also be committed to true scientific experimentation. It is necessary to keep a log of the changes to specific variables to identify what has been tested against what and all of the results.  It is true that effective A/B testing is not for the easily annoyed or impatient. You can test almost everything about a page, but that would be a waste of time.  Stick to the more obvious changes as mentioned above.  The more testing that is done, the more expertise is gained.

This entry was posted on Thursday, June 17, 2010 and is filed under Pay-Per-Click.

Related Posts

Your thoughts? Please comment below.

One Comment