If you read Google’s official guide on performing A/B testing for your AdSense ads earlier this month, you’d probably already have a strong sense of what split testing is. But if that has been your only exposure to it, then I’m afraid that there’s a chance you might have been fooled to believe that split testing is done only for adverts, or just AdSense in particular.

Fundamentally, a split test is a means to statistically test multiple variants of the same thing against each other. For bloggers, that “thing” might be your ads, affiliate offers, headline colour, navigation, and more. And an A/B test is the most basic (and perhaps most useful for most bloggers) form of split testing as it pits only two variants to see which one is better. There are more advanced versions, e.g. multi-level tests and Taguchi (multivariate) testing, but most of us won’t need it (yes, this is also a good excuse since I barely understand them).

But as important as theory is, getting split tests to work for you is the more relevant part. The AdSense team offers us a method to split test their ads using Javascript, which you could extend for conventional text and navigation by using document.write(). This, however, won’t work if your visitor has Javascript turned off. Luckily, bloggers have a more powerful tool at their disposal: The server-side programming languages powering their weblogs. WordPress, TextPattern and Drupal users have PHP, Movable Type users have Perl, while Typo users have Ruby on Rails. I’ll extend on these in later posts.

Interestingly, there are even a bunch of free tools available to supplement your A/B tests. Perry Marshall’s SplitTester tool helps you determine whether your A/B test results are significant in the long(er) term, while click tracking tools like Crazy Egg (which I looked at earlier) can be used to measure and graphically display the results of your split tests.

And if you have the cash to splurge, paid software are available. Check out Split-Run-Test (US$47) for a self-hosted solution, or Verster (US$87.98 for 20,000 page views) for a hosted one. I feel both are overkill for the average blogger though.

All this doesn’t mean split testing is always a Good Thing™, especially when it comes to ads. Even with the rigidity of A/B testing (i.e. only one change between the two tests, both variants must run simultaneously and with equal emphasis, etc.), there are still too many variables involved. Traffic is rarely stable, whether volume or quality. Ad quality varies from ad type to ad type. Goals can be hard to define, e.g. targeting your headline might reduce ad clickthrough. For more, I suggest reading Anthony Lawrence’s thoughts on the issue.

This pretty much sums up an overall picture of split testing. Admittedly, it isn’t as targeted to bloggers or as concise (a.k.a. not boring) as I’d have liked. But it has gotten me to plan a couple of more practical posts on A/B testing, which I hope you’ll find useful when they get published in the coming weeks. If you have specific ideas on what I should write on though, I would love to hear them.

If you found this post useful, keep updated with future posts by subscribing to blogHelper (for free) through RSS or email.