In Internet marketing, as with any other type of marketing, continuous A/B testing of new ideas against your control marketing campaign ensures that you are always getting the best conversion rates possible. In classical marketing, a small percentage of each campaign audience receives the new campaign, while the bulk of the campaign audience receives the time-tested standard campaign, and at the end of the campaign the conversion rates are compared. If the new campaign has a higher conversion rate, it becomes the new standard. While this method works well for slower media such as direct mail campaigns, it is inefficient for use on the Internet.
Faster tests make money sooner
Rather than running an entire series of test campaigns and calculating conversion rates at the end of the series, using Google Analytics to conduct your A/B test allows the conversions to be calculated after just one day. If one of the choices is clearly outperforming the other, the percentage of people who are shown the different options can be automatically adjusted to favour the best performing option. This allows you to be getting more conversions and making more money much earlier in the testing process. The tests are also generally shorter, since they can be ended as soon as a clear winner is identified.
Test more options at once
Split tests are normally referred to as A/B tests, because you are comparing two options. However, you don’t have to limit yourself to two choices when using Google Analytics to conduct the test. Their model is known as a multi-armed bandit model, referring to a bank of slot machines or “one-armed bandits” that a gambler can play in rapid succession. Instead of simply changing the wording on your headline and testing both versions, you can simultaneously change the wording, the font, and the colour, or try multiple variations on the wording against each other.
Test a variety of elements
Google Analytics has taken the idea of split testing far beyond simple ad copy or marketing text. You can test a variety of graphical elements to determine which pictures sell better. You can test different versions of navigation to see which creates higher user engagement. You can test your search protocols and determine which algorithms return results that encourage conversions. You can test your formatting to see if different layouts produce better conversions. Best of all, unlike hand-coded A/B tests that require you to create separate pages on your website for each possibility to which the visitors are then randomly directed, all of these tests can be located on the same page with the Content Experiments API dynamically loading and tracking the appropriate option.
The Internet Marketing Academy