Successful companies must be able to get most out of the website traffic they receive. That means converting the greatest number of website visitors into actual business leads as is possible. However, the process of creating a website can be arduous, and no matter how much time you put into it, it\u2019s hard to say whether the individual elements of your website are ideal.<\/p>\n
How do you know what visitors best respond to? This is where A\/B testing and multivariate testing can offer a lot of value. Also known as split or bucket testing, A\/B testing serves website visitors different variations of a webpage with different versions of a particular element. Let\u2019s say your website has a Call to Action on a particular page, and you want to test a couple of options to see which converts most effectively. You could change the verbiage used in the Call to Action, the position on the webpage, the size of the text, or even the color.<\/p>\n
For right now, let\u2019s just consider an A\/B test, so we will only test one element at a time. If the CTA currently says \u201cSign Up Now,\u201d let\u2019s say that two alternative verbiage variations we could test are \u201cSign Up Now for Free\u201d and \u201cGet Started Now.\u201d Using A\/B testing (there are a bunch of platforms that can make this process easy to implement, or you can create and serve up these variations yourself), we can show one of these three different versions to each unique website visitor. Then, we can simply wait until there is sufficient data to describe how well each variation performed.<\/p>\n
In a multivariate test, you can modify more than one element at the same time. For the past example, you can test the three different CTA wording variations, but also the color of the CTA and the position on the page. To test this, every single possible combination of these elements will be served up to website visitors, and again, the test results will show which arrangement performed best. The downside to multivariate testing is that since there are more variations, the test requires significantly more data, and therefore visitors\/time.<\/p>\n
This process can be utilized across your website, regardless of goals. What\u2019s important to consider before getting started though is what you are looking to accomplish by testing these website elements. You could test every aspect of your website if you wanted, but it takes time, energy, and money. Make sure that you are testing elements of the website that tie directly to your business goals.<\/p>\n
Something else to consider before beginning A\/B or multivariate testing is how long the test will take, and as part of that, how many samples your test will require to be valid. This depends on a few factors: your website traffic, your baseline conversion rate, the detectable conversion lift (impact) and the number of variations you want to test. If your website is only receiving a few dozen visits each day, a test could take several months to run. For the action you are looking to impact, if the conversion rate is already pretty high, the test will require fewer samples. If you only want to detect a higher conversion lift, say a 30% increase versus a 5% increase, the test will again require fewer samples to be statistically significant. Finally, the fewer variations of website elements that you are looking to test, the fewer samples you will need in your website experiment.<\/p>\n