Your Website SEO Checkup: Part 7 of 8 – Testing | A/B Testing | Multivariate Testing

Testing can be a never ending battle, depending on how thorough you are looking to test. The idea is to find the most effective web page that will lead to the best results. Your results may be, increased time on site, increase in page views, increase in # of signups, etc…

There are two main ways you can go about testing your website. Below are the two best ways to test and what scenario you should use each.

A/B Testing

A/B or split testing is probably the simplest form of testing there is. You are basically developing two different versions of a web page and seeing which version produces better results. Those results could be measured in click through, time on site, bounce rates, or conversions.

You want to make sure that when performing an A/B test, that you have a control to measure from. Normally this will be your existing web page. Your objective is to find the winning combination that produces the best results. You want to test your control page (Page A) with a test landing page (Page B). The page that performs the best, wins. If Page B performs better than the original, obviously go with B. The key is to only test one element at a time.

For example, you may own an ecommerce store and would like to test your “Buy Now” button. You may run a test on that actually button, changing the wording to be “Order Now”. You may also run a separate A/B on the color of that button, changing it from blue to yellow. Unfortunately A/B testing does take time because you are only testing one element at a time, but it can be easy to determine if the new page is a winner.

Multivariate Testing

Multivariate testing is much more in-depth because of the amount of different elements that are being tested at the same time. Multivariate testing allows you to accelerate the learning curve to see which elements work best from which pages. It allows you to test more elements much faster.

Say you are running 3 or 4 different page tests. With multivariate you are able to see which elements from which pages are the best combinations. Your outcome may be, Landing Page A: element 1, Landing Page B: element 5 and Landing Page C: element 3 is the best combination. With multivariate testing you have multiple variables on each test page.

What You Should Test

Kichus blog illustrates a good example of what you should test.

Additionally you may also want to look at testing some other elements: