A Small Business Search Engine Optimization Expert’s Guide on How to Handle a Failed Split Test

Staff Writer: Carolyn Johnson

 

Date: 7.27.2012

 

 

 

When a small business search engine optimization firm undergoes either multivariate or split landing page testing, success and increased conversion rates are often the result. However, there are times when, despite your best efforts and intentions, your split test will fail to be effective. Knowing what causes split test failure and what to do about it will make you more prepared for when this dreaded event comes around.

 

There are a number of causes attributed to split test failure.

 

  1. The test design is poor.

 

Before testing, you ought to perform research and analytics on your website’s performance. If the website and test design is built solely for aesthetics and without effective content, then the split test runs the risk of failing.

 

In order to acquire the research for well-informed test design, there are several routes you can go, including user testing, clickstream analytics, data from previous tests, heuristic usability research, and your own editorial staff. A small business search engine optimization firm performing more research before split testing will lead to saving time and money, as well as preventing a failed test.

 

  1. Variations have been left untested.

 

If you make just a few variations to the testing page before going about performing the split test, your results can change drastically. The test will result in the original site beating the variations performed on the site. If you’re only testing for small changes, split testing probably isn’t the right way to go. Multivariate testing, on the other hand, allows for testing of smaller variations and pinning down whatever is responsible for conversion rate changes.

 

With all this said, you shouldn’t be apprehensive about making the changes you believe will improve conversion rates, even large, bold changes.

 

With split testing as an A/B/n test, you’re often in the dark about which variation on the page is responsible for outperforming the original page. Testing the entire page will help you make variations and test with such apprehensions, allowing you to make the necessary changes for high conversions.

 

  1. The test was ended before procuring results.

 

A common error made during the split test is when the webmaster ending the test before its duration is complete. This occurs often because the webmaster already feels confident that the variation page will result in higher conversion rates. If you don’t let the test run its course and see significant results, aborting it prematurely could be costly.

 

Now that we know a few of the events that contribute to a split test failing, there’s the question of what to do next.

 

First off, segment each part and variation of the site and begin analytics. You can also use behavior analytics to determine how visitors used your website on what landing pages were most commonly used. Clickstream analytics are also a viable option here, along with behavior metrics. Try to isolate the aspects of the site that saw gains and find why and how.

 

You need not wallow in the failure of a split test. On the contrary, you can learn a great deal even when your split test fails. After all, split tests are generally used to determine what doesn’t work.

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *