Search Engine Optimization (SEO) isn’t an easy task to carry out. That’s why most businessmen and women employ an agency offering SEO services to help them better optimize their website in order to improve their visibility in SERPs and pull in more organic traffic.
What makes SEO difficult? Google! Google is arguably the most used search engine in the world and in order to better serve its users, Google makes hundreds of changes to its algorithm yearly. Since SEO performs better when you have a better understanding of Google’s current algorithms, as an SEO user or expert, you are left with two ways to deal with Google’s ever-changing algorithm—you can either choose to make reactive changes to your website only after it has been affected negatively or become proactive with split testing to ensure that it’s doing its job.
Statistical split testing has to do with splitting pages of similar intent on your website into two groups, making changes to one of them, and monitoring the impact the change has on a key SEP metric such as click-through rate (CTR), and organic traffic, or keyword ranking positions. These small changes or tweaks improve your website for users and also ensure that future algorithm updates will affect your website positively instead of affecting it negatively.
However, there are times we observe traffic spikes within an SEO A/B split test—the spikes indicate that there are outlier URLs within the data set and it’s caused by misalignment of data. But what exactly are outliers in SEO split testing? Continue reading to find out.
What are outliers?
In its general sense (statistically), an outlier is a data point that differs considerably from other observations. An outline could be a result of variability in the measurement or an experimental error; sometimes the latter is often removed from the data set. In statistical analysis, an outlier can cause serious problems.
Types of outliers
The outliers in SEO split testing will be explained based on the following types of outliers in SEO A/B split testing:
To perform SEO A/B split tests, a group of pages will be split into two even-handed groups (control & variant) based on the organic traffic.
Each test usually has a minimum requirement of 100,000 clicks to the set of URLs being split tested in order to have a statistically significant test result.
If spikes occur, the traffic anomalies within the group are divided and the traffic outliers are excluded or removed from the control and variant groups.
If you run an SEO split test before, during, or after a holiday, you’d discover that your test has a traffic spike in them. That’s as a result of the URL(s) associated with the holiday. That spike in traffic throws off the test results.
To work around it, identify the URL(s) in question, review them to determine if they are seasonal due to the holiday, and then remove them from the test group.
With SplitSignal, you’ll be able to filter your pages and remove or exclude pages that are outliers before re-running the test. If after doing that, you still notice a spike in traffic, then you’ll have to review your URL(s) further as there could be additional pages that are outliers. Once you find them, repeat the process above.
outliers can occur in page elements when testing changes in page elements. For page outliers, ensure that all the URLs in your control & variant groups have the element on the page you need to change. That’s because the pages included without the element will be seen as an outlier.