SEO Split Testing: How to use A/B Testing in SEO

SEO Split Testing

SEO Split Testing allows you to perform various experiments to find out how to optimize your website’s ranking on Google. Check out!

Certainly, you already know the A/B Tests, those that are done to determine which variable of the strategy generates more results in a given metric. But have you heard of SEO Split Testing?

If you don’t know, you don’t have to worry: it is a new concept that started to gain visibility recently, but already offers very attractive results and that can further optimize your website’s results.

Continue reading this post to understand what SEO Split Testing is and how it can help you achieve better ranking in search engines.

What is SEO Split Testing?

A / B tests are those in which we compare different variables in our strategy to see which generates the best results. We can test, for example, if using an emoji or not in the subject of an email increases its open rate (spoiler: yes, it does! ). Or we can evaluate whether it is better to use a contrasting or harmonic color in the purchase button of an online store.

SEO Split Testing is the application of this same logic, but in modifications to the structure of your website to find out which alternatives generate the best ranking in Google. These tests aim to find out which formats generate the most results (in this case, better ranking in Google) to optimize the page and increase the flow of organic traffic to the site.

Why do SEO Split Testing?

Performing tests to improve your ranking in search engines is something that requires large investments. This task takes a long time to carry out each experiment, in addition to the work of formulating the hypotheses, preparing the site, collecting the results and analyzing the data in order to find useful information.

If it generates so much work, why do SEO Split Testing? Why not opt ​​for a WordPress ready template? The answer is simple: testing gives benefits that no template can deliver.

Data-based decisions

Managing a website and a Content Marketing strategy involves making decisions often. There are hundreds of variables that must be evaluated and monitored, the consequences of which may involve success or failure.

No decision should be made using instinct as the basis, as this generates unpredictable results. Although the results are positive in the beginning, there is no guarantee that they will continue in the medium or long term.

Only data-based decisions guarantee solid, positive and sustainable results – and for that, it is necessary to carry out tests and experiments. If you test the elements of your site regularly, you will be able to gather a diversity of data to direct your actions in the best possible way.

Greater control of consequences

To explain one of the advantages of SEO Split Testing, let’s tell a story about Pinterest, that social network where we share images and videos of different topics.

Pinterest’s team of engineers had been working on-site improvements for some time and, for a number of reasons, decided it was better to render the platform’s media files in JavaScript. The technical part of the matter is beside the point, but the most important thing is that, before implementing the change throughout the site, they decided to do a little test.

Initially, they thought Google would crawl Pinterest content with the new rendering system as normal, with no differences from the old method. However, during testing on a limited number of pages, they realized that there was a substantial drop in access.

The new rendering system did not communicate very well with the search engines and the pages started to rank very poorly. Therefore, they chose to cancel the change plan and decided to study the subject further to understand what, in fact, happened.

For us, the most important thing is how this SEO test prevented a catastrophe on Pinterest. If the entire site had received the new system, visits would have plummeted and the platform would have a lot of work (and cost) to reverse the scenario.

So it’s important to test often, especially before major changes to your site. Thus, we can anticipate and control possible damage that would be caused (as well as implement changes that we know to be positive with more security).

The greater organic flow of visitors

If you find that implementing an H2 subtitle in an article on your blog can increase that page’s placement by 10% on search engines, what would you do? Logically, it would implement the feature on all pages. As a result, you would see all of your articles ranking better.

With this positioning, the number of visitors to your blog would increase, generating more fuel for your Content Marketing strategy. What if, after that, you found that some other specific change helped to improve the positioning of your articles by 7%? I would certainly implement this change and see the cycle repeat itself, again increasing the organic flow of visitors.

The basic idea is that the use of SEO Split Testing is so that we can understand what works and what doesn’t to rank your site better. If we identify the best options, then the site can be optimized and receive more visitors.

Preparation against updates

One of the fears of those who have a Content Marketing Strategy is to see all their work losing reach with one of Google’s algorithm changes. In some cases, an update can bite a portion of your organic traffic – in others, it can increase your reach.

Optimization tests help to mitigate the effects of these updates. First, they allow us to more easily identify what has been changed. After that, with the knowledge of previous tests, we were able to formulate hypotheses that help guide in the recovery (or optimization) of the effects caused by the algorithm change.

What is the logic behind this technique?

Now that you know what SEO Split Testing is and its benefits, you need to understand how it works. Basically, SEO Split Testing is a variation of the Experimental Methodology applied to website optimization.

If you remember your high school classes, you will remember that Experimental Methodology works by manipulating a variable within a set of rules to see what significant changes are caused by this change.

To guarantee the scientific rigour of the experiment, we always maintain a control group that remains in the same environment, but without changing the same variable. SEO Split Testing is just that. We modify a variable on a page (or a set of pages within a website) and maintain another set of equal weight without the change as a control group.

Then, we compare the two results to understand if there has been a significant change that indicates good SEO practice. As we created a control group with other pages of the same strength, there is no risk that the study will be contaminated by seasonal changes, such as the poor performance of competitors.

For example, suppose you created Group A and Group B for the SEO test. Group A is the modified one and Group B is its site without changes. If Group A had a statistically significant increase, it is a sign that the change is important to improve its ranking. If the reason for the traffic growth was something seasonal and unrelated to the change, it would have affected Group B as well, since both were in the air at the same time and the only thing that separates them is the object tested.

That’s why SEO Split Testing is a safe way to find out how to optimize your website’s ranking on Google.

How to do SEO Split Testing in Google Optimize?

There is a very useful free tool for performing website performance optimization tests: Google Optimize.

With it, you can perform A / B Tests to improve the elements of your landing pages, articles and other types of content. It is typically used to optimize metrics such as conversion rates, average visit time and bounce rate, but it can also be used for SEO Split Testing.

The way to do this is to modify on-page elements of your website and then measure the results using the number of organic visits as a measure of success. How to do this? Let’s see in the step by step!

Create a Google Optimize account

The first step is to create a Google Optimize account to run your experiments for free. If you already have a Google account (a Gmail, YouTube or something) you can use it to facilitate the process. Just go to the official Google Optimize website and log in with your Google account.

Link your Optimize account with your website’s Analytics

That done, the next step is to link your Optimize account with your site ‘s Google Analytics data. With this synchronization, it is much easier to obtain the data to attest to the results of your experiment. To do this, go to the right side menu, select the option ” Link to Google Analytics” and click on ” Link Property”.

A list will appear with all the Analytics properties linked to your account and you must choose the one for the website you want to test.

Create an experiment

The next step is to create your first test. To do this, click on the blue button “ Create Experience” or “ Come on!”, if this is your first project.

A screen will appear asking you to select the name of the experiment in question and the URL of the page that will receive the experiment. If you are testing on-page elements of an article, for example, you will need to enter the specific URL instead of your blog’s homepage.

Select the type of experiment

The next step is to select the type of test that will be performed. There are three main options:

  1. A / B tests: allows you to compare the performances of a single variable. For example: inserting an image in the first paragraph increases the time it takes to read an article;
  2. multivariate tests: allow to compare different variable elements. For example: if you insert an image in the first paragraph of a text and a subtitle in H2 it can generate a better performance;
  3. redirect tests: allow you to compare the performance of completely different pages, from content to design.

The correct choice will depend on the type of element you want to compare. A / B tests are more friendly for beginners, especially if only a variation of the tested element is used.

Determine the tested variable

Next, Optimize will ask you to enter a variant (that is, a modified page) for the experiment. Click on “ Add Variant” and give the test page a name.

The good practice asks to insert the element that differs in the name. For example, if this page has images with captions, while the original article does not have this component, the idea is to name it “ Page with Captions”.

After naming it, you must click on “ Edit” and install an Optimize extension for the Google Chrome browser, which will allow you to modify the site to create the page that will be tested.

Modify the test page

The next step is very intuitive. Optimize will open the page to be tested and ask you to insert the chosen changes. Just click and drag or insert new color codes to make the changes you want to put into practice in the test.

Set a goal

With everything specified, the next field to fill in is the “ Objective”, which you can choose from a list programmed by Google.

Basically, the Objective is a metric that will serve to decide if the tested variant is more or less efficient than the original page. In addition to it, there is a field for writing the hypothesis you want to test, for organization purposes.

Start the experiment

Finally, the last step is to put your experiment on the air to start collecting data. You can do this manually by clicking on “ Start” or schedule a time for the test to be performed.

Analyze the data

When you think you have collected enough data (it may take a few days or hours, depending on the flow of visits and the elements tested), end the experiment and analyze the information collected to conclude whether your hypothesis was correct or not.

What to keep in mind when doing these experiments?

Before starting to carry out your SEO Split Testing experiments, you need to keep in mind some considerations so that your work produces the best possible results.

Let the test run as long as necessary (but no more!)

The length of your experiment is crucial to the accuracy of the results. You cannot let the test run for a short time or for too long.

The ideal time will vary depending on the number of visitors your site receives. The most recommended is to have enough for a robust statistical base (1000, 2000 visitors), but it can be less if your blog receives fewer users.

The ideal is not to leave the test running indefinitely so as not to appear that the site is undergoing changes or to confuse users in any way.

Track usability elements tested by Google Optimize

Anyone who has experiences with Google Optimize knows that the tool tests some elements more linked to the user experience on your website. Some of these components are very important for the ranking of your pages, such as the length of stay for each visit or the bounce rate.

However, there are still other very important elements that affect the ranking of your site, such as changes to the page code. To test these components you will need to do manual tests. This means that it will be necessary to change the site code so that only a few pages have that change and establish a control group to ensure that your results will be reliable.

Read the Google good practice manuals

Whenever you do tests to optimize your site, read the manuals of good practices that Google offers, including when using Optimize. This is necessary because, in structuring your experiment, you can make some mistakes that are penalized by Google’s algorithm.

For example, the search engine suggests that you use canonical URLs when performing a Redirect Test. Thus, you inform Google’s crawlers that those pages are for testing and avoid any unforeseen events.

SEO Split Testing is an incredible tool to ensure that your website is optimized to rank better and win more users. We also saw that it is not a very difficult feature to use, is it? You can start testing right away!

Leave a Reply

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