Changing the way you receive insight through tools and analytics
Continues learning of on-going lifetime risk in keyword
Niche/market monitoring and prioritization
Early detection of anomalies of Google SERPs
Home SEO Guidelines Panduan A/B Testing
Written by cmlabs

Panduan A/B Testing

Last updated: Mar 15, 2021
Disclaimer: This cmlabs SEO guidelines contain comprehensive information about SEO, such as an introduction and general guidelines. You may be sent through SEO Terms in from third parties or links. Such external links are not investigated, or checked for accuracy and reliability by us. We do not assume responsibility for the accuracy or reliability of any information offered by third-party websites.

Written by cmlabs

A Guide on Performing A/B Testing

Before you decided to perform the latest page testing for the first time, there are several things that you have to know beforehand so you won’t unwittingly break one of Google’s guidelines. Read the following guide on A/B Testing to learn about the topic further.

Test Variations in Content and Site URLs

This guide gives an in-depth explanation about how to ensure that testing variations (A/B Testing) in pages’ content or URLs will not have a significant impact on your Google search performance. The test does not give any detailed instructions on how to actually build or design the testing process. However, you can find more resources about the testing process in this guide.

About A/B Testing

Website testing is performed whenever you try out different websites (or parts of websites) and then compile data about the users’ reaction on each version. Most of the time, you will need some sort of software to help you compare behavior between two different variations from web pages (it can be from parts of a page, entire pages, or entire multi-pages flow), as well as tracking which versions are the most effective for your users.

A/B Testing is when you try to run the test by creating several page versions, each with their own URLs. While your users are trying to access the original URLs, you will redirect some of them to each URLs variations, then you will compare the users’ behavior to observe which pages are the most effective.

Multi-variations testing is when you use software to change several different parts of your website quickly without interrupting other processes. You can test changes in multiple parts of your web page, such as heading, images, and the “Add to Cart” button. The software will then display variations from each of those parts to the users in different combinations, then it will automatically analyze which variation statistics are the most effective. A single URL is all it needs; while other variations are attached to the pages dynamically.

Depending on what types of content you are testing, this won’t be a problem if Googlebot decides to crawl or index several of your content variations while you are still in the middle of testing. Minor changes—such as sizes, colors, button placements, images, “call to action” texts (“Add to Cart” vs “Buy Now!”)—can have a significant impact on users’ experience when they visit your web pages. However, they won’t matter as much in terms of snippet or ranking for the search results.

Aside from that, if you have crawled enough sites to detect and index your own experiment results, Google may index the subsequent update that you make in a short time after concluding said experiment.

The Best Methods when Conducting the Test

Here are a list of the best practices that you can try to avoid the unwanted effects on your Google Search while testing site variations:

Don’t Hide Your Test Pages

Do not display different URLs to the Googlebot, from the ones that you present to your users. This particular activity is called Cloaking, and it violates the general Webmaster Guidelineswhether you are still in the middle of testing or not. You have to remember that violating Google Guidelines can get you and your site in trouble, for instance you could get a penalty or even worse, Google could just remove your website from their search result.

Cloaking still counts even if you decided to do it by server logic, by using robot.txt, or any other methods. Instead of cloaking, use the links or redirects as explained in the next point.

Use rel=”canonical” Links

When you do an A/B Testing using several URLs, you could use the rel=”canonical” link attribute on all alternative URLs to prove that the original URL is the preferred version. It is better if you use rel=”canonical” instead of noindex meta tag as it complements your current purpose. atribut link rel=“canonical” di semua URL alternatif guna menunjukkan bahwa URL asli merupakan versi yang lebih disukai. Sebaiknya gunakan rel=“canonical” daripada meta tag noindex karena lebih cocok dengan tujuan Anda saat ini.

Let’s take an example, if you perform a variation test on your home page, you will definitely want the search engine to index your home page, you also want Google to understand that all test URLs are all duplicates or just similar variations to its original URLs and should grouped together into one, with the original URLs as the canonical. Using noindex instead of rel=”canonical” in such situations can sometimes end badly for you.

Use 302 Redirects, and not the 301

When you perform an A/B Testing that redirects your users from the original URLs to the URL variations, then use the 302 redirect (temporary) rather than 301 (permanent). This will let search engines know that the nature of the redirect is only temporary—the redirect will only be applied as long as the experiment is still in session.

Any search engine should be able to keep the original URL in their index, instead of replacing them with a redirect target (test page). You can also use JavaScript-based redirects.

Perform the Experiment as Necessary

The amount of time needed to do a proper test varies depending on several influencing factors such as conversion ratio and how much traffic your website gets daily. A good testing tool would notify you when you have gathered a good amount of data to draw credible conclusions.

After concluding your test, you should update your site as soon as possible with the desired content variation and also immediately delete all testing elements, such as alternative URLs or scripts and test markups.

If Google found a website that still runs its experiments for longer than necessary, Google will consider such a site to be an attempt to trick search engines and then it will take the appropriate measures to deal with the situation. This is especially true if that web page displays content variation a large amount of percentage from your users.

For Further Information

To learn more about this topic, you can do several of the following.

A/B Testing on a Website

Agency Website

An agency website has to run the necessary page version adjustments to increase its users’ experience.

E-Commerce Website

To attract customers, an e-commerce website has to perform various testings especially when there is a newer web version.

Brand Website

The growth of a brand name has to be complemented with an upgrade to its website version, and as such it also needs A/B testing.

Blog Website

Bloggers may not necessarily need to perform testing but just in case they want to upgrade their sites, then they have to run a test to see whether their readers prefer the new blog version or not.

For pricing, licensing, renewals, quotes, payments, and other related questions, please contact us at
Share Now

Copied to Clipboard !