Minimize the effect of A/B testing on the search results in Google of your site

This page describes how to ensure the minimum impact of testing of various page content or URLs on the results of Google search. Instructions are not given here to create or design tests, but you will find additional resources about testing at the end of the page.

Testing Review for your site in Google

Testing on the site is a process in which you try different versions of your website (or part of it) and collect data on how users react to each option.

• A/B Testing is when you test two (or more) versions of changes. For example, you can check the different fonts on the button to see whether the clicks on the button will increase.
• Multifactal testing is when you test several types of changes at the same time to understand how each of them affects, as well as what synergies can occur between these changes. For example, you can check a few fonts for the button, as well as try to change (or not change) the font of the rest of the page. Is it easier to read a new font, and it should be used on all pages? Or is the advantage that the font of the button stands out against the background of the rest of the page, attracting attention?

For testing, you can use software that compares user behavior with various versions of the pages of your site (part of the page, a whole page or multiple streams) and tracks which version is most effective.

You can conduct tests by creating several versions of the page, each with a unique URL. When users try to access the original URL, you redirect them to variation URL, and then compare user behavior to find out which page is more effective.

You can also conduct tests without changing the URL, dynamically inserting variations onto the page using JavaScript to choose which version to show users.

Depending on what kind of content elements you test, Google may not pay significant attention to indexation of some variations of content while the test still continues. Small changes, such as the size, color or location of the button or image, or the text on the “Add to the basket” button against “Buying now!”, Can significantly affect the interaction of users with your page, but often do not affect the snippet or position in search results.

In addition, if Google regularly indexes your site, the changes that you make during testing will most likely be quickly indexed after the experiment is completed.

The best practices to minimize the impact of testing on the search results of your site in Google

Here is a list of the best practices to avoid negative impact on the results of your site search results when conducting experiments with variations:

Do not use cloating when testing on the site in Google

Do not show different URL Googlebot sets and users. This is called cloaking and contradicts our spam politicians, regardless of whether you are conducting a test or not. Violation of these politicians can lead to a decrease in the positions of your site or remove Google from the search results - which is clearly not the desired result of your test.

Cloking is triggered regardless of whether you use server logic, Robots.txt or other methods. Instead, use links or redirects, as indicated below.

If you use cookies to control the test, keep in mind that Googlebot usually does not support cookies. This means that he will see only that version of the content that is available to users whose browsers do not accept cookies.

Use Rel = "Canonical" to specify the canonical URL in Google

If you conduct a test with several URLs, use the attribute Rel = "Canonical" on all alternative URLs to indicate that the original URL is a preferred option. We recommend using Rel = "Canonical", and not the Noindex meta-tag, because it corresponds to your intention in this case. For example, if you test various versions of the home page, you do not want the search engines not to index your home page. You just want them to understand that all test URLs are variations of the original, and they should be grouped along with the original, which will be canonical. The use of Noindex instead of Rel = "Canonical" can sometimes lead to unexpected negative consequences.

Use 302 redirects, not 301

If you conduct a test and redirect users from the source URL to a test page, use redirect 302 (temporary), not 301 (permanent). This informs the search engines that the redirect is temporary - it will only work during the experiment, and they should leave the original URL in the index, and not replace it with the target URL (test page). JavaScript -based redirects are also acceptable.

Conduct an experiment only as much as you need for your site in Google

The duration of the experiment depends on many factors, such as the conversion and traffic of your site. A good testing tool will tell you when you collected sufficient data to get a reliable result. After completing the test, upgrade the site with the necessary variation of the content and remove all testing elements, such as additional URLs or scripts and dough marking. If we find that the site is conducting an experiment for too long, it can be perceived as an attempt to deceive the search system, and we will take measures to protect the quality of the search results in Google.

Additional information about testing your site in Google

• Google Analytics article on content experiments
• Google Analytics Tools for Content Testing
• Ask questions about testing on the Analytics Help Forum forum
• Ask questions about the impact of testing on the search results in the Google Search Central Help Forum forum.

Contact the SEO companion CEO On any question, you can by email: info@seo.computer, WhatsApp: +79202044461

ID: 155