How to make an ab-test in google analytics not for a specific page, but for a common element of all pages (for example, headers)?

Question

Google Analytics

I study the topic of split testing in Google. Faced a problem.
Initially, it would seem that everything is simple - I created two or more versions of the page, specified them in the experiment settings, placed the Google code in the original page and let's go! But this is only if we are testing a specific page with a specific address. But what if we want to check the impact on the conversion of any element, say, from the site header, i.e. common element for all pages? For example, the inscription "free shipping". In this case, you no longer specify the address of the original page.

Answer 1

In another thread, he gave the same advice)
Try like this
If the server received a cookie with the number of the test variant (the cookie for each test must be named differently), then it displays the desired variant, otherwise it returns a blank page on which 2 actions are performed: the cookie is set in depending on the variant installed by Google, and the page is immediately updated so that the request goes to the server with a variant cookie. The option is defined as in the second solution.
2. Google itself offers us a simpler and more civilized way. But it focuses primarily on small changes. I'll get ahead of myself - for big changes it can also be used if you load the changed part of the page with Ajax. The essence of the method is to determine the option and make some changes through js (for example, the inscription on the button and the picture change).
An example of defining a variant here https://developers.google.com/analytics/solutions/...
An example of changing a page here https://developers.google.com/analytics/solutions/...