Website testing is often performed by domain owners to improve their conversion figures. It involves using different version of your webpages and seeing which variant returns the best results. Conversions can be any human involvement metric which the domain owner chooses to test.
It can be an incredibly effective tool in finding a combination of web design components that achieve better conversions. Otherwise, you essentially have to play a very slow process of elimination as to which element is the problem.
There are two primary methods used to perform these types of tests. Below we cover what they are and how to perform them without Google thinking you’re trying to trick its webcrawlers, and subsequently penalising you in the rankings.
Multivariate and A/B Testing Defined
A/B Testing – involves creating different versions the same page with different URLs and randomly redirecting various visitors to these different versions. Then analysing the results and reviewing the most effective page.
Multivariate testing – uses only one URL for the page. Webmasters make small changes to the page in real time and use software to show these combinations to visitors, then statistically analyse the conversion results.
All types of testing that involve displaying different versions of sites and pages to visitors in order to determine their effectiveness is also known as “split testing.”
Avoid Cloaking
Google take particular exception if you show one piece of content to visitors and another to their webcrawlers. To search engines, it seems as if you’re attempting to unethically optimise your website while trying to retain the readability for human visitors.
The bots won’t be able to tell if you’re running a test, so the possibility exists that you may end being heavily penalised or even completely dropped from the rankings.
Use Canonical Pages
For those running A/B testing, using various URLs of the same page, use the rel=”canonical” link attribute to define the preferred version of your page.
This will tell webcrawlers what your intention is. That you’re testing your pages and want them to know that the various URLs are near duplicates of the same page. That way they can be grouped together as such, and only the “canonical” page will be considered official and ranking worthy.
Use Temporary Redirects (302s) Rather Than Permanent (301s)
Google advises against the use of 301 redirects, as they are permanent and an old method used by spammers to avoid SEO penalties. Using these may give Google bots the wrong impression of your site.
Rather use a 302 redirect, which is only a temporary measure, and keep the original URL in the index. This can help webcrawlers know that testing is being conducted, and allows old URL’s to be restored when necessary.
Keep Testing Time To a Minimum
The longer you test the more suspicious it will begin to look to search engines. Don’t run it any longer than necessary to achieve indicative results with the test. Otherwise Google might interpret it as spamming, and take the appropriate action.
Afterwards, all the desired content and structure changes should be implemented and the test elements and alternate pages should be removed.