A/B testing is an effective way to improve your site’s user experience and its ability to convert users to clients.
As A/B testing experts, we are often asked about the impact of A/B testing on our clients’ organic search rankings. If SEO is not taken into account, an A/B testing campaign can impact the visibility of the site, notably for tests based on URL redirects.
This post is a good opportunity to review A/B testing best practices for SEO and help you do what’s best when it comes to optimizing conversions, without jeopardizing your rankings and web traffic.
General SEO recommendations
To start, let’s review some general recommendations from Google.
On its blog, Google shares rules to be respected so that its algorithms do not penalize your site. The main rule concerns opening your test to the search engine’s robots, who must navigate on the same version of your pages as your visitors.
So, one of the first best practices for SEO is to not exclude Google’s bot from your A/B tests. Even if your A/B testing solution offers some advanced user-targeting capabilities, like user-agent detection, do not use them to exclude Googlebot.
It is also recommended that you do not display pages that are too different from one another to your users. For one, it will be more difficult to identify which element(s) had a greater impact on the conversion rate. Second, Google may consider the two versions to be different and interpret that action as a manipulation attempt. Losing ranking may result or, worst case scenario, your site may be completely removed.
Depending on your objectives, the A/B testing setup may differ and each way of doing things can have an impact on SEO.
Best practices for A/B tests with URL redirects
A/B testing using URL redirects, also known as split testing, is one of these methods. Instead of using a WYSIWYG (What You See Is What You Get) editor to design your variation, you redirect users to a completely separate page, often hosted on your site, that has its own URL. Using this method is justified if you have a lot of changes to make on your page; for example, when you want to test a different design or another landing page concept.
This use case is the most prone to error and can have a dramatic impact on your search engine ranking, namely your original page being removed from the Google index, and replaced by your variant page. To avoid this, remember the following points:
- Never block Google’s bots via your site’s robots.txt file with the Disallow instruction or by adding the noindex command on your alternate pages. The first prevents bots from reading the content of targeted pages, whereas the latter prevents them from adding the pages to Google’s index. It’s a common error, as the site publisher is afraid that the alternate version will appear in results. If you respect the following instructions, there is no reason for your alternate version to “rank” instead of your original version.
- Place a canonical attribute on the variant page and set the value to the original page. This tells Google the original page is the one it must take into account and offer to internet users. Search engine bots will understand that page B has no added value compared to A, which is the only version to be indexed. In the case of a test on a set of pages (e.g. you want to test 2 product page formats across your catalogue), you must set up this matching for each page.
- When a redirect test is completed, you must put into production the changes that have been shown to be useful. The original page A is then modified to include the new elements that foster conversion. Page B, meanwhile, can either be redirected to page A with a 301 (permanent) or 302 (temporary, if the page will be used for other tests) redirection.
Best practices for standard A/B tests
- The number of elements called by the overlay is limited given the overall page and that the test does not overhaul the page’s structure or content.
- The overlays do not delete or hide elements that are important for the page’s ranking and improve its legitimacy in the eyes of Google (text areas, title, images, internal links, etc.).
- Only run the experiment as long as necessary. Google knows that the time required for a test will vary depending on how much traffic the tested page gets, but says you should avoid running tests for an unnecessarily long time as they may interpret this as an attempt to deceive, especially if you’re serving one content variant to a large percentage of your users.
While it’s better to avoid overlay phases that are too heavy on pages generating traffic, you have complete freedom for pages that Google’s bots do not browse or that do not have an SEO benefit (account or basket pages, purchase tunnel pages, etc.). Don’t hesitate to test new optimizations on these pages that are key to your conversion rate!
What about mobile SEO?
Using your A/B testing solution to improve the user journey on mobile devices is a use case that we sometimes encounter. This is a particularly sensitive point for SEO, since Google is rolling out its Mobile First Indexing.
Until now, Google’s ranking algorithm was based primarily on the content of a site’s desktop version to position it in both desktop and mobile search results. With the Mobile First Indexing algorithm, Google is switching this logic around: the search engine will now use the mobile page’s content as a ranking signal rather than the desktop version, no matter what the device.
Therefore, it’s particularly important to not remove from mobile navigation – for UX reasons – elements that are vital to SEO, like, for example, removing page-top content that takes up too much space on a smartphone.
Can personalization impact your SEO?
Some A/B testing tools also offer user personalization capabilities. AB Tasty, for example, helps you boost user engagement via custom scenarios. Depending on your visitors’ profile or their journeys on your website, you can easily offer them messages or a personalized browsing experience that is more likely to help them convert.
For instance, highlighting customized content with an interstitial (pop-in) presents a challenge in terms of SEO, notably on mobile. Since January 2017, Google considers it to be harmful to the user experience, since the page’s content is not easily accessible. So personalized interstitials must be adjusted to Google’s expectations. Otherwise, you take the risk of seeing your site lose ranking and the resulting traffic.
Note that Google seems to tolerate legal interstitials that take up a majority of the screen (cookie information, age verification, etc.) for which there is no SEO impact.
To learn more, download your free copy of our A/B testing 101 ebook.