A/B testing (or multivariate testing) is good practice to determine which aspects of your site need tweaking. But running multiple versions of a website can get you out of Google’s good graces (mainly because serving up different versions of the same page to different users has, in years past, been used for nefarious SEO purposes).
So here’s what Google has to say about running split tests and maintaining good SEO:
Website testing is when you try out different versions of your website (or a part of your website) and collect data about how users react to each version. You use software to track which version causes users to do what you want them to do most often: which one results in the most purchases or the most email signups, or whatever you’re testing for. After the test is finished you can update your website to use the “winner” of the test—the most effective content.
A/B testing is when you run a test by creating multiple versions of a page, each with its own URL. When users try to access the original URL, you redirect some of them to each of the variation URLs and then compare users’ behavior to see which page is most effective.
Multivariate testing is when you use software to change different parts of your website on the fly. You can test changes to multiple parts of a page—say, the heading, a photo, and the ‘Add to Cart’ button—and the software will show variations of each of these sections to users in different combinations and then statistically analyze which variations are the most effective. Only one URL is involved; the variations are inserted dynamically on the page.
So how does this affect what Googlebot sees on your site? Will serving different content variants change how your site ranks? Below are some guidelines for running an effective test with minimal impact on your site’s search performance.
No Cloaking— Cloaking, showing one set of content to humans, and a different set to Googlebot—is against our Webmaster Guidelines, whether you’re running a test or not. Make sure that you’re not deciding whether to serve the test, or which content variant to serve, based on user-agent. An example of this would be always serving the original content when you see the user-agent “Googlebot.” Remember that infringing our Guidelines can get your site demoted or removed from Google search results—probably not the desired outcome of your test.
Use rel=“canonical”– If you’re running an A/B test with multiple URLs, you can use the rel=“canonical” link attribute on all of your alternate URLs to indicate that the original URL is the preferred version. We recommend using rel=“canonical” rather than a noindex meta tag because it more closely matches your intent in this situation. Let’s say you were testing variations of your homepage; you don’t want search engines to not index your homepage, you just want them to understand that all the test URLs are close duplicates or variations on the original URL and should be grouped as such, with the original URL as the canonical.
Using noindex rather than rel=“canonical” in such a situation can sometimes have unexpected effects (e.g., if for some reason we choose one of the variant URLs as the canonical, the “original” URL might also get dropped from the index since it would get treated as a duplicate).
Use 302s, not 301s -If you’re running an A/B test that redirects users from the original URL to a variation URL, use a 302 (temporary) redirect, not a 301 (permanent) redirect. This tells search engines that this redirect is temporary—it will only be in place as long as you’re running the experiment—and that they should keep the original URL in their index rather than replacing it with the target of the redirect (the test page).
Only run the experiment as long as necessary.
The amount of time required for a reliable test will vary depending on factors like your conversion rates, and how much traffic your website gets; a good testing tool should tell you when you’ve gathered enough data to draw a reliable conclusion. Once you’ve concluded the test, you should update your site with the desired content variation(s) and remove all elements of the test as soon as possible, such as alternate URLs or testing scripts and markup. If we discover a site running an experiment for an unnecessarily long time, we may interpret this as an attempt to deceive search engines and take action accordingly. This is especially true if you’re serving one content variant to a large percentage of your users.
The recommendations above should result in your tests having little or no impact on your site in search results. However, depending on what types of content you’re testing, it may not even matter much if Googlebot crawls or indexes some of your content variations while you’re testing. Small changes, such as the size, color, or placement of a button or image, or the text of your “call to action” (“Add to cart” vs. “Buy now!”), can have a surprising impact on users’ interactions with your webpage, but will often have little or no impact on that page’s search result snippet or ranking. In addition, if we crawl your site often enough to detect and index your experiment, we’ll probably index the eventual updates you make to your site fairly quickly after you’ve concluded the experiment.