Does A/B Testing Affect Your SEO?

A/B testing does not hurt your SEO when you run it the way Google recommends. Google's own guidance says testing is compatible with search if you avoid cloaking, use rel="canonical" on variant URLs, use 302 redirects, and end experiments once they have done their job. And when we checked Google results for two eCommerce shops running live experiments in Omniconvert Explore, the new content from the tests was never indexed.

This article covers why SEO still deserves your attention, which parts of a test can touch your rankings, Google's four rules for testing, and the real-world data from our own check.

Why SEO still matters when you run experiments

If you live in an AI bubble like we have been in recently, you might believe that search engines (Google or others) have become obsolete. However, search is still as relevant as ever.

While we cannot claim to know what the future holds, the numbers reveal a different story than the one spreading in my bubble. In November 2023, Search Engine Land reported, based on Similarweb data, that ChatGPT received only about 2% of the monthly traffic that Google did. The influence of ChatGPT on Google's search dominance was exaggerated.

Therefore, your SEO strategy is still very much important, and you should not neglect it.

At the same time, A/B testing is still relevant, as it challenges you to test assumptions and find the winning combination of website elements that can deliver the best possible conversion rates. But is SEO affected by A/B testing?

Does A/B testing affect SEO?

Your site ranking is affected by many factors, including page speed and your existing website content. An experiment can influence both.

Page speed

When it comes to poor website speed, the culprits are usually the hosting services, your files' types and sizes, plugins, and even your traffic volume. A testing tool adds its own script to the page, so it becomes one more thing to keep lightweight. Check load times while tests are live, not only before you launch them.

Content

Nowadays, many (if not all) A/B experiments change your content one way or another: a new headline, a different product description, an extra block of reassurance copy. The question is whether search engines see and index that changed content, or keep ranking the original.

Google describes two common ways to run a test, and each carries a different SEO consideration:

Google's four rules for testing without hurting SEO

These rules come from Google's official page A/B Testing Best Practices for Search.

1. Do not cloak test pages

   Do not show one set of URLs or content to Googlebot and a different set to humans. That is cloaking, and it is against Google's spam policies whether you are testing or not. Do not use your testing tool, server logic or user-agent detection to keep Googlebot in the control.
2. Use rel="canonical" on alternate URLs

   If a test uses separate URLs, add a rel="canonical" link on each variant that points to the original page. Google recommends this rather than a noindex tag, because it tells search engines the variants belong with the original.
3. Use 302 redirects, not 301 redirects

   When you redirect visitors from the original URL to a variant, use a 302 (temporary) redirect. It tells search engines to keep the original URL in the index. A 301 (permanent) redirect signals that the variant should replace it.
4. Run experiments only as long as necessary

   Once you have enough data for a reliable result, update your site with the winning variation and remove all elements of the test as soon as possible. Google says an experiment that runs for an unnecessarily long time may be interpreted as an attempt to deceive search engines.

When we use Explore for A/B testing, we take extra steps to ensure your rankings are not affected from a technical perspective. Still, it is always good to look at results and get real-world confirmation that SEO is not affected by experiments.

Real world experiments: what Google indexed

One set of such data came after we tested new content for two eCommerce shops and checked SEO rankings before and after the experiments. Here is the methodology we used.

Sample A

We took an eCommerce shop with high traffic that had Explore installed. It had an experiment running for 6 days with traffic evenly split at 50% each (control vs. variation).

Sample B

We used another eCommerce shop with substantial traffic that had Explore installed. It had an experiment running for over 30 days with 100% traffic allocation. This meant that for more than 30 days, all users saw only the changes from the experiment, with the control receiving no traffic.

What we found

In both cases, fresh new content was inserted that could not be found anywhere else on the page or website. Because Google is the dominant search engine in the countries we sampled, we focused our checks there.

• Google did not index any content added through the experiment.
• Google only found and returned results based on the content from the control group.

Sample B is worth a second look. Even with every visitor seeing the variation for over a month, search results reflected only the original content. That protects good rankings, but it also means a long-running winner does nothing for your SEO. It is also the kind of extended experiment that Google's fourth rule advises against.

What to do after a test wins

So there you have it. From a content perspective, if your website has good rankings, running experiments with Omniconvert Explore keeps your good results.

On the other hand, if your content performs poorly from an SEO perspective, it would not be wise to keep a winning experiment live for too long. Better to make those changes natively on your website as well.

1. Confirm the result

   Make sure the test reached a reliable result before you act on it. See our A/B testing best practices for how to judge a winner.
2. Build the winner into the site

   Add the winning content or layout to your templates or CMS, so it is part of the page itself and not injected by the test.
3. Remove the test

   Stop the experiment and remove its elements, including any variant URLs, redirects and canonical tags created for it.
4. Watch rankings and speed

   Monitor search performance and page load times over the following weeks, so you can see the effect of the new content on both conversions and SEO.

Frequently Asked Questions

Vladi Iancu

Head of Implementation