Call us on:
01392 349 580

SEO-friendly A/B Testing — Does it Exist?

After the dreadful jolts from the Panda and Penguin updates in the past two years, SEOs have become a lot more apprehensive about their next course of action. In many cases, this “too careful” approach has given rise to some irrational fears, causing them to miss great opportunities as they are less willing to leave their safe harbours and take risks to explore new growth potential of their websites.

Although conversion rate optimization (CRO) with A/B testing is quickly claiming its place in the mainstream, it is still one of those opportunities many webmasters are passing on, because they are not sure how the ‘G god’ might perceive it.

This is sad because those who have tried it are strong proponents of the powerful approach of this testing method. It isn’t surprising to see these people raving about their increase in conversion rates, especially because they didn’t have to invest too much of their time and effort.

A/B testing tools in the market make things easier by providing WYSIWYG editors to test changes on your webpage without having to fiddle with the code or HTML. It only takes 10 to 15 minutes to create a test.

In most cases, the only thing holding back many webmasters from trying A/B testing is their hesitance to play around with the SEO of their site. This is a fair concern. After working hard on achieving their coveted rankings or SEO metrics, no one would want to be hit with penalties.

But, if you reason things out, SEO and CRO essentially emphasize the same thing: improving the user-experience on your website. You’re making it easier for visitors to find what they are looking for. There doesn’t seem to be any reason Google would use this against you. In fact, look around all you want to find if any website has ever been penalized in search engine rankings or SEO metrics because they were testing. Chances are, you’ll not be able to find it.

In fact, the introduction of Google Content Experiments (previously Google Website Optimizer) in GA only made it more clear that Google is all in favour of testing. So, do you really think Google would penalize your website or ranking because you are using any other testing tool and not the one offered by Google? Sounds absurd to me.

Moving ahead, the fundamentals of CRO and SEO are also very compatible:
Both will give you better results when you focus on one product or topic.
Subtlety doesn’t work in either case. Clear and to-the-point copy and headlines is what help you steal the show.
Complicated flash animation and content presentations are a complete no-no.
Clarity in your copy and using headline tags properly will benefit your site in both the cases.

Addressing SEO Concerns Related to A/B Testing

What if your variation page gets indexed by the search engines?

“But how can this be possible? Don’t they say that these tests redirect using JavaScript and the variations are thereby not crawled or indexed by the search engines?” Yes, a lot many of you might believe this.

Sometimes, however, your variation page can be indexed by the search engines. This especially is true for multivariate tests because they do not redirect visitors from one page to another but only show a different version on the same page itself.

How do you solve this mystery then? The answer is understanding that A/B tests are not conducted in void. Real time visitors on your website are a part of the test, even when you do not allow search engine bots to crawl through those test variations. So maybe some visitor mentioned you on their blog and unknowingly linked the variation page URL of your website, or maybe someone bookmarked the variation page for later reference. And soon, your variation page will be indexed.

How can you prevent this?

Add the variation page URLs to the robot.txt file of your site: This will disallow indexing and crawling by search engines when you start your test.

Set 301 redirect to your variation page URLs: Once your test is complete, set the winning version to the main page’s URL. And don’t forget to add a 301 (permanent) redirect of all other variation pages to the main page’s URL. This channels all future traffic that attempts to visit any of the alternative URLs or variation pages of your site and takes them to the right homepage. You also will not need not bother about any dead-end traffic because of broken links.

Many people may even suggest that you should use the 302 (temporary) redirect over the 301 (permanent) redirect. Don’t get confused. The 302 redirect suggestion works best when your test is still running. The temporary redirect tells the search engines that this arrangement is in place only as long as the test is running. And the original URL should be used in the index instead of replacing it with the variation page. Redirects based on JavaScript also work well.

Is it OK to use a noindex meta tag?

Not Really. rel=canonical tag is your best bet in this case. It makes it clear to the search engines that you are conducting a test and that all test URLs are a mere variation of the original URL. Set your control page URL as the canonical and group the other variations to be treated accordingly by the Google bots.

Adding a noindex meta tag can sometimes be tricky because it would make even your control page go un-indexed — and you sure wouldn’t want that.

Even after you’ve followed all of these tip, you should also make sure you are not running any test longer than necessary. Running a test for longer than you should may be seen as an attempt to deceive the search engines. Their suspicion becomes more valid when one variation is viewed by a huge percentage of your visitors. If Google suspects that you are running a test for an unusually long duration, you are warned that you might face penalty. Unfortunately Google is not clear on what this acceptable duration should be.

Be sure to remove all test-related elements as soon as you’re done testing. This includes everything ranging from alternate URLs to testing markup and scripts.

What if you set things up for Google bot to always see your original content page?

That’s cloaking, my friend. Don’t show crawlers what humans are not seeing on your website. Webmasters cannot screw things up any more than distinguishing which test variant is seen, according to user-agent. Some of the webmasters even make sure that Google bot sees only their original content, which we know is not permissible. Breach Google’s guideline and wait for what happens next. Come on! We all know what they can do. Your site will be demoted or probably even removed from search results altogether.

So yes, you need to take a few precautions to conduct SEO-friendly A/B tests. And it may seem like too much effort right now. But once you begin to witness the increasing revenue figures, you’ll know it was all well worth the effort.

About the author: Smriti Chawla is a content marketer with Wingify, the company behind the success of Visual Website Optimizer, which is one of the leading A B testing tools in the industry.
Related Posts


Leave a Comment

Want to join the discussion? Please fill out the form below to leave your comments on this article.