Patrizia Vassallo

Associate Manager, Analytics

Q&A: Implications of Testing and SEO

An inherent goal of all website owners is to optimize pages to increase traffic and thereby increase conversions. Marketers are familiar with optimizations attained through SEO – and all these small modifications, if done properly, can lead to increased page rankings, greater visibility, and ultimately more traffic and conversions. Another effective way to optimize web pages is through split testing. However, some brands are hesitant to split test optimized and high ranking pages.

The most common question our teams receive is, “Will Google penalize our website if we A/B test this?” The short answer: no. Pages carefully sculpted to attain high performance in SEO are in no danger if the process of testing is understood and best practices are followed. Below, I address some of the common questions relating to testing and SEO.

Q: First and foremost, what is split testing?

A: Split testing involves taking two or more variations of a webpage (inclusive of the control) and directing a proportion of all or a subset of your traffic to each of these variations (for example, including only 10% of all traffic to be split 50/50 in an A/B test).

There are three ways to look at split testing:

A/B/n
This is the most straight-forward and widely-used type of split testing, given the ease of analysis when traffic is split only a handful of ways. A/B/n is often used when changing one element on a webpage or when a more drastic variant is introduced (for instance, redirecting to an entirely new design).

Multivariate
It's more accurate in measuring the interaction effects between independent elements and effectively determining exactly which change or combination of changes resulted in the greatest lift in conversion (if any). However, multivariate testing (MVT) can become complicated quickly. MVT requires that a website has a large flow of traffic to direct to each of the variations, the number of which is determined by following a full-factorial methodology (i.e., three elements to change on a webpage would result in nine variations, including the control).

Personalization
Although personalization is mainly concerned with delivering relevant experiences and messaging to each visitor, many personalization technologies provide the ability to designate a holdback for each personalized audience. As a result, personalization becomes a form of split testing where the content of the page will ultimately differ from the original page content.
 

Q: Can my tested pages be considered duplicate content?

A: In the case of a standard A/B/n test, multivariate test, or personalization campaign, driving traffic to a slightly modified version of the exact same page will not be considered duplicate content. Although Google may see the variation in question, the control will remain as the only indexed version.

However, it is important to be cognizant of how long tests have been running and to either set a predetermined sample size per variation following a chi-square test for significance analysis, or to regularly check into results for tests following a sequential statistics analysis.

Q: Should I be worried about cloaking?

A: Detrimental impact on the site and individual page ranking can only occur if Google perceives the changes on the page(s) in question to be cloaking. Cloaking involves showing one version of a page to a user-agent (Googlebot) and a different version to actual human visitors. As long as the content of the test is similar in intent to the original page and not malicious or harmful to users, then the page will not be penalized. Above all else, maintaining similar content in the variants is the most important factor to consider when split testing.

Q: What about redirect tests?

A: Redirect tests are common for full design overhauls where the intent of the page is the same, but the presentation differs entirely. When redirecting any proportion of your visitors to a new URL using a front-end testing platform, it is important to place a rel=”canonical” link attribute on the redirect URL that points to the original page. Doing so will indicate to bots that the control page should be indexed. In server-side redirects, using a 302 (temporary) redirect as opposed to a 301 (permanent) redirect is recommended.

Additionally, it is recommended to redirect to a URL that is similar in structure to the original URL, as well as to redirect to a page on the same domain. For example, it is advisable to redirect acme.com/redirect to acme.com/redirect2 as opposed to redirecting to riseinteractive.com or acme.com/uxanalytics.

Q: What if I am personalizing my visitors' experiences?

A: Presenting personalized experiences to visitors works in the same manner as straight-forward A/B testing. If personalizing a page to present a more relevant CTA or a more relevant hero image to visitors who have shown certain behaviors on a website, then Google will not read this as duplicate content. However, if a personalization strategy involves directing certain audiences to a version of the page that is entirely modified, then it is advised to use a rel canonical tag on the variants to ensure Google will not index these over the control (this is preferred over a noindex meta tag where Google may not index the page entirely).

Q: What is the best approach?

A: In addition to following the best practices outlined above, it is paramount to work closely with SEO stakeholders to account for and address any concerns that may arise as testing is underway. This is one of the benefits of working with an agency that handles optimizations of a website through both SEO and conversion rate optimization (CRO). Furthermore, ensure your team is set up for CRO and SEO success by testing ideas that are backed by sound quantitative and qualitative support. This can be achieved by implementing a comprehensive UX analytics program that follows a framework of strategic planning of testing, per research conducted on the website and with the visitor personas at hand. Along with practiced deliberation in ideating for website testing, managing the length to analysis for each individual test is also important to ensure SEO is not impacted by long-running tests.

For more information on testing best practices, check out these tips from Google. To learn more about how you can implement SEO best practices and testing on your own site, reach out to our team at Rise.

09/21/2017 at 07:26

Subscribe to
the Rise Blog

Sign up to receive valuable industry content, delivered straight to your inbox.

or Close this form

 

Want More?

Subscribe to the Rise blog and we'll let you know when new articles are available.

You might be interested in:

Blog / SEO, Customer Experience

How Faster Page Speed Can Impact Your Business

Read More

In the past several years, the number of online searches via mobile devices has rapidly grown [...]

Blog / Paid Search, SEO, Trends

The Next Chapter of Search Marketing: Signal-Driven Search

Read More

As the marketing landscape expands and new digital channels and formats are introduced [...]

Blog / Personalization, Strategy

Is Personalization the Right Play for Your Brand?

Read More

2015 has shaped up to be the “Year of Personalization.” As marketers, it is a [...]