Head of SEO
When it comes to getting more traffic, clicks and conversions, a digital marketer’s job is never finished. We’re continually hypothesizing, testing, analyzing and tweaking, in a bid to increase the website in question’s performance.
Using A/B testing allows us to scientifically trial out our hypotheses and understand whether the tweaks we are making will statistically impact performance.
Recently, we’ve seen lots of noise about SEO A/B testing, and some subsequent confusion on what the difference between conversion rate optimization (CRO) and SEO A/B testing is.
In this blog, we’ll aim to shed light on each technique, how they differ, and what they’re used for.
In short: CRO focused testing creates multiple variations of the same page, while SEO focused testing splits groups of pages into control, or variant. In SEO A/B tests, there is only ever one version of each page. This is because CRO is testing how changes impact the users’ behavior, while SEO testing looks at how changes impact traffic and rankings.
CRO is the act of optimizing your site for the user, in order to increase the number of visitors who convert. A conversion could be anything from signing up to a mailing list, to requesting a call-back, to completing a purchase.
By creating multiple variations of the same page and assigning different users either the control, or the variant, we can see how these changes impact the number of people converting.
When a visitor lands on the page, they are randomly shown either the control page, or the variant – the browser cookie ensures that this is the variation they see even if they leave and come back to the site again.
Once the test has finished, we can analyze the data, segmenting the traffic by variation to see how it was impacted. If a hypothesis is proven to be statistically significant (the variant page has more conversions), we can roll them out permanently, and across multiple pages. Likewise, if conversions drop significantly, we know not to make these changes.
CRO has many benefits. Conducting CRO tests forces marketers to take a scientific look at the data they have available and use this to guide their insights, rather than make guesses or assumptions which drives their strategy.
By testing out hypotheses prior to making site-wide changes, CRO allows brands a safety net to ensure there is damage limitation for any adverse effects of changes and also prevents budgets from being wasted on rolling out changes that will have a negative (or no) effect.
Implementing changes which have been statistically shown to improve conversion rates can improve all of your existing marketing efforts, as it capitalizes on existing traffic (whether that’s from paid, organic search, social or email) to improve revenue, lower cost per acquisition, and increase profits.
Although CRO and SEO testing both use A/B testing, their goals – and how they are carried out – are very different. CRO aims to improve the conversion rate once visitors have reached the site; SEO focused testing aims to increase the number of visitors to the site, by implementing technical changes that will improve the website’s position in the SERPs.
Rather than creating two variations of one page, SEO split testing takes a family of pages (e.g. ‘shoes’) and actions the changes to half of the pages (‘trainers’) to create a variant and keeps half the pages (‘heels’) the same to create the control. Every visitor will see the same version of the control (heels) and the variation (trainers) as other visitors, but the two pages will appear differently to each other.
We run SEO tests this way to avoid creating duplicate pages, as this could interfere with the test results – Google doesn’t like duplicate content, and older pages tend to rank better. By splitting up identical groups of pages into either a control, or variant, we isolate the technical changes made and can directly attribute an increase or decrease in search performance to these changes.
To check if the tests have been successful, we compare both the control and the variant’s actual traffic against their forecasted traffic. If the variant group outperforms its forecasted traffic and the control pages don’t, we can assume that the changes made have statistically improved performance and can roll these changes out across all pages.
Rolling out the technical changes needed to conduct SEO focused tests can be time-consuming, and brands often struggle to get the time and resource required from their development team to carry out these tests.
However, technical SEOs can implement changes, without using developers, through Content Delivery Networks (CDN) and EDGE technology. This allows changes to be applied to cached versions of the website.
SEO A/B testing can be timely and expensive to carry out, as it requires a lot of web development time to implement these changes. It is possible for SEOs to use Content Delivery Networks (CDN) to roll out these changes to cached versions of the website, essentially modifying the pages before it reaches Google (or the user), without needing to make any changes in the back-end of the website.
If these changes are shown to statistically improve traffic, they can be implemented sitewide – it’s much easier to get development time if you can prove that the changes will increase traffic and revenue.
In summary, there are several key differences between CRO and SEO A/B testing. Both tests are important steps in improving online performance but target very different metrics.
|CRO testing||SEO testing|
|Each individual page has multiple variations: a control, and at least one variant||Multiple pages within a family of pages are grouped into either: the control, or the variant. There is only ever one version of each individual page|
|Optimizes for visitors once they are on the page by improving page usability and user-experience||Optimizes for Google (and searchers) by improving technical aspects that will increase their position in the SERPS|
|Aims to improve the number of people converting once they are on the page||Aims to increase traffic|
|Uses Google Optimize and Optimize 360 to check for statistical significance||Forecasts traffic and then checks for statistical significance against forecasting using Google Analytics/ GA 360|
Want more SEO insights from our Head of SEO, Rob Marsden?Subscribe to our newsletter