How Is SEO Split Testing Different from CRO Testing?


In this post, we will discuss the differences between CRO (conversion rate optimization) testing and SEO split testing. This thorough overview can help you determine if SEO split testing fits into your SEO strategy.

While these tests sound similar, there are significant differences.

Distinct Differences Between CRO Testing and SEO Split Testing

CRO Testing

In CRO testing (sometimes called CRO user testing or CRO A/B testing), you duplicate a web page or an email, changing one element. You can then send traffic to both versions of the page or email. The comparative conversion rates then reveal which version provides a better conversion rate. 

In CRO testing, we’re solving for a user:

  • You are testing two versions of a web page or an email.
  • You are testing for web traffic you already have or an email list.
  • You are attempting to increase your conversion rate with on page elements.
  • CRO testing provides a conversion win on one page or email.
  • A CRO test does not scale across many pages or email messages.
  • CRO testing requires a new test for each page or email you want to test.

SEO A/B or Split Testing

We cannot duplicate a page in SEO split testing, which would be considered cloaking, a violation of Google’s webmaster guidelines

This can lead to your website being entirely removed from Google’s index or otherwise affected by an algorithmic or manual spam action.

In SEO split testing (sometimes called cohort SEO split testing), we test a change across a group of URLs in the variant group compared to a control group with the intent of looking for a statistically significant result.

In SEO A/B testing, we are solving for Googlebot.

  • You are splitting a group of URLs into two groups (control and variant).
  • You are testing for an increase in organic web traffic.
  • You are attempting to increase the number of clicks from Google’s search engine results.
  • SEO split testing allows you to roll out a positive test to sections of your website.
  • SEO A/B split testing scales optimizations across enterprise websites.
  • In SEO split testing, you can have hundreds to thousands of URLs within a test.

There are several types of SEO A/B split tests:

Pre and Post SEO Split Testing

In this form of SEO split testing, the SEO marketer makes a change on a single page, then takes a snapshot of the change, makes a note of what the Google Analytics (GA) traffic is, and may also measure the number of clicks to the URL from Google Search Console (GSC).

The SEO then comes back a month or two later and measures for any difference in GA or GSC traffic based upon the change.

This form of SEO split testing is very close to CRO testing; it is challenging to ensure that there are not other variables in play that might affect the test result. 

This form of testing does not lend itself to scaling SEO testing across your site.

Cohort Analysis in SEO Split Testing

A cohort is a group that shares common characteristics, usually within a specific timeframe. In SEO split testing, the cohort characteristics are the clicks from Google’s search engine results to the URLs in the testing group. The timeframe is the 100-days before the split test.

How to Conduct Manual SEO A/B Split Testing

In this version of SEO split testing, the SEO has a data science team split a group of URLs into two groups (control and variant) with an equitable traffic model for the past 100 days. This can take months to accomplish manually.

Next, the SEO has the development team make an SEO testing variant change to an element of each URL within the variant group. The extent of this task depends on the number of URLs within the variant group.

Then the SEO, development team, or analytics team sets up tracking for the test results with a dashboard such as Databox or Tableau. 

Then it’s finally time to launch your test!

Upon completing your test, have your data science team analyze the data using a Casual Impact statistical model to ensure that you had statistically relevant results.

Lastly, the data science team should provide you with the analysis and results within your dashboard.

This manual process is very labor-intensive and can take many months to set up, run and analyze the test results.

Cohort SEO Split Testing with SplitSignal

SplitSignal is a client-side SEO split testing platform that runs cohort analysis on tests and changes to on-page SEO optimizations.

How Does SplitSignal Run SEO Split Tests?

In SplitSignal, we take a group of URLs and equitably split them into two groups (control and variant) by organic clicks from Google Search Console. This takes minutes and is quickly done.

We then make one on-page bulk change to the URLs in the variant group via a JavaScript snippet code in the head of the web pages in the split test. 

What on-page changes are available for testing in SplitSignal?

Most of the element changes are visible on the URLs within the variant group besides page title changes and meta description changes. You can have a test up and running in minutes.

Googlebot then crawls the variant group of URLs during the test (usually within the first seven days), picking up the change.

SpitSignal tests are by default set to 21 days but can run up to 42 days long.

SplitSignal employs a statistical Bayesian Casual Impact model (Google’s standard for SEO split testing) to measure whether optimizations within SEO tests will impact clicks for Google’s search engine results.

SplitSignal puts all this test data into a test dashboard with test analysis. 

With SplitSignal, you can get SEO A/B split tests up and running in minutes and have tests completed within days, not months.

What Is a Good Website for SplitSignal?

The best sites for SplitSignal have a strong concept of templatized pages and 30K of clicks a month to a group of URLs (100K of clicks over the past 100 days).

Read our knowledge base article on whether your website is a good fit.

SplitSignal Metrics

In the SplitSignal dashboard, we show you whether or not your test is statistically relevant. We also supply the below metrics, which allow you to measure what change has occurred if any.

  • Type of Results: see below
  • Effect Size: the percentage change of organic clicks from Google
  • Absolute Effect: the number of organic clicks from Google
  • Confidence Level: Based on the Casual Impact model, any result of 95% (or above) is considered statistically relevant.
  • Duration: the timeframe the test is run for. Tests can run from 14 to 42 days.
  • Variation Group Traffic: The left side of this graph shows the 100-day pre-test traffic between the control and variant groups. The right side measures the testing period and the difference between the control and variant groups.
  • Real Clicks: the number of clicks before and after the test launch
  • Predicted Clicks: the number of clicks if there was no test (calculated by the Casual Impact model)
  • Additional Clicks: the cumulative difference between the actual and predicted traffic (what it would look like if there were no test). Additional clicks equal zero until the test launches.
  • Control Group URLs: the URLs included in the control group
  • Variant Group URLs: the URLs included in the variant group
  • Excluded URLs: the URLs that are excluded from the split test. URLs can be excluded for traffic that is not equitable in the pre-test 100-day period. URLs can also be excluded for redirects, URLs not found, or URLs blocked by page-level meta robots or the robots.txt file.
  • Visitation by Googlebot: the First Visit shows when Googlebot first visited the URL after the test launch. It may take a few days before changes appear in Google SERPs. We provide a metric based on the percentage of URLs visited by Googlebot’s IP address. We also provide a URL list and the date of Googlebot’s visit.
  • Google Search Console Data: Paid users of SplitSignal can receive Google Search Console (GSC) data for tests upon request.

Available GSC Metrics:

  • Clicks: provided in the tool
  • Clickthrough rate (CTR): available for download
  • Impressions: available for download
  • Average Page Ranking: available for download

SplitSignal Types of Test Results

In the SplitSignal dashboard, we provide you with the following three different types of SEO split test results. In addition, we provide the above metrics to allow you to measure changes based on your tests.

  • Positive Test: a positive change shows there is a lift in clicks to the URLs within the variant group
  • No Change Test: a test where there is no change in clicks from the variant on-page test
  • Negative Test: a negative change shows there is a decrease in clicks to the URLs within the variant group

Read our knowledge base article on how to read SplitSignal tests.

SplitSignal’s SEO A/B Split Test ROI Calculator

SplitSignal provides an ROI calculator, which allows you to showcase SEO split tests with positive results with additional organic clicks from Google, the average order value (AOV), and conversion rate.

You can then forecast a 12-month positive revenue to an optimization you implement with your dev team.

This ROI calculator can help justify optimizations based upon SEO best practices.

ROI example:

Start SEO A/B Split Testing with SplitSignal Today

Are you interested in scaling SEO to see a lift in organic traffic across large groups of pages on your website? 

Are you tired of falling back on SEO best practices to explain why a specific SEO optimization is critical to your dev team?

Are you looking for an easier way to show statistically significant SEO split test results that prove why an optimization would provide organic lift and an increase in revenue?

If you are looking for a way to show SEO ROI to your organization’s leadership, reach out to us today to request more information.



Source link

LEAVE A REPLY

Please enter your comment!
Please enter your name here