Conversion Rate Optimization: A/B Testing versus Multivariate Testing

Conversion Rate Optimization: A/B Test versus MV Testing

A/B testing or multivariate testing? These two methods of website testing have proven very effective in practice. We at internetwarriors explain when you should use which and how you can apply both types of testing with Google Analytics and AdWords.

The conversion rate measures how many visitors a website operator manages to persuade to perform a specific action. It is always only a segment of the total potential of website visitors who respond to such a call to action on the site. A website aimed at converting a download or completing a purchase should fundamentally be designed for the goal of conversion and use optimized visitor flows for focused implementation.

The Leverage of Conversion Rate Optimization: Testing

Visitor success is measured by the conversion rate. Fundamentally, optimizing the conversion rate requires patience and a good sense. Different variants of assembling a website also entail different success factors and should therefore be well considered in advance. Test methods help website operators try out different versions of a landing page and are thus indispensable for optimizing the conversion rate of the site. If one variant is preferred by visitors, then it should become the permanent variant on the website. A/B testing or multivariate testing are tools for targeted conversion rate optimization. A/B testing compares two different versions of a landing page or ad, whereas multivariate testing tests individual elements of a single landing page or ad, such as the "Buy Now" button in different positions and colors.

The A/B Test versus MV Testing

A/B testing, also known as split-run testing, divides the target group into two subgroups, Group A and Group B, during the test phase. There are also two versions of the landing page, the original and the alternative. During the test, 50% of users (Group A) are directed to the original page and 50% (Group B) to the alternative page simultaneously, and reactions are compared. Examples of A/B testing can be alternative navigation or a new search function. The test itself happens automatically and can go so far, depending on the test software used, that, for example, only Google visitors or only returning visitors are considered for the test and subgroups.

The advantages of A/B testing are that it can be conducted quickly and is suitable for low visitor volumes since only one variable is tested. At the same time, it takes the perspectives of the target group into account.

However, disadvantages of this method are that too significant changes during tests could deter existing customers. Additionally, 50% of the information is automatically lost if only the winning variant is pursued after the test.

Multivariate testing (MVT) checks multiple possible versions of a single landing page to identify website weaknesses and user dissatisfaction. Here, various small changes to a variant are tested, such as replacing text, placing images elsewhere, or displaying the entire landing page or individual buttons in different colors. Since this test method examines several versions of the landing page simultaneously, significantly higher traffic is needed to achieve statistical relevance for each tested version.

By using multivariate testing, more precise statements can be made about the change in conversion or bounce rates. However, these tests are often only used on pages with high visitor traffic, as it would otherwise take too long to achieve statistical relevance.

Sometimes A/B testing is also combined with multivariate testing by first examining broad adjustments through A/B testing before refining details with multivariate testing subsequently.

Testing with Google AdWords & Analytics

Both Google AdWords and Google Analytics enable the creation of a testing environment. However, with Google AdWords, the focus is on ad and campaign settings.

Budgets, CPC bids, ad texts, and also different target URLs, thus different landing pages, can be measured with a selectable portion of the purchased traffic. The aim of an AdWords test should be to optimize the relevance chain between ad text and landing page. Does the user find on this page what was promised in the ad, and does it ultimately lead to conversion?

For classic landing page testing, we recommend using Analytics. In mid-2012, the Google Website Optimizer, originally native to AdWords, was integrated into Analytics. Analytics enables both A/B and multivariate tests. The landing page variants can be tested for bounce rates, views, and session duration or the set targets. In contrast to AdWords testing, the entire traffic can be used for analysis here afterward. A minor drawback is that the Analytics test requires installing an additional code snippet on the website. We therefore also recommend using Google Tag Manager here.

What can we do for you?

Do you want to know your customers better and use these insights to optimize your shop? Do you want to generate more leads and sales by creating a better usability experience for the customer?

Contact us, we develop concepts for you on how to optimize your tracking and derive valuable insights about your customers and your website with our help, and thus further develop your e-commerce business.