Multivariate testing is another useful option for testing your landing pages. This method is quite similar to A/B testing but with crucial differences. In multivariate testing for landing pages, a web page is treated to be composed of different elements such as headlines, Forms, buttons and Sales content, which actually it is. These elements affect the conversion rate of the page. A website is decomposed into distinct units. Different variations of those units are created and tested.
Let’s say we divide a sales landing page into headline, sales content and form. Different variations are created for each one of them. For example,
Headline: headline 1 and headline 2
Sales content: Sales content 1 and sales content 2
Form: Form 1 and form 2
You can create some more versions of the elements, if you wish to.
We have three variables in here, each having two different versions. Our goal in multivariate testing is to find the right combination of these versions which gives us the highest conversion rate. The different variations of page that we can create here are:
- Headline 1 + Sales content 1 + Form 1
- Headline 1 + Sales content 1 + Form 2
- Headline 1 + Sales content 2 + Form 1
- Headline 1 + Sales content 2 + Form 2
- Headline 2 + Sales content 1 + Form 1
- Headline 2 + Sales content 1 + Form 2
- Headline 2 + Sales content 2 + Form 1
- Headline 2 + Sales content 2 + Form 2
The traffic is split between each of these eight versions of the page to find the highest converting version. Threshold traffic is set as in A/B Testing.
How To Do Multivariate Testing for Landing pages
First of all, you need to find a tool that supports multivariate testing. The most common of them are Google Website Optimizer and Visual Basic Optimizer . Not all A/B testing tools support multivariate testing for landing pages.
Once you have decided which tool to use, you need to select the elements which you want to include in the test. Now there are many elements in a web page such as footer, headline, sidebar, navigation buttons etc. You cannot choose each one of them. Creating their variations and testing them live would be a massive task. Select only those elements which you think are important to your conversion goal. Below are some of the testing elements:
- Headline and heading.
- Call to action buttons.
- Text copy (content, length, size)
- Form length.
- Image (Size, quality, placement)
Once you have decided the elements, the next task is to distribute the traffic in each of the combination. As in the above example, we had three sections with two variations. The total number of combinations landed to 8. The number exponentially increases with an increase in sections and variations. 3 sections with 3 variations give 27 combinations. Therefore, the traffic requirement decision in multivariate testing is quite critical as each version definitely needs a fair share of traffic to produce distinctive results
Types of Multivariate Testing for Landing Pages:
Based on the distribution of traffic, we have different types of multivariate testing :
A. Full factorial Testing:
This is the most common form of testing method in multivariate testing. When people refer to multivariate testing landing pages, they refer to full factorial multivariate testing. In this form, the entire website traffic is equally distributed amongst all the different combinations. If there are 16 combinations, each version will receive one-sixteenth of the entire website traffic. Since each combination gets the same amount of traffic, it provides equal opportunity for each version to showcase its effectiveness. The metrics data (bounce rate, form fill rate, sales etc.) obtained are legitimate to determine which combination works best for the traffic. You may discover that a certain image has no effect on conversion rate, while a headline may be very influential. This method deals with hardcore facts and stats, therefore it is the most preferred way of doing multivariate tests.
B. Partial or Fractional Factorial Testing:
The entire website traffic is diverted to only a fraction of combinations. The conversion rate for unexposed combination is deduced from the ones that were included in the test. For example, if there are 16 combinations, the traffic is split equally among only eight of those. The metrics data for remaining eight combinations is derived from mathematical assumptions. This method is not recommended since inferences are needed to be done to come to a conclusion.
C. Taguchi Testing:
Very few marketers use this form of multivariate testing. This method uses higher mathematical assumptions to derive the effectiveness of different versions. Read more about Taguchi multivariate testing for landing pages.
What Makes Multivariate Testing Different from A/B testing?
AB testing vs multivariate testing
1. Traffic Requirements:
As explained above, the traffic requirement in both the cases is entirely different. The number of combinations needed to be tested grows exponentially in multivariate testing. One can test 3 or 4 versions with A/B testing and hundreds of combinations with multivariate testing for landing pages. The amount of traffic and time required is definitely high in multivariate testing.
A/B testing is usually conducted for testing large radical changes (such as completely changing the landing page or displaying two more CTA buttons). Multivariate testing is used to microscopically refine and optimize an existing design. A/B testing is global whereas multivariate is local.
The biggest advantage of multivariate testing over A/B testing is that it can identify which parts of the webpage are most influential on conversion goals. Many tools give you a metric which can identify which parts played a major role in conversion. This is not possible in A/B testing as all the smaller variations are lumped into one bigger page variation.
It is very important to determine the right amount of traffic that is needed to run multivariate test. This is highly required to obtain significant and hard data. Use A/B split and multivariate testing for landing pages duration calculator , to estimate how much traffic your test will require.
Preview all the combinations, personally. Filter your combinations and determine the best which are worthy of inclusion in the test.
Multivariate testing really helps you to identify the converting elements in a website. With proper multivariate testing landing pages, you are sure to increase your conversion rate.
If you are looking for a step by step process to optimize the conversion rate of your website, you can visit our Conversion Rate Optimization Guide to turn your visitors into customers. Please feel free to leave any comment you have in your mind.