Starting with A/B or Multivariate testing for your Landing page

If you understand the importance of landing pages and why you should not direct your campaign traffic to your homepage, then you must know that creating landing pages is only half the solution you are looking for,…..that is improving conversions.

Common Fallacy

“I have created a landing page template and use it with different pictures/styling for various online campaigns.”

Well that might not always work. Even if you have one best performing version, chances are that for different campaigns it would not perform as expected. WHY?

1. Every campaign has a different objective. example: Registration or Buying products.
2. Every campaign would have a different target audience. example: coupon specific crowd or premium customers.
3. The landing page would not have same content/images/styling.
4. When user lands on landing page 99% chances are that the user context is different that other landing pages.

To sum all above, there are many variables which impact conversion rate so assuming that one best performing template would work always would be foolish.

The only solution we have is:

• A/B/N Testing
• Multivariate testing

You can google and find more about these. The main difference among the two is: A/B/N testing is mostly done to test radically different versions of same page whereas multivariate testing is often used to test incremental features.

Which strategy to choose?

1. A/B testing: If you’re creating landing pages for the first time or are unaware of your target audience.
2. Multi variate: If you already have a landing page template based on previous data or on site structure.

Example:

If you’ve created a landing page where you want users to register and are confused whether you should use “Sign Up now!” or “Register Now!” . This is a perfect case for Multi variate testing. Whereas if you’ve created two versions of same landing page with a lot of changes then A/B testing would be more suitable.

How to start doing A/B or Multi variate testing?

Start using Google Website optimizer {It is also included in Google analytics as content experiments}, its a free tool and best in the industry for testing landing pages. I have tried many other tools Crazyegg, Clicktale, Visualwebsiteoptimzier etc all have their own usp’s but GWO is the best. Lets begin :

How to start an A/B test?

You need to upload both your landing pages on your website, in case of WordPress users you need to create two different versions of the same page with different template. Then, open your Google analytics->Content experiments and create a new experiment.

So, you just simply enter different page’s url’s (NOTE: The pages should be publicly accessible by the Google optimizer crawler). Then, you’re given three types of code. One which is placed on the original version of the page, the second which is placed on the second version of the page and third which is placed on the conversion page, such as the registration success/completion page.

You’ll then choose the percentage of traffic you want to test your experiment on.

Then you’ll place the code on your pages and start the experiment.

How to start a Multi Variate Test?

Multivariate test has most of the steps same as above but the variation creation is different. In this case you need you add following tags, where ever you need to do testing. For example you need to decide whether you should use Sign Up now or Register now in your main call to action button.

After experiment setup, Your button HTML code would look like this:

<a><script>utm_section(‘Button’);</script>Signup Now!</noscript></a>

Next in Website optimizer, you’ll click on verify and start creating new versions.

In new version you can enter any HTML that will be replaced within the </script> and </noscript> tags.

So, in our example, I would create a new version and enter simple text “Register Now!”. Now our experiment would start

How does GWO work? {For more technical folks}

Notice that there are three kinds of scripts provided by GWO or Google content experiments.

1. Control script : This script determines whether user should be included in the experiment or not. Which page/variation the user should see.

2. Variation script: A simple page view counting script.

3. Conversion script: Records goal conversions and variation to which the Goal is attributed to.

So, now we’ve setup experiments and they are running. You can check by using multiple browsers and if you’re lucky then you’ll see different versions of the page.

Now, the reports section. GWO clearly does all the statistical significance tests to decide which one is a winner/a looser.

Tools like GWO help us in optimizing landing page keeping all other environmental variables constant. Since the traffic is randomly distributed, you can assume that the nature of the traffic, intent of the users landing on different variations of the pages is same. This gives us perfect result in which feature of landing page or which landing page version is performing better.