Testing is critical to the success of any marketing campaign. Testing provides actionable data that translates into increased ROI. It is results oriented and provides insights that allow you to improve any marketing campaign in ways that are measurable.
Split Testing and Multivariate Testing
Basic testing of ad campaign elements (such as ad copy) is often done through what is known as split or a/b testing. Split testing is simply testing two advertising elements against each other under similar conditions to see which performs better.
An example of split testing is taking identical PPC ad copy and driving it to two different landing pages to see which page converts better, or creating two PPC ad copy variants and sending them to the same landing page to see which variant gives you a higher CTR (click-through rate).
Multivariate testing is the testing of more than one variable at a time, and requires sophisticated statistical analysis to determine results. Until now, multivariate testing resources were cost prohibitive.
Google Website Optimizer
Website Optimizer is a free multivariate testing application from Google. It allows users to test different combinations of content on a site or landing page with the goal of increased conversion rates. It is available to all AdWords users and provides analytic power on par with any website optimization software package out there.
If you are using PPC or email campaigns to drive users to a landing page with the goal of conversion, Website Analyzer is a godsend. Through well thought out landing page design and testing you can get feedback that will significantly impact conversion rates.
Define Conversion Goals
The first step in testing is defining conversion goals. A landing page should have one central goal – the action you want the visitor to take. This action defines the conversion. This could be a purchase in a transactional campaign, or an email address in a lead generation campaign.
The landing page is what you will be testing, measuring, and tweaking. This is the test page in the optimization experiment. The original content becomes the control, and variants of page elements such as copy and images compete with each other for improved results.
After completing the call to action, users arrive at a conversion page. Usually this page is an acknowledgement that the transaction has occurred, such as a 'thanks for your order' message. By definition, any traffic that lands on this page is converted traffic.
Once you have identified your test page and conversion page, decide what elements you want to test.
Deciding Which Elements to Test
Google Website Optimizer allows you to test up to eight elements at a time. Here are a few elements consider testing:
- Calls To Action (such as “Buy Now” buttons)
- Images and Hero Shots (i.e. product images)
- POA (point of action) assurances such as guarantees
64 Tips for Getting Started with Google Website Optimizer and 101 Easy ways to Use Google Website Optimizer provide more test ideas.
After identifying the test elements, you are ready to set up your experiment.
Setting Up the Test
On your test page, paste the 'control script' code before the closing HTML head tag, and paste the 'tracker script' code at the bottom of the page before the closing HTML body tag. Also, place the 'tracker script' code before the closing body tag on your conversion page.
Next, place the 'section script code' before and after the elements you will be testing and the variations of those elements. As an example, place the code before your original headline and a headline variation that you want to test.
Brian Eisenberg of Future Now has written an excellent white paper called 10 Minutes to Testing that provides a detailed overview of setting up tests with the Website Optimizer.
That's it. Validate the code and start the test.
Check the Results
Website Optimizer reports are intuitive and provide actionable data you can use to improve your results. Remember to let the test run for long enough to yield results that are statistically valid.
The report consists of two tabs: The 'combinations' tab, and the 'page sections' tab. Here is a screenshot of the 'combinations' page:
This page shows the performance of different combinations of the tested elements. The "combinations" column lists the results of your control page, followed by the best performing combinations in order.
The "estimated conversion rate range" shows how well each combination performed based on the percentage of impressions that resulted in conversions. For example, in this screenshot combination 11 had a 38.9% conversion rate representing a 24.9% improvement over the control page.
The "chance to beat original" is the probability that a particular combination will beat the control. The "chance to beat all" column is the probability that a combination will be more successful than any other combination tested.
The "conversions/ impressions" column enumerates the number of impressions each combination accumulated and how many times those impressions converted.
Here is a screenshot of the 'Page Sections' tab: