After you have developed a testable hypothesis, it is to time to create the variations of your control page. Follow these recommendations before creating the variations:
1) Brainstorm
Your test is as good as your ideas. Testing tools like VWO, Optimizely and GA Content Experiments can only implement your ideas. It is your team’s responsibility to come up with variation ideas.
a) Ask your team to list variation ideas
b) Provide an explanation for each variation
c) The recommended variation should be based on data (analytics or experiment results)
2) Start with the best practice list
Follow the best practices for creating web elements and page layout. For example, short forms have proven to convert more than long forms.
Now create the best practice list for each of the following elements:
Copy (Font/Color/Content)
Headline (Font/Colour)
Button (Shape/Colour/Edge)
Call to Action (Use of Verb)
Forms (Number of Fields/Button/Call to Action)
3) Check the Control Page
Check your control page and evaluate whether the page follows the conventions mentioned in the best practice list. Otherwise, develop your variations based on the list.
4) Define Performance Metrics
Before starting the experiment, define the performance metrics. For buttons, it is the percentage of visitors that have clicked (Download Buttons) or percentage of visitors that have purchased (Checkout Buttons).
5) Test Few Elements
It is common to see companies starting tests with many variations. Firstly, this will increase the time for completing the experiment. And most importantly, you will learn very little about the visitors. Pick one element, say Call to Action in a Button. Create variations only for that element.
For example, if you are managing an e-commerce site the ‘Add to Cart’ button contributes directly to the Business’s bottom line. Many experiments have proved that ‘Add to Cart’ is the most effective ‘Call to Action’ (CTA). But can you create a variation that beats ‘Add to Cart’?
6) Pick Experiment Type
After you have defined performance metrics, created variations and picked the elements for the test, select the experiment type. You can pick A/B or multivariate testing.
If you are testing more than one element, choose multivariate testing. If you want to see the impact of changing one element, choose A/B tests.
7) Dramatic or Subtle Change
When you are testing one element of the page like Call to Action, you have a pre-defined set of CTAs. But if you are testing page layout, subtle changes will not give any marked improvement. For such elements, make dramatic changes.
For Example:
1) Test three-column layout with two-columns.
2) Change background colour
3) Increase the Font of the Copy
4) Change the Navigation
8 Evaluate and Learn from the Experiment
Most often experiments are run indefinitely without any evaluation or marketers stop the experiment when they find a winner. The idea behind Conversion rate optimization is two-fold: find a winner and learn from the experiment.
Make sure that the marketing manager creates a Case Study about the experiment. The Case study should include the elements tested, the duration of the test, the variations created, the logic behind variations and the conclusions.
9) Make Changes and Observe
Once you have found the winner, make the changes. Sadly, on an average, 80% of the experiments fail. If the variation has improved conversion by over 20% then don’t hesitate to make the changes. Monitor the conversion trend for the next three months.