The best way to approach your A/B method is to take the scientific approach or better known as “The Scientific Way”.
Ask a Question
Research the behavior of the users a website is targeted at and understand where they tend to bounce off the path leading to the goal. For example, you observe that a lot of visitors either don’t fill a required form or start filling it and then abandon it somewhere in between.
Do Background Research
A good place to start your research is by analyzing your site analytics. You can determine many things about how your visitors interact with your site. For example “The signup form has too many fields leading to a high abandonment rate.”
Construct a Hypothesis
A hypothesis is an educated guess about how things work. The hypothesis might be very specific or it might be broad. For example “Reducing the number of fields in the signup form will reduce the form abandon rate.”
Test Your Hypothesis by Doing an Experiment
Your experiment tests whether your hypothesis is supported or not. It is important for your experiment to be a fair test. Based on your hypothesis, you create a variation in which you reduce the number of form fields. You split the website traffic 50/50 between the original and the shorter variation, and wait for the experiment to run until it has achieved statistical confidence (95% confidence is the accepted standard and your chosen software should report this). Remember to always test against the original (the “control”) at the same time so you can compare results. This way you’ll know whether a variation is better or worse than the original.
Analyze Your Results
Once your experiment is complete, you collect your measurements and analyze them to see if they support your hypothesis or not. You see whether the variation performed better or worse than the control.
Conclusion
If the variation performs better, you implement it for all users. If it performs worse, you learn that form length is probably not responsible for the high abandonment rate and go back to identifying the problem.
