During your campaign, you can add two types of email templates while changing the subject and email body at any step and run them both. After your campaign is launched you will get a detailed report via the Tracking tab and will be able to see which of the test options works better: more email openings stand for a better subject and more replies - for a better email body.
Reply splits the contacts in your campaign as evenly as possible, with each variant going out to approximately 50%. The system chooses who gets which variant randomly. For example, if you have a list of 200, the system will automatically send version A to 100 and version B to the other 100.
Email steps are split up randomly according to the number of variants on the step. Each step is randomized independently from each other, so if Step1 variant was A, it doesn't mean this person gets A variant in the following steps as well.
A/B testing only works if you add people in batches. For a single prospect, the variant will always be A.
Since the A/B testing is random, you won't know who receives which template variant.
We recommend adding a short list of people for testing templates as it's impossible to remove variants within active campaigns (though you can alter the contents to mirror that of the preferred variant). 100 to 200 people is a large enough test sample, and then you'll be able to duplicate the campaign and remove unnecessary variants.