Every landing page is automatically created as an Experiment.
The testable content is defined with Landing Page Versions. In a world where you only have one enabled version, 100% of the traffic will simply go to that version.
For landing page experiments, a pull is defined as a unique visitor, and the conversion is defined as the signup.
Using a Bayesian testing methodology will optimize for signups. This is the default methodology, but it is not appropriate for testing things like price changes or the impact of discounts on revenue.