How do you test a call to action without changing the whole video?

Keep the demonstrated problem and product action consistent, then vary the requested next step or its wording deliberately. Record the destination behind each version so a change in landing experience is not mistaken for a copy effect.

Choose a measurement that corresponds to the request. A save-oriented prompt and a purchase-oriented prompt have different immediate jobs. Compare their business contribution where you can observe it, but do not treat more comments as proof that a purchase call to action was worse.