How often do you split test?

Same here. In the beginning, I test a lot. At some point, you reach the point of diminishing returns, where running tests on things, even if it does yield some improvement, isn't worth the opportunity cost.

One piece of advice I'd like to share here on this day before Christmas (in the States, at least), is to start at the wide end of your funnel and test your way to the other end.

For example, in the basic Traffic > Optin Page > List Sequence > Sales Page funnel, start with the Traffic > Opt-in section. Don't worry about what color Guru X says your Buy button should be until you've optimized your opt-in process. Start with one source of traffic and test until the opt-in rate levels off. Then try another traffic source on a copy of that opt-in page as control, and test against that.

Once you have a steady, reliable stream of opt-ins, work your way through your sequence looking for the point where people drop off. Look at the messages on either side of that point, and test variations (one at a time) until the bottleneck is resolved.

Once you have a reliable, predictable stream of people to your sales page, start testing that page in earnest.

It takes some time, but if you start at the wide end of the funnel first, it takes less time than if you try to do everything at once. Especially if you're in the "try anything that pops into your head" mode (been there, done that).

Do your testing systematically, and with a specific result you want to improve.

Merry Christmas, Happy Holidays, best wishes whatever you celebrate...