What to Test First, Second, Third
The correct answer is of course that you should start testing where you have the most traffic and/or the biggest leak.
But let’s assume for a minute that you have no data. Maybe it’s a brand new site. Changing the color of your button might result in a 1.2% improvement, but a better offer can bring upon a 320% improvement. In that case you should start your testing with variables that can really boost your results.