Don't trust just one option
Many of our recommendations are based on our experience from long-time work on various internet projects. However, the best is to confront these recommendations with tangible data from your web analytics. One of the best methods, how to get to these data, is A/B testing.
Principle of A/B testing
For A/B testing, two (or multiple) versions of landing page or other element are tested in simultaneous experiment to find out which version brings better results.
What it looks like in practice
Incoming traffic is divided equally between tested variants, until we gather enaugh of statistically significant data to evaluate. The test usually takes pre-agreed time.
The output is a comparation of precisely defined metrics, for example conversion rate or earnings. Better values will show the winner of the test and this variant will generally continue to be used.