Every decision you make for your site design should not be based on subjectivity (i.e. personal taste, what you like as an individual), or indeed what I might personally prefer, but it's what the users like, that should count most.  

The data generated by the users is what Ezoic uses to make its decisions, so testing layouts via multivariate testing systems like Ezoic is all about the numbers. For example:

  • If a site has a naturally high bounce rate (let's say a reference site, an online tool, or a calculator), it should 'look' different in its optimized form than if it was a specialized blog with higher page views per visitor.
  • Similarly, if 90% of your visits are 'new' vs. repeat visitors, the optimal layout would be different in terms of ad placements.

The point we're trying to make is that the 'look' of a site is subjective to the publisher, but should be directly related to visitors' actions and user metrics.

It is very difficult to find the right layout by testing things manually (or guessing the right layout based on personal taste).

We believe you should really consider using multivariate testing to help you improve user experience (i.e time spent on site, page views, bounce rates...) and of course increase your ad revenue.