Except that the question that the experiment wants to answer isn't what happens when you present users with the option. That would only become relevant later. The question right now is whether Cliqz' address bar autocomplete results are superior to the status quo.