This Ariely experiment is unconvincing because it's quoted with percentages.

Are the free customers switching down from the luxury product, or are they new customers who otherwise would have gone for neither option? The percentages withhold that information.