Web design not only presents the customer with product information but also communicates the brand values and philosophy to the customer. The design features of the web page make the most impact when it comes to forming the first impression of the brand. In this project, we investigates what web design features are indicative to how expensive a brand looks.
We extracted features such as whitespace, #colors, #products, and mean H/S/V from web design and got perceived price rating of brand websites from a crowdsourcing platform. A model then was proposed to analyze what web design features are discriminant to the perceived price of a brand. We implemented random forest with backward elimination (4-fold cross validation) and ran the model on watch and car brands. Features like #colors, #products, whitespace, and SDev of saturation are discriminant to perceived price rating.
Late-Breaking Work in CHI 2016 pdf