Machine learning can be used to personalise images using the Five-Factor personality model.
Choosing the right images is a key part of digital marketing as they can be more effective than text at grabbing attention and forming first impressions. Previous research has already managed to assess the general aesthetic appeal of pictures and even to match the right images to audiences with different demographic characteristics like age and gender.
But this study goes a step further and investigates personalising images to consumers with different psychological attributes, using the well known Five-Factor personality model. The authors report that they were able to predict how a particular image would appeal to people with different personality types, that this personality matching increased the positive impact of an image and, crucially, that this effect spilled over into improved attitudes and even purchase intention for the brands and products using them.
This insight could allow marketers to optimise their digital content by choosing the most effective visuals for different segments of their audience. But, the paper explains that the impact is fairly small, and marketers would require the data to assess the personality features of both a selection of images and their target consumers, and the process would need to be automated to add value at scale.
Read the original article:
[45 minute read; article may be behind a paywall]