Marketing professionals and amateur Instagrammers alike have long scratched their heads trying to determine exactly why one particular image received more likes than others. Now there is finally some logic to be applied to this thought process, thanks to a new research paper by Aditya Khosla.
“What makes a photograph popular?” Asks Khosla in his introduction. “Can we predict the number of views a photograph will receive even before it is uploaded?” Khosla, along with Atish Das Sarma and Raffay Hamid, analysed approximately 2.3 million images from Flickr to “demonstrate that we can reliably predict the normalised view count of images… using both image content and social cues”.
Among some of the conclusions reached: colour plays a significant role in how people respond to an image, with warm colours like bright colours such as red and pink recommended, and green or blue hues less likely to receive likes. With regards to the actual content of the photos, Khosla found that a wide range of subjects racked up views, including “bikinis” and “miniskirts”, perhaps unsurprisingly.
Khosla, who is supported by a Facebook Fellowship, has previously developed a tool which modifies headshots to make them more memorable. If you’re curious as to how well your images might fare according to Khosla’s findings, you can upload them here to gage their potential popularity score.
If you don’t feel like being blinded by science, there is a much simpler way to (roughly) calculate the number of likes your images will receive, with the “Instagram Like Predictor”, an infographic over at The Bold Italic. This predictor factors in the “free likes” you get from friends and family members, in addition to other variables such as props (coffee and pastries always get likes) and craftiness (a picture of something you made yourself will be very popular).
“Can photographers be aided with suggestions on how to modify their pictures for broad appeal vs artistic appeal?” Khosla asks. “This could be an interesting research direction as well as a promising product.”