How Much Is an Image Worth in the Sharing Economy?

Param Vir Singh, Carnegie Bosch Associate Professor of Business Technologies, speaks about his research on the economic impact of images.

Video Transcript

There's a lot of informational data shared online now in photos. There are about 2 billion images that are shared every day. So we tried to figure out what kind of information we can learn from images. I'm applying methodologies from machine learning and deep learning to understand the economic impact of images.

So for long, we have known that product pictures or images play a huge role in impacting demand, but there hasn't been a way to figure out how much of this impact is there or what aspects of the image really play a role in effective demand. So we study this thing in the context of Airbnb.com. We want to figure out what aspect of this image leads to more demand and then we want to optimize these images if we can. We can even generate new images. So initially the first thing we started with was to look at whether the quality of the image plays a role.

So what we found was if you get an Airbnb professional photographer to take your property images versus your own images, this will lead to about an 8% increase in demand.

If you can guide the person's eye on a diagonal across the room, that is the longest line in the room. The same room would look much more spacious. It's exactly the same room, nothing changes; it's just how you took the picture. We found that has a big impact on whether your property gets booked or not. We looked at symmetry in a room; that had a big impact, like how you organize things within the room. We found that color tones have a big impact. If you have warm color tones, that has a huge impact. We actually figured out if they were to optimize these things, it's close to about $80 million that they should be making more than what they're making right now, if they would just optimize post-processing images.