Deriving Data from Images

Google recently acquired a company called Skybox Imaging for $500 million. While admittedly a small deal by the standards of Google, the potential of this acquisition is so mind-boggling it is amazing it has not received greater attention. You see, Skybox makes satellites. And it’s just a year or so away of having six satellites orbiting the earth that will be able to photograph, in high resolution, every spot on Earth and do it twice a day.

In many respects, the innovation here is speed of refresh, not resolution of the images. There is nothing particularly special  about the optics used by Skybox. The excitement comes from the frequency of update. That’s because if you check on things frequently, you can see changes easily. And from those changes you can infer meaning.

Already, we know that satellite photos can be used to calculate the square footage of a building (based on the roof surface), a potential boon to roofing contractors, who can now do estimates from their desks. But you can also start to see angles for data providers as well: the size of a company’s facility can let you infer a lot of valuable things about that company. And Skybox will potentially take this concept to the next level.

Overlay satellite photos with map data (something Google routinely does now), and you now know who owns the property you are looking at. Checking the number of cars in the parking lot twice a day could allow you to infer number of employees. Over time, you could infer if the company is growing or shrinking.

One hedge fund (allegedly) now uses satellite photography to check the number of cars in lots at big box retailers to infer sales. It’s suggested that Skybox can assess the quality and yield of crops while still in the ground, as well as the amount of oil being pumped around the world (by analyzing storage tanks). Consider construction data, where new home starts and completion rates could be accurately measured on a daily basis. Consider measuring the truck and rail traffic into manufacturing plants over time to assess financial conditions. Let your mind roam, because that’s what Skybox is all about.

And lest you think I am alone in this geeky view of things, consider this statement by Skybox co-founder Dan Berkenstock, “"We think we are going to fundamentally change humanity's understanding of the economic landscape on a daily basis.”

The key to all this magic is software that is smart enough to interpret photographic images. This is where images get turned into data. And once that data is overlaid on maps, giving it both context and other data such as ownership, you quickly move to actionable data.

I’m focused on commercial applications for Skybox. For those considering the consumer implications, privacy concerns abound. For the moment it seems, we have to rely on Google not to be evil. And in the interim, there’s still a lot of work to be done to get this infrastructure fully in place and to determine what can be measured as well as what is worth measuring. But as a potential new source of high-value business intelligence in structured form, Skybox is painting a very pretty picture of the future.