By Charlene Li
(Just getting around to posting this -- I wanted to try out a Keyhole more extensively, as well as some of the competitors.)
Besides being an amazing piece of technology, Keyhole uses geographic information as a foundation to display information – some examples CEO John Hanke demoed at the Web 2.0 conference included real-time traffic information and the location of bloggers. What Keyhole does well is amass and organize large amounts of data (sounds like Google!) – they just organize it by geography instead of by PageRank. John Batelle has a nice write-up on the announcement.
Another company that does something similar is TerraFly – see my image captures below of my office location from both of these services.
From having used both of these services briefly, Keyhole wins on sheer “wow” factor – there’s nothing quite like flying in over your building, and it has quite a few data points uploaded into the free service (the image from Keyholes shows a few local restaurants and subway stations). But Keyhole requires a cumbersome download while Terrafly works simply within the browser.
But why does Keyhole matter to Google? Besides getting some amazing mapping software (imagine plotting your driving directions in 3D), Google needs to support its nascent local search and advertising efforts. For example, Google’s AdWords allows geographic targeting, but only within a certain radius of a specific address. That’s well and good, but geographic targeting could benefit from an overlay of other types of information.
What if Google could overlay Claritas PRIZM segments against users IP addresses – and allow advertisers to adjust their bids up if certain households searched on specific terms? For example, if a search user was identified to be in the “Landed Gentry” segment, the advertiser may be more willing to pay more for this person’s attention.
Let’s take this a step further – what if an AdWords advertiser could review the performance of its search ads by geography – and through that, uncover trends in terms of income, education, ethnicity – and figure out what those people had in common? Then, the advertiser could also ask Google to find more of the same type of people, again using geographic data to target them. For large advertisers, this is something they already do through credit card purchase histories but currently can’t do online because of the anonymity of the Web.
What about privacy? Depending on how you look at it, it’s the beauty or scary part about this. Much of the information already exists in databases like PRIZM that are available to the public (for a price). But because it’s not targeted at a specific household, but rather, a geographic area, the data doesn’t necessarily invade a person’s privacy. It simply makes that connection that because of where you live, you’re more likely to have certain characteristics.