Over the past few years, there has been a virtual explosion of location information. Technology companies are collecting all kinds of place-specific information from cell phones, from logged point-of-purchase sales information, from traffic patterns on streets and sidewalks, and from business and events listings. But all of this information has thus far been somewhat disaggregated, and while much of it is available through public APIs, there is also a lot of data that is proprietary and hidden.
PlaceIQ CEO Duncan McCall, who has worked at a number of startups in the geo-location space, thinks the time is right to make sense of all that information. Street Fight recently caught up with McCall (who will be appearing at the 2011 Street Fight Summit) to talk about how we can use location data to get a better understanding of what a place is like, and how it changes over time.
Where did the idea for PlaceIQ come from?
About the beginning of last year I decided that the time was right, and that there was all this digital information being created about places and locations. It has influenced geocode, photos, social information, check-ins, tweets, all the way down to POI information and how many smart phones are in an area. All that information now is location-aware.
So the idea was to harness all the different platform information, make sense out of it — because it’s always very unstructured and different types. If you put them all together, you look into it the spectrum of hyperlocal, my thesis was you could actually build an interesting data products that didn’t have any privacy concerns — because there’s no PII in there — that would allow you to profile a location.
We ended up choosing the construct of hundred-meter tiles — hundred-meter by hundred-meter, about the size of half a city block — looking at all the geo-information in and around those tiles. The idea was you could infer powerful things. You could infer the types of people that are there, the conversion intent, what are they buying, what are they looking for, what are they searching for, how does the complexity of that location change.
How important is time of day changes in the hyperlocal environment?
Time of day is a really important thing for us. You can imagine a tile in a downtown city during the day — there’s a lot of workers there, changes at lunch time, changes in the morning, people commuting. It’s very different on the weekend. Once we understand all those dynamics then there’s the concept of how events change it? When the ball game’s on, when there’s a band playing in a local bar.
Using data we can start to understand all of these temporary environments. We even aspire to understand how weather changes people’s behavior, conversion intent, retail purchases. So with our technology we look at all the different dynamics, and the change over time is a huge one for us. We essentially can look at a location through any hour, any day of the week.
Is the information you use from publicly available APIs or do you take proprietary streams from different companies as well?
The majority of it is proprietary. Proprietary means everything from licensing POIs all the way through. We have a partnership with Skyhook and we use their location data — which is a fantastic data set. On its own, it’s not that interesting, but with our data it’s really powerful because they can tell us anonymously how many people are really in this location — because we know how many smart phones are in a location — which is a proxy for business.
We can spot patterns and say, “Why is it busy here and what are the patterns during a certain month of the year in terms of when these people leave the work area, and go to commute area, and when are the busy times, and what are the parts that are busy?” We want to target people in certain areas around a city, we can use the Skyhook data to understand when it’s actually busy there.
Looking at all the geo-information … you could infer powerful things. You could infer the types of people that are there, the conversion intent, what are they buying, what are they looking for, what are they searching for.
If it’s busy in a certain area, why is it busy? What’s going on there? White type of people are there? So, it’s everything from free data, partnership data, open data, all the way through to completely proprietary data sources we’ve licensed and built partnerships with.
How can your data inform mobile advertising?
On mobile right now, there are no really pervasive targeting mechanisms. There are no cookies and pixels like there is online that will evasively dissolve. So the majority of mobile impressions don’t have much context. They say, “This is an iPhone user and maybe they are on a game or sports related mobile app.” That’s all we know about them.
What we can do is by anonymously understanding the context of their location — we don’t track them, we don’t know who they are, we don’t follow them over time — we take that time and from the publisher and supply side, and we can turn that into context and say: “This consumer is most likely in the financial and tech world, and most likely on vacation because we see them on a hiking trail. They are in an area that scores extremely high for tourism right now. So, they could be a tourist.”
With that information we now can empower the mobile ad ecosystem to understand more about the impression and now connect it with a more relevant ad, which is better for everyone including the consumer (because they are seeing something that’s slightly more relevant).
The other side is we can empower brands, agencies, and advertisers to target robust, specific, markets. So a campaign we have right now is the ability to target late night diners. We work with a brand and the agency to be able to say, “These are the types of targets we want to shoot for. Which is people coming out of bars and night clubs, late night diners, recreators, college student, blue collar workers, shift workers, commuters, etc. People searching for fast food.
We can find all the hyperlocal tiles in that city that have that behavior by certain timed areas, and use our data to serve ads. We are now serving ads to those specific locations. We don’t know who those users are. We don’t pretend that we have 100% hit rate in terms of every single person that we say is a late-night diner. But we can enable them to extend their mobile media that actually target the tiles that have a high percentage for late-night diners.
The focus on location in the past few years has been really intense. Where do you see location-based services and data headed in the future?
The thing that is fantastically interesting for us is to be able to understand mobile ad conversions. What types of ads to people click on by location? Now we can correlate by that location to all the deep signals there and we can say, “We can now see what type of ads do college students click on when they come out of bars? What kind of ads to financial and tech analysts click on when they are at work or commute? Tourists? People at the beach? People at theme parks? You name it.”
We think there’s a very strong correlation between – especially on mobile – people’s environment and intent and their conversion metrics because people behave very differently when they are tourists or locals or when they are out drinking. So, it’s really interesting for us as we start to inject this conversion data and build up an understanding in an anonymously, privacy, and friendly way of what people click on by location. I think that’s going to lead to a whole bunch of targeting efficiencies and a much better experience for consumers.
This interview has been edited for length and clarity.