Data That Thinks ‘Big,’ But Acts Local
Earlier this summer, I moderated a panel at Street Fight Summit West about how local media is colliding with the emerging super-sector known as “Big Data.” The idea is that local content and ad delivery continue to get smarter with more context around people, places and how they’re connected.
This kicks into ludicrous speed as smartphone penetration breaks the 50 percent barrier, resulting in lots more signals to play with. So a batch of companies has emerged to capture, scrub and apply that data — building on what others like Localeze have been doing for years.
“Demographic information including age ranges [and] geographies, can be valuable predictors,” said Localeze CEO Jeff Beard. “Yet, if you can merge this information with socio-economic and lifestyle data such as affluence, shopping trends, businesses can build a much more precise privacy-friendly profile for targeting.”
Others in this batch include Sense Networks, PlaceIQ, Locu, Factual, and Urban Mapping — not to mention the mobile local ad networks that are building similar capability in house (xAd, WHEREads, JiWire, etc.).
This will define the mobile & local promise we keep hearing about — that location targeted ads will boost relevance, performance and thus monetiztion. But one thing often missing from the discussion is what the heck we mean by “location targeted?”
Some define it as reverse IP lookups, which has a knack for telling you that someone is somewhere in the American Southwest. Geofencing is more granular but that’s still at the mercy of the information that app publishers pass along to ad servers (“This app would like to use your location”).
Meanwhile, these methods lack deeper data to extract meaning from location. That’s where our big data players come in, with a goal to boost contextual, historic, demographic or behavioral understanding of a user (as Beard explained), or a spot on the map.
The latest example is JiWire’s “Location Graph.” It profiles users — moms, students, business travelers, etc. — based on patterns of where they’ve been. That derives predictive modeling around future behavior, and thus ads that will most resonate. “Just like social graphs look at interconnectedness of people, [our] patterns show linkages between places,” JiWire CEO David Staas told me.
This uncovered the fact that zoo-goers are more likely to visit family restaurants; or that sixty percent of women eat at the same three restaurants each month. The data draw from 500 million profiled devices and 3 billion location tags, resulting in a 40% performance lift over geofencing alone.
“If you’re targeting someone in Grand Central Station, is it a cab driver? a backpacker? a commuter? That’s where geofencing falls down.” — Sense Networks CEO David Peterson
Though privacy flags go up during such discussions, Staas and others are self interested in following responsible data use, such as anonymized pattern detection. Some steer clear in other ways.
“We target the location and not the user,” says PlaceIQ CEO Duncan McCall, describing the strategy of deriving probability from points on the map, rather than user behavior.
Sense Networks, meanwhile, brings together data sets to not only profile user groups, but to innovate new areas such as “place retargeting” — like online display retargeting but in the real world.
“Marketers want to market to people,” says Sense CEO, David Peterson. “If you’re targeting someone in Grand Central Station, is it a cab driver? a backpacker? a commuter? That’s where geofencing falls down.”
This holds up, however a classic local challenge remains: the more you divide audiences, the more you segment yourself out of impressions. That’s an issue for brand advertisers that emphasize sheer reach. (one reason for low advertiser demand).
Staas and Peterson both answer that audience profiling speaks the language of brand advertisers: demographics. This frames it as buying certain audiences — which they’ve been doing for years — under the premise that location defines those audiences.
The scale question can be answered similarly.
“10 percent of mobile advertising has true location targeting,” says Staas. “We can use that to power the location graph and create user profiles, then use it to reach the additional 90 percent of ad inventory. Once I know an individual is a mom, I can reach [her] any time, any place.”
Mike Boland is senior analyst at BIA/Kelsey, where he heads up the firm’s mobile local coverage. Previously, he was a tech journalist for Forbes, Red Herring, Business 2.0 and others.