Are Centroids Misleading Your Advertising? (And What is a Centroid?)

In location-based advertising, accurately determining where your target users are throughout the day is a crucial factor in a campaign’s success.

Increasingly, location data is a key ingredient in mobile advertising. BIA/Kelsey recently estimated that U.S. mobile ad spending on local, or location-targeted, placements would rise 56% this year to $6.7 billion, or 37% of all mobile ad dollars.

Advertisers are increasingly delivering personalized and relevant experiences because they can use location to know exactly where their consumers are, right? Not as often as you might think. Ad tech, it turns out, is actually relying heavily on a fuzzy location measure that has the potential to weaken location-based ad campaigns, resulting in wasted ad impressions – and money.

The basic premise of all location-based marketing is delivering a relevant message, to the right person, at the right moment in time and in the correct location. There are two ways that a publisher can obtain someone’s location: Using an IP address or via a device that has location services turned on (on-device location). While both of these methods result in a latitude and longitude, only one is the accurate and precise location of a device.

Eighty four percent of location is obtained by publishers using an IP address. This method is neither precise nor accurate. Most of the time, the latitude and longitude returned using IP location is the center of a civic area – called a centroid – with a large margin of error.

As the name implies, centroids are often the center point of large areas. The average margin of error range is a square mile, but there have been scenarios where centroids are placed in the middle of the country and in the wrong state.

Centroids are largely to blame for muddying the waters of location-based advertising – meaning that determining the location of any given person is fuzzy, rather than precise and accurate. When most people think of location they think of an exact point that is both precise and accurate. In reality, location has levels of precision and not all location data is created equal.

Remember, we’re trying to get the right message to the right person in the right place at the right time. If place is fuzzy, it’s the digital equivalent of dropping leaflets from a plane into that area. On-device location enables advertisers to know the accurate and precise location of the user’s device. This means you can have finer level detail of the place such as venue category, name and address.

Assuming people are at the location of a centroid makes for an absurd scenario – here are a few centroids that show up around the country:

• In New York City, centroid logic determines that the city’s residents spend their whole day at a Japanese restaurant called Kanoyama, then they magically teleport home
• In Dallas, everyone is apparently hanging out at the Brown Sugar Spotlight Club
• In Denver, the city’s population can be found at the ninth Hole 9 on a Golf Course
• In Iowa, everyone is spending time together at the Iowa County Home Cemetery

What can advertisers do about centroids?

Ad tech companies can use centroid location to build only high-level personas and target campaigns on a regional level with localized offers. But, even when using this region-level location, what do you really know about a person’s whereabouts?

On-device location data delivers location specificity at a much more granular level to derive hyper precise locations.

Here’s just one examples of the difference a couple of decimal places can make when determining location: A user attends a baseball game at Fenway Park. A latitude/longitude with four decimal places would position a user at Fenway Park – practically on top of the famous Green Monster. But, if the latitude/longitude is truncated or rounded down to two decimal places, we’ll be led to believe that the user is actually standing on Huntington Ave, right in the middle of Northeastern University and steps away from the Museum of Fine Arts, Boloco and Au Bon Pain. The specificity is gone and the user’s movement is nebulous.

Any profiling of this user based on the two digit latitude/longitude will be incredibly misleading. Meanwhile, if the location was derived from an IP address that provided a latitude/longitude for the zip code associated with Fenway Park, the user will be placed in the middle of a centroid associated with that zip code, in this case the Residence Inn.

Brands and agencies assume that location enabled impressions include accurate and precise location, but some publishers may be providing inaccurate location data without even realizing it.

Putting useful contextual power tools like venues, geofencing and personas on top of a data point makes it possible to provide better experiences and targeting. The mistake is thinking that on-device location and centroid can both provide the same level of granular targeting.

Mike Schneider is CMO of Skyhook Wireless.