March 29, 2017

Mobile Marketing 2.0: It’s Not Just Where You Are But Whom You’re Near

Assoc. Prof. Zsolt Katona
Assoc. Prof. Zsolt Katona

By Pamela Tom

How GPS network science is evolving to help mobile marketers prosper

Companies love knowing where potential customers are hanging out at any given moment. By “geo-fencing” or using GPS data, companies can target customers based on their exact location and send promotional messages directly to their phones.

Location data can also help to discover customers’ personal preferences. Research now shows that consumers are more likely to respond similarly to a mobile marketing offer or coupon if they have recently been in the same physical place.

A new study found that location data is a better predictor—up to 19% more accurate—of consumer behavior than information about demographics (age, income, education) and psychographics (values, lifestyle, and personality).

Zsolt Katona, an associate professor of marketing at UC Berkeley’s Haas School of Business, calls this response a “co-location effect” in his study, “ Predicting Mobile Advertising Response Using Consumer Co-Location Networks.” The paper is co-authored with Peter Pal Zubcsek of Tel Aviv University, and Miklos Sarvary of the Columbia Business School.

The researchers hypothesize that redemption responses may be similar because people with similar tastes typically visit the same places. “If you have similar tastes, you are more likely to go the same location, even if you do not actually know each other," says Katona. "That may cause you to respond similarly to the same coupon. We saw the biggest effect with people who are closely located, within a roughly 400-foot radius.”

The researchers used GPS data obtained by a major Southeast Asian cell phone company and modeled how 217 study participants in a metropolitan area responded to a variety of coupon  offers for coffee, food, and entertainment sent out at random times and random locations.


While like-minded consumers may visit similar locations, the co-location relation between consumers is not necessarily social. That’s good for marketers who don’t have access to people’s social networks. In order to study how well customers respond to coupons while controlling for similar habits of socially connected individuals, Katona and his colleagues constructed two participant networks: co-location and referral.

Co-located participants must have been at the same location during at least one of the GPS observations the day before the offer.

Illustration: Dulce Lopez

Participants in the referral network were deemed connected if one of them had invited the other to the program.

When participants recently visited the same location and received the same mobile coupon, they reacted to the offer at a higher rate and with more similar responses than those who were not near each other. For instance, a participant in the proximity of someone with a 20% coupon redemption rate (in the consumer packaged goods or CPG category) redeemed the same offers at roughly double the rate of participants who had not visited the same location.

The research team also conducted similar tests to determine if consumers are more or less likely to redeem offers in a similar fashion if they had met in so-called “hot spots”—popular and populated locations such as those in urban centers—compared to “cooler” or less-visited locations.

The study found that it is easier to predict customers’ behavior when they have visited the same cool spots rather than the same hot spots. By definition, hot spots attract everyone irrespective of tastes and preferences.

In contrast, when people tend to go to the same cool or non-popular places, they are more likely to be similar in their preferences.





Topics:    Research News