Using Placed Optimization, which enables programmatic in-flight optimization, AdTheorent is able to develop predictive models that identify consumers with the greatest likelihood of visiting a retail location. The goal is to increase in-store visits, lower the cost per store visit for advertisers, and drive incremental customers into stores.
Apart from the retail sector, the real-time capability also benefits restaurants and auto brands, according to Josh Walsh, co-founder of AdTheorent.
As part of the integration, AdTheorent will receive data on a regular basis from Placed about features that drive consumers’ off-line behaviors. The data is used to inform AdTheorent’s predictive models, resulting in campaigns optimized by many data points including demographic, psychographic, geographic and creative variables.
The predictive modeling is different from geo-fencing, which is the practice of establishing a virtual perimeter around a radius such as a store location. With geo-fencing, when a consumer target enters the geo-fenced area, a trigger is sent and an ad is shown. “In addition to the geo-fence that’s set up based on campaign parameters, AdTheorent layers predictive targeting within the geo-fence,” Walsh told Real-Time Daily by email. “As such, AdTheorent’s machine learning platform identifies consumers within the geo-fence with the highest percent chance of visiting the desired location, and serves the ad to those consumers.” The predictive targeting is more effective in driving visits since it consists of more informed targeting vs. a geo-fence alone, Walsh said.
With Placed, AdTheorent late last year conducted a six-week campaign for a national retailer to drive in-store visits using predictive geo-targeting that delivered more than 87 million impressions. AdTheorent was able to grow in-store visits by 59%, at a cost of 33 cents per visit.
Walsh attributes the success of the campaign to data optimization. “AdTheorent’s integration with Placed allows us to receive ongoing data throughout the campaign flight that is used to inform our predictive models. This allows us to optimize using a myriad of data points such as demographic, psychographic, geographic and even creative type,” Walsh said.