Location “Crop” Circles

By Saiyam Shah

Location intelligence reveals a great deal about consumer preferences and interests. This real-world location data is the bridge to understanding consumer behavior.

For location data to be meaningful for advertising, it must be tied to locations that are the union of advertisers and consumer spending. Valassis’ Lifestyle Map is highly deterministic, in that, multiple data sets are connected to granular parcels, connecting data points to devices.

A key factor that distinguishes Valassis as an industry leader is our ability to identify fraudulent activity and abnormal data. This includes isolation and removal of location signals that include appended random numbers or appended fixed numbers or artificial information. These fraudulent adjustments to a location signal will appear to display a device in an inaccurate location. If a marketer uses this data for a campaign, they may not reach their target audience or meet the performance objectives of their campaign. Valassis has processes and algorithms in place that continuously monitor and correct these anomalies thus allowing us to keep our clients’ best interests front and center.

Beginning in October 2020, the Consumer Graph Engineering and Data Science teams at Valassis observed certain mobile applications across different OS platforms sending inaccurate location signals to gain more revenue for opportunities. We believe we are the first in the industry to notice this anomaly. Several opportunities have been attributed to ellipse-shaped points in urban and rural areas despite no true and accurate information being available from these opportunities. This inaccurate information could lead to wasted spend, focusing impressions away from the in-market audience and ultimately not meeting the clients’ needs. We started monitoring this pattern closely and have noticed it spreading since the past October and being used by more applications with time.

Figure 1: Demonstration of location points being wrongly placed

Some of the examples of the plots that demonstrate this aforesaid behavior are included below:

Figure 2: Plot showing ellipse-shaped points in a small portion of a city

Figure 3: An enlarged picture of the location points from the above plot showing the formation of an ellipse

Figure 4: A much wider area displaying several circles throughout all the regions from a single application

As seen in the above images from our observed data these “crop” circles are easily visualized! To address this problem, Valassis has designed an algorithm that finds applications contributing to such location-based anomalies. This algorithm applies techniques from graph theory and computer vision to novel geospatial data structures. With this algorithm we can identify all such applications and clean out anomalies from our location intelligence products.

Figure 5: Before (with anomalies)

Figure 6: After removal

With this cleaned data, marketers are assured that their reach to the customers is not compromised in any manner and opportunities reach the right consumers.