Massive Mobile Data – Not All Sources are Created Equal
What is Massive Mobile Data?
Retailers, restaurant, and service providers are abuzz over an emerging technology, known as Massive Mobile Data (MMD), and its uses and applications with respect to better understanding customer profiles, shopping habits and patterns, customer residence and workplace locations, cross-shopping, target marketing, assessing staffing requirements based on peak activity periods, etc. As we blogged about earlier this year, massive mobile data consists of geographic observations regarding the location of consumers based on two primary data sources:
- Cellular data (think AT&T, Verizon, et al.)
- Location-enabled applications (in phones, in vehicles, etc.)
Theoretically, this data tracks the whereabouts of cell phone users and vehicles over time, including whether they have visited your stores or restaurants, as well as any/all other stops they made to and from your locations. While many articles have been written, and many presentations have been made touting the theoretical benefits of this data, the purpose of this blog is to help the reader come to grips with the very significant differences related to how MMD is collected and processed amongst various compilers, and the implications of these differences – differences that directly impact the accuracy and ultimately, the value associated with such data.
Not all MMD is created equally.
Most providers of MMD rely at least in part on their ability to track consumer behavior via their cell phones by leveraging cellular data from carriers. In its most primitive state, tracking is estimated via a triangulation process which seeks to determine a consumer’s location based on their positioning within the network of cell towers. Unfortunately, this approach is only accurate (on average) to within 100 meters approximately 50% of the time. The primary implication of this accuracy limitation is that a cell phone at the center of a busy commercial intersection could suggest that its user is patronizing any or all of the retail and restaurant developments located at the intersection, and provides no perspective as to whether the user patronized your location.
Significant improvements to accuracy are achieved when utilizing location-enabled applications. These apps can leverage GPS, location beacons, and Wi-Fi to track consumer locations. Based on the technology employed, app-derived location detail can be accurate within 10 feet of a consumer’s actual position! That said, not all MMD compilers/providers are created equal. Two primary concerns that should be considered:
- Several MMD providers rely on apps that must be a.) open and b.) have an engaged advertisement operating within the app to track the locations of consumers. Data leveraged from such providers would not be able to track a consumer that visits a store or restaurant but is not using the relevant app. Reliance on the utilization of apps that must be being open and in use when tracking the location of consumers dramatically undermines the utility of such data as it only provides “snapshots” of consumer locations when the app is being leveraged.
- Other MMD providers leverage applications that are embedded in “smart cars” and fleet vehicles. While data derived from these sources can provide a robust, real-time source for automobile traffic (a welcome alternative to the infrequent collection of data by way of rumble strips), it is not helpful in identifying those consumers who set foot within a store or restaurant.
Extensive diligence by Intalytics revealed one MMD provider (Cuebiq) that has surmounted this shortcoming by developing an SDK (software development kit) embedded in apps that provide an ability to ping a mobile device for its location regularly throughout the day regardless of whether or not the app is in use. This approach enables Cuebiq and Intalytics to understand the whereabouts of anonymous users - be it at home, in the workplace, at an entertainment venue, another retailer or restaurant, etc. As early adopters and partners, only Intalytics has access to up-to-the-minute data from Cuebiq. Obviously, current data is mandatory to accurately assess any event that happened relatively recently such as a new store or competitive opening, cross shopping between an anchor and a store, the impact that a marketing campaign had on driving shopping activity, etc.
Further, Intalytics is the only retail predictive analytics company that leverages Cuebiq location data in its rawest form (latitudinal/longitudinal coordinates), enabling our analysis to be as granular and precise as possible. That said, both Cuebiq and Intalytics take a proactive approach to consumer privacy as it relates to this data. As a member of the advertising industry privacy watch group The Network Advertising Initiative (NAI), Cuebiq ensures that all the data it collects meets NAI’s specific privacy regulations and that users’ personally identifiable information is kept out of Cuebiq’s data collection pool. Intalytics has adopted a similarly stringent approach to privacy – at no time will address-level data (lat/long coordinates) about consumers be provided to clients and/or other third parties.
Utilizing MMD, Intalytics immediately realized dramatic improvements with respect to all aspects of our analytical deliverables (e.g., the impact of competitor and sister unit cannibalization, the strength of competitors and co-tenants by brand, customer profiling, the ability to target prospective customers in a more granular fashion, assessing the interplay that exists between a consumer’s home, workplace, shopping and entertainment preferences, etc.) and most importantly, the accuracy of our forecasting models.
Additional benefits of the partnership between Intalytics and Cuebiq will be discussed in future blog posts. Please feel free to contact us to learn more about how this data can potentially be of benefit to your organization.
Updated November 2, 2017