Ad Attribution via Geolocation Data Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Advertisers face challenges in determining the impact of online ads on in-store visits, as existing methods are ineffective in tracking visits that do not result in purchases, limiting their understanding of ad effectiveness.
Innovation Solution
An attribution system that uses geolocation data from mobile devices to attribute ad impressions to physical visits to target places, leveraging data collection and inference pipelines to determine the probability of visits and time spent at these locations, and incorporating third-party data to enhance attribution metrics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If email address reconciliation is used to track ad impact, then purchase behavior can be captured, but non-purchase store visits cannot be measured
Solution Approach 1:
The patent applies universality by using geolocation data to track multiple types of consumer behaviors (store visits, purchases, time spent) through a single attribution system, rather than relying on email reconciliation which only captures purchases. The system universally attributes various consumer actions to ad impressions across different devices and channels.
Solution Approach 2:
The patent uses geolocation data as an intermediary to connect ad impressions to offline store visits. Instead of directly linking email addresses to purchases, the system uses location data from mobile devices as a mediator to attribute store visits to specific ad exposures, enabling measurement of both visits and purchases.
2Loss of information
If cross-device tracking is implemented, then ad impact on offline behavior can be measured, but data collection complexity increases
Solution Approach 1:
The patent extracts geolocation data from mobile devices as a separate, trackable element to attribute offline behavior to online ads. By extracting and independently tracking location information, the system can link ad impressions to store visits without requiring complex cross-device identification across all consumer electronics.
Solution Approach 2:
The patent changes the measurement parameter from email address matching to geolocation-based attribution. This parameter change simplifies the tracking mechanism by using location data that is naturally captured by mobile devices, rather than requiring complex device fingerprinting or user identification across multiple devices.
3Measurement precision
If geolocation data collection is expanded to all users, then attribution accuracy improves, but privacy concerns and data security risks increase
Solution Approach 1:
The patent implements partial action by collecting geolocation data only from a panel of users who have opted in, rather than universally tracking all consumers. This partial sampling approach provides sufficient attribution data to measure ad impact while limiting the scope of data collection and reducing privacy concerns.
Solution Approach 2:
The patent applies self-service by requiring users to voluntarily join the geolocation data collection panel, giving them control over their own data. Users who opt in are actively participating in the measurement process, which reduces privacy concerns compared to covert tracking, and allows the system to function with a representative sample rather than requiring universal participation.
Data Source
AI summary
A system and method for attributing in-store visits to exposure to advertisement (“ad”) impressions associated with an ad campaign are disclosed. The system gathers impression data and uses that data to identify users who were exposed to the ad impressions. The system then uses location data, activity information and in some instances beacon data points reported by mobile devices of the impression users to determine if the impression users visited a target place during a conversion window. Based on the impression users who were exposed to the ad impressions, the system establishes a control group of users who were not exposed to the ad impressions to perform quasi-experimental analyses to assess whether the ad impressions had any impact on changing the impression users' physical in-store visitation behavior.