Overview

Topsort provides in-store ads solutions that integrate with existing offline display technologies, allowing retailers to show ads on their in-store screens. This system can leverage Topsort’s auction engine or dynamically adapt ads based on audience demographics.

Core Capabilities

Topsort’s platform is designed for scalability, capable of processing high-volume data feeds for real-time and post-facto attribution, managing large data streams for various retail media components like catalog, price, and ad inventory.

When a user can be identified across both online activities and in-store purchases, such as through loyalty programs, attribution for in-store sales can be computed based on prior online ad exposures. The platform supports both real-time and post-facto data ingestion from loyalty programs and POS systems to provide full funnel attribution, and it can integrate with existing data lakes.

The solution also offers computer vision-based metrics for in-store audiences, including total presence and views within defined ad zones. It tracks customer behavior and journeys within the store using people detection and tracking technology, which is crucial for understanding customer paths and computing in-store attribution.

Integration and Implementation

Clients have the flexibility to integrate only with Topsort’s in-store ad capabilities. Implementation can begin when the client’s screens are set up and configured. The system operates on a high-level API designed to be independent of the specific underlying measurement technology used in the store.

Presence events, which are the in-store equivalent of online impressions, are generated by the measurement technology layer and sent to Topsort for audience and attribution calculations. Purchase events from in-store points of sale are reported with an added placement field to identify the specific point of sale.

Privacy Considerations

Privacy is a key consideration, and the system employs opaque track IDs and short-lived data structures to track individuals across different zones without storing images. The attribution methods are designed with privacy in mind, aiming to obscure the direct relationship between in-store tracking IDs and personal user IDs.