Sensormatic Solutions Adds Machine Learning to Shrink Analyzer

Sensormatic Solutions, the global retail solutions portfolio of Johnson Controls, continues to improve the capabilities of Shrink Analyzer, its cloud-based, loss prevention (LP) application.

The application’s recent development efforts enhance internal theft analysis, using machine learning (ML) to continuously help improve loss event identification accuracy, accelerate evidence-building processes to aid in prosecution and support early detection and remediation. Attendees of NRF PROTECT 2025 explored these enhancements in-person.

“To effectively address total retail losses, retailers’ data ecosystems need to have context,” said Tony D’Onofrio, president at Sensormatic Solutions. “They need to be able to see all areas of operations—from the floor to storage and beyond—and evolve alongside changing needs. Shrink Analyzer’s new ML capabilities represent a significant breakthrough in how Loss Prevention teams can identify, validate and accelerate case-building efforts.”

Since its launch last year, Shrink Analyzer has been helping retailers develop deeper understandings of how shrink happens within stores, the items most at risk and the ways employees contribute to total retail losses.

Using radio-frequency identification (RFID) tags and sensors for item-level data, Shrink Analyzer can turn existing analytics infrastructure into powerful LP insights to help retailers enhance the value of their investments.

The addition of ML enables Shrink Analyzer to learn from what’s happening and help separate significant loss events and notable patterns from the noise of day-to-day operations. As a result, the enhanced solution will help retailers to:

• Boost data integrity: RFID technology captures hundreds of data points per second—far more than any person can sift through. Shrink Analyzer’s new ML-powered Shrink Confidence Score helps simplify aggregation and data cleaning processes to increase accuracy and boost case values—a top priority for 49% of retailers using RFID for LP, according to a new study with VDC Research.

• Identify and address suspicious store associate behaviors: The Sweetheart Detection feature brings the financial impact of sweethearting and self-checkout scan-avoidance—as well as the identities of repeat offenders—into focus. By pairing ML technology with item-level sales, inventory, exit and associate/shopper behavior data, Shrink Analyzer can enhance retailers’ ability to identify evolving tactics and take action to stop them.

• Improve investigation efficiency and outcomes: Shrink Analyzer’s ML capabilities can identify previously hard-to-detect loss events, which can be used to index video and automate case building. This helps drive productivity and can help accelerate law enforcement processes.

As with previous iterations, the newly enhanced application is inventory-platform agnostic, is compatible with Sensormatic Solutions suite of RFID hardware and can be integrated alongside any retail analytics ecosystem. Sensormatic Solutions plans to make these ML-based capabilities available later in 2025.