Retailers are suffering significant financial losses as a result of an increase in internal theft and organized retail crime (ORC). Shoplifting incidences in the US increased by 93% in 2023 compared to pre-pandemic levels, indicating a huge spike in retail larceny. According to the National Retail Federation, retailers reported an average of 177 theft offenses every day, with other industries reporting over 1,000 cases daily.
The British Retailers Consortium (BRC) reports that retailers have voiced their displeasure with the police response to these occurrences, in addition to the general rise in offenses. According to the BRC study, retailers reported more than one-third of violence and abuse occurrences (171,608) to the police in 2022–2023, with an estimated 8% (38,135) leading to prosecution and 19,067 to conviction.
According to a report from the ONS Centre for Crime and Justice in the UK, shops cited a lack of staff time and a lack of expectation that reporting such instances would have any impact as their primary excuses.
Matthew DeMello, the editorial director of Emerj, recently spoke with Joe Troy, the senior risk manager at Amazon, and Mike Matta, the CEO and co-founder of Solink, about the transition from conventional video surveillance to AI-powered systems that serve as real-time decision engines for retail operations.
Their discussion demonstrates how surveillance is transformed from passive monitoring to an intelligent, real-time system that improves decision-making, lowers risk, and generates business value across departments thanks to AI-enabled video.
- Prioritizing signals to cut down on noise: Loss prevention teams may concentrate on handling actual occurrences rather than managing noise by using AI to filter and rank security data.
- Accelerating risk detection with AI: Real-time alarms delivered by AI-powered video significantly cut down on investigation time and stop losses before they get out of hand.
Signal Prioritization To Lower Noise
Episode: Using AI-Driven Video Security to Combat Retail Crime and Prevent Losses — with Mike Matta of Solink
Mike Matta, co-founder and CEO of Solink, is the guest.
Knowledge: Business development, leadership, and entrepreneurship
Brief Overview: Mike has extensive experience in the security and technology sectors. He worked on business development and technology integration in key positions at Bell Canada and Salesforce.com before to co-founding Solink. He graduated from the University of Waterloo with a Bachelor of Applied Science (BASc) in Electrical Engineering.
Security cameras now serve as “Google Analytics for physical spaces,” providing insights into operations, IT, and human resources, Mike explains at the beginning of the program. Businesses can spot trends like understaffing or queue abandonment and act quickly by gaining visibility. He asserts that cameras are now essential to vision-enabled operations, which facilitate more intelligent, cross-functional decision-making, and are no longer merely a component of security.
He argues that instead of relying on costly, difficult-to-implement specialist solutions, CEOs are increasingly embracing this video data to evaluate customer flow, product movement, and interaction patterns. Video becomes a straightforward yet effective source of departmental knowledge without the need for additional hardware.
He describes how video has emerged as the most important sensor in contemporary retail settings, offering proactive threat response, real-time operational insight, security, and informed decision-making.
He explains the shift by highlighting a number of crucial elements:
- Cameras are becoming the most valuable sensor in stores.
These days, cameras offer a rich data feed that displays not just what is happening but also how customers act within the store. They essentially act as the company’s eyes by providing a real-time feedback loop that is not possible with POS or inventory management systems. - Using AI to scale awareness.
Even though cameras record everything, nobody can view hundreds of streams from different stores. AI helps merchants focus on the correct situations without becoming overloaded by surfacing just pertinent material. - Using generative AI to transform video into a workforce.
Generative AI on video allows the system to act on the data it is analyzing in addition to providing alerts. It turns video from a passive sensor into an active virtual employee by identifying occurrences, interpreting events, and even speaking through speakers to discourage crime. - Using confirmed threats to replace bogus alerts.
False positives from traditional alarm systems frequently result in low-priority police responses. Agents may now visually verify actual threats in real time with AI-verified video alarms. Law enforcement can react more quickly, which leads to quicker arrival times and higher rates of apprehension.
In the end, he clarifies that although retail settings provide a great deal of data, the difficulty lies not in a lack of data but rather in having the right data at the right moment. Security personnel may now concentrate on actual dangers rather of being distracted by irrelevant occurrences or false alarms thanks to AI’s ability to prioritize and filter information.
Today, handling noise—false alarms, insignificant occurrences, and incidents that hardly ever call for action—is the primary responsibility of a security leader. Typically, only around 5% of what appears is a genuine signal that has to be addressed. We can reverse that ratio with AI.
AI enables loss prevention and security professionals to concentrate nearly totally on significant concerns by prioritizing and eliminating noise. The skill set needed is radically altered by that development. Teams make better decisions and have a greater effect when they devote most of their effort to responding to actual problems rather than triaging false positives.
— Mike Matta, Solink’s CEO and co-founder
Using AI To Quickly Detect Risk
Episode: Joe Troy of Amazon Discusses the New Era of Loss Prevention with Video Intelligence
Joe Troy, Senior Manager of Site Risk at Amazon, is the guest.
Knowledge of Risk Management, Asset Protection, and Loss Prevention
Brief Recognition: Joe has over 20 years of expertise and is skilled at establishing standards to improve customer satisfaction, discourage theft, and boost sales. He has previously held a variety of positions pertaining to asset security and loss prevention with well-known businesses like Rent the Runway, J. Crew, Walmart, and Toys R Us. He has a Master’s degree from the University of New Haven and a Criminal Justice degree from Temple University.
According to Joe, traditional surveillance is frequently seen as labor-intensive, reactive, antiquated, and connected to manual video monitoring. However, AI is altering that. It’s changing the way physical industries like retail, hospitality, logistics, and facilities management deal with staffing, safety, and investigations.
He outlines three main advantages of AI: it can expedite video review by automatically labeling clips, provide real-time safety and fraud alerts, and free up personnel to concentrate on service and training. Additionally, the change improves customer happiness and worker morale by allowing security teams to interact with employees more.
Joe describes how the use of in-store data has changed over time, moving from manual tools like traffic counts and heat maps to more sophisticated, AI-powered video analytics. Traditional tools were unable to differentiate between a person, an object, or a compliance issue, but they could display where customers loitered or how they moved across aisles. Businesses may now determine whether an object is a regulatory violation, a safety problem, or a potential consumer by integrating video into the system.
Retailers can now make better judgments on aisle layout, shelf placement, and store operations thanks to the increased visibility, which is a major improvement over decades-old technologies.
Joe lists several crucial tactics for using AI-enhanced video solutions in companies that can be reluctant or change-resistant:
- Present It as Facilitation Rather Than Enforcement:
Present AI as a tool to assist teams rather than as a way to keep an eye on them. Its purpose is to eliminate uncertainty from decision-making by demonstrating what is effective and correcting what is not. - Connect AI to Specific Business Objectives: To begin, pinpoint a specific, customer-focused issue, like increasing foot traffic or streamlining line management, and show how AI directly solves it.
- Pilot and Create Champions: Start tiny pilot programs and acknowledge initial successes. Make use of those success stories to develop internal advocates who can discuss how AI improved their outcomes or saved them time.
- Make the Data Actionable Right Away: Don’t merely display dashboards; instead, directly connect insights to a choice the team can make this week, particularly if it fits with their performance indicators.
It’s critical to realize that the goal is to remove manual labor so that teams can concentrate on the things that really drive the company, not to replace human intuition. AI-powered video is a benefit rather than a danger.
With careful implementation, it develops into a cross-functional intelligence layer that facilitates improved operations, marketing, training, and customer experience decision-making.
-Joe Troy, Amazon’s Senior Manager of Site Risk