Skip to content
Microsoft Industry Blogs

The National Retail Federation completed a loss prevention survey of 63 retailers in 2019. The top two risks (and therefore priorities) for 50.8 percent of retailers in the last five years were internal theft and cyber-related incidents (such as data breaches). Retailers reported an average inventory shrink of 1.38 percent, which is on pace with historical norms and equates to approximately $52.4 billion annually. As a result, loss prevention (LP) teams are pulled in many directions as they combat various types of shrinkage.

But, the problem in the stores and warehouses is even more acute. In fact, 42.9 percent of retailers indicate that they have noticed a greater increase in in-store fraud incidents, which impact sales. To combat these challenges, best-in-class retailers are making investments in innovative AI, especially computer vision systems. The technology promises to not only cut out the losses but also ensure the safety and security of customers and associates. By using predictive patterns, image recognition, and intelligent item tracking, retailers are reducing the loss of products due to theft or error. The added bonus is a better shopping experience for customers and associates.

Even though POS analytics is being used by 55 percent of retailers, other emerging LP tools are gaining in adoption. Such tools include computer vision, intelligent item tracking via RFID, electronic security tags, weight sensors, as well as artificial intelligence. These tools are helping lower shrinkage while enhancing the in-store shopping experience. Advanced analytics and IoT allows for much more. The following use cases are helping retailers embed new and more secure processes in their stores:

  1. Intelligent and artificial intelligence-backed LP insights in automated/semi-automated stores and warehouses: As retailers gradually start introducing more semi-autonomous stores and warehouses, they will need to track multi-sensory and multi-device data. Such diverse data will lead to predictive intelligence that stores (and warehouse teams) can act upon to prevent fraud incidence before it happens. Field employees can stop fraud incidences before incurring large amounts of inventory or financial losses. Such proactive and predictive tracking relies on data collected from store traffic, shelf-level activities, and employee movement (in stores and warehouses). Data can be gathered from many sources: cameras, Bluetooth devices, shopping basket weight sensors, wireless access points, mobile devices, robotics, infrared, and blockchain applications. Applying advanced analytics will help retailers track and trace fraud incidences in the aisles, backroom, front of the store and other areas within warehouses/stores. It’s no surprise that 62 percent of retailers are hiring LP talent with analytical skills, followed by 40 percent who have skills related to executing effective cybersecurity. (The question of privacy must be addressed. Conscientious enterprises such as Microsoft already put a premium on ethical AI.)
  2. Upgrading intelligent LP assets in legacy stores and warehouses: According to the survey, the top three LP systems in use are burglar alarms (92 percent), video recorders (84 percent), and armored car deposit pickups (68 percent). These legacy LP systems have undoubtedly proven effective in putting higher-level LP processes in place within stores and warehouses. But such systems are reactive, and mostly used for forensics. Retailers need to invest in new data sensors. They must also implement intelligent LP processes that integrate new (and old) devices. The changes include multi-sense, serverless and artificial intelligence-led approaches. The results are transformative changes to store and warehouse loss prevention. Fraud is stopped in its tracks.

To reduce LP threat incidences and execute proactive LP processes in stores and warehouses, retailers have two tasks. First, upgrade stores and warehouses to include intelligent devices. Second, they must develop new AI and analytics processes that evolve continuously (as fraud will also evolve.)

Download our latest playbook covering AI.