Inside RetailNext's Loss-Prevention Engine
A cashier rings up a refund, voids a transaction after the fact, or completes a sale with no shopper standing at the register. Loss-prevention leaders now want to know within hours, not weeks, whether the exception is an honest mistake, a training gap, or theft. RetailNext's Asset Protection platform answers by linking each flagged transaction to the camera frame that captured it, then ranking stores and cashiers by risk so investigators start where the evidence is heaviest.
The platform runs on two interlocking systems. The first is AI behavioral analytics, which RetailNext describes as combining "best-in-class video analytics, Wi-Fi detection, on-shelf sensors, and data from point-of-sale systems and other sources" to track how shoppers move through stores. The same sensor stack that counts foot traffic and draws heat maps has, in some form, been watching the register for more than a decade.
The second system is POS exception reporting, and this is where Asset Protection does its loss-prevention work. Rather than streaming every transaction to a reviewer, the platform surfaces a defined set of high-risk events (cash refunds, post voids, discounts, no-sale events, and transactions with no shopper present at the POS) and links each one to the corresponding camera view. The result is a single interface where investigators move from alert to footage without tab-switching.
That single-interface design predates the Asset Protection branding. When RetailNext shipped RetailNext 4.0 in June 2013, the release already carried "enhanced POS exception reporting" with "Top-N queries for the most important metrics" and the ability to "export selected lists of POS transactions along with associated video events." It added legacy DVR and NVR integration so retailers could pull video from existing camera systems without buying new hardware, plus full mobile video support on iOS and Android. The 2013 announcement came from the NRF Loss Prevention Conference, where the company moderated a panel on "How Linking In-Store Analytics to LP Data Uncovered Lost Profit Opportunities."
The financial backdrop makes these efficiency gains material. Retailers including Target, Walgreens, and Home Depot have publicly leaned into AI-powered detection. In a CNBC Television segment, a former director of Walmart's asset protection division noted that "1% of your inventory if you're at the volume of Walmart or some of the major tier ones, that's a staggering amount of money," and pointed to Lowe's absorbing roughly a billion dollars in shrink. The same segment put U.S. retail shrinkage at $112 billion in 2021, with half of that figure attributed to organized retail crime, a number described in the discussion as an overgeneralization.
Adoption has tracked the dollar exposure. RetailNext's install base, more than 560 brands across 100+ countries, drawing on more than 65,000 sensors in retail stores, and tracking more than 500 million shopper trips a year, per the company's Black Friday 2025 release, is also why RetailNext traffic reads are treated as the holiday bellwether by outlets like CNBC. The install base is the moat.
The early-traffic data from Black Friday 2025 hints at how the retailer base is using the platform in practice. Apparel was nearly flat on Black Friday itself at −0.7% year over year, while discretionary categories like home fell 9.8% on the weekend average and health and beauty dropped 4.7%. Saturday traffic dropped 8.6% across the board, with the Midwest off 42.1% after a snowstorm. Those are the kind of weather- and tariff-driven swings an exception-reporting layer is built to absorb: anomalies get flagged, ranked, and routed to the right store and cashier instead of getting lost in the noise of a busy weekend.
RetailNext's majority growth investment from Battery Ventures in January 2025, covered in a BusinessWire announcement, signals that the install-base expansion is being capitalized for the next phase of product rollout.
Why Solution Architects Are the New Bottleneck
The Asset Protection platform's spread is showing up where retailers feel it most: in job postings. With deployment cycles pulling LP teams away from manual video review and toward data-driven exception handling, retailers are hunting for people who can stand up the system, wire it into existing POS and video stacks, and tune its risk rankings once it's live.
RetailNext's footprint explains the pressure. That's not a single-installation sale. It's a fleet of integrations, where each store has its own POS variant, camera vendor, and loss-prevention workflow, and each one needs someone who can configure exception rules, validate behavioral models, and translate findings into action for store managers and LP investigators.
Once LP leaders see exception-flagged transactions paired with the corresponding video clip in one interface, searchable and ranked by risk, the bottleneck moves from finding evidence to acting on it. Acting on it requires architects who can map a retailer's existing transaction codes to the platform's exception taxonomy, decide which high-risk flags warrant real-time alerts, and build the dashboards that LP analysts actually use. That's a specialized profile: equal parts data engineer, retail operations analyst, and integration consultant.
Vendor consolidation in video surveillance and RFID has been pushing retailers toward unified platforms that bridge AI analytics with physical security. Each unification project creates a fresh demand spike for architects who can translate between legacy DVR/NVR infrastructure and cloud-native analytics, which is the exact gap RetailNext's legacy DVR integration capability was built to fill back in 2013.
Solution architects are in demand across industries as companies push to put technology to work for innovation and efficiency. Within retail-tech security, the role is unusually specific. A generalist SA can land a deployment; a retail-LP SA can land one that actually reduces shrink. The differentiator is domain knowledge: knowing which POS exception types correlate with employee theft in apparel versus grocery, how to calibrate behavioral models across regions with different shopper patterns, and how to present risk rankings to investigators evaluating dozens of cases a week.
For retailers evaluating the platform, the practical question isn't whether the AI works in a demo. It's whether they can staff the team to operationalize it. That same January 2025 round, earmarked for international expansion, product innovation, and acquisitions, signals that the company expects that hiring pull to intensify. The bottleneck is shifting from technology to talent, and retailers that move fastest on the recruiting side will be the ones that convert shrink reduction into a durable margin line.
What the Incumbents Are Doing Differently
The loss-prevention market has not stood still as RetailNext extended its behavioral-analytics footprint. Axis Communications, Checkpoint Systems, and Hikvision are the incumbents most often named alongside it in vendor rosters, and each is leaning harder into the same AI-on-the-edge pitch that now defines the category. Genetec is also active in the convergence of physical and IT security. The vendors chasing retail shrink are increasingly pushing AI-native features as the differentiator.
The competitive pressure cuts both ways. RetailNext's counter is its sensor hardware plus cloud retention of high-resolution color video on a SOC 2 Type II–compliant platform, packaged as a single interface for video and POS. It's a stack play rather than a point-solution play.
The wider backdrop tilts further toward incumbents in some uncomfortable ways. Hikvision cameras have absorbed high-profile cyberattack volumes, and Stanford cut ties with Flock Safety, switching to a different surveillance vendor. For retailers evaluating loss-prevention platforms, that kind of news reframes the buying decision: capability matters, but the durability of the vendor, the cleanliness of its security posture, and its track record with privacy regulators now sit on the scorecard next to detection accuracy.
Where the Platform Stops and the Questions Begin
The same data linkages that make AI-driven asset protection effective (pairing point-of-sale transactions to time-stamped video, ranking stores and cashiers by risk, and pushing exceptions to mobile devices) also define the system's boundaries. RetailNext's platform is built to integrate with what's already in a store: POS systems, EAS infrastructure, door alarms, safe openings, and panic buttons. What happens outside that perimeter is the story's edge, and worth naming explicitly.
Privacy posture and data-handling claims
RetailNext positions its Asset Protection platform around cloud video retention, a SOC 2 Type II–compliant infrastructure, encrypted communications, and configurable privacy controls designed to support GDPR and other regional data-protection requirements. Those are vendor-stated controls, not independent audits surfaced in the research, and the distinction matters for any retailer evaluating the platform under a tightening state-level patchwork. Retailers running behavioral analytics on employees and shoppers should expect their privacy, labor, and works-council teams to scrutinize retention windows, access logs, and the scope of "POS exceptions" that include cashier-level behavior scoring.
The broader counter-moves in this space underline why those questions aren't theoretical. Stanford cut ties with Flock Safety and shifted to a new surveillance vendor, and Flock's cameras have become a midterm campaign flashpoint over voter privacy concerns. For retailers, the practical takeaway is that any AI loss-prevention deployment doubles as a cybersecurity and privacy posture decision, not just a shrink decision.
Integration scope: what the platform covers, and what it doesn't
The platform's documented integration surface is purposeful but bounded. It pulls POS data, EAS signals, door alarms, safe events, keypad activity, and panic buttons into a single investigative view; it links transactions to video and surfaces predefined high-risk exceptions (refunds, post voids, discounts, no-sale events, and transactions with no shopper present). It does not, on the evidence in the research, replace upstream supply-chain inventory systems, RFID pipelines, or EAS hardware, which are adjacent layers where competitors like Checkpoint Systems, Axis Communications, and Zebra Technologies operate.
It also matters to be clear about the metric the platform owns. Reporting shrink in dollar terms without naming a deployment invites the kind of comparison the research doesn't support.
What this article deliberately does not cover
A few adjacent topics will appear elsewhere in coverage of AI in retail, and they are out of scope here. Holiday traffic reads, including RetailNext's own Black Friday weekend figures, which showed U.S. in-store traffic down 3.6% year over year on Friday and 8.6% on Saturday, with the Midwest off 42.1% on Saturday after a major snowstorm, measure shopper behavior, not loss prevention, and don't speak to how Asset Protection performs. Omnichannel fulfillment economics, e-commerce conversion, and the role of AI in pricing or merchandising sit in a different category of retail AI and aren't addressed by this platform.
What remains in scope is narrower and more useful: the privacy and integration constraints that determine whether a RetailNext Asset Protection rollout actually delivers, and the explicit acknowledgment that shrink-reduction percentages, cybersecurity exposure, and adjacent AI applications belong in separate reporting rather than being folded into a platform-specific story.
Working in frontier tech? Zero G Talent tracks the openings: see every open ASML role, browse frontier tech jobs, openings at Stripe, and the people building the field.