Retail

Turning Shelf Blind Spots into Retail Intelligence

Edge AI on existing phones lifted planogram compliance by 35% across 4,500+ stores and exposed phantom inventory. See how computer vision turns the shelf into a business asset.

Reading Time

6 Minutes

Table of Contents

Authored By

Lukasz Piotrowski

CEO & Founder

Note: This is the English version of the article originally published in “Stacje Paliw – Sprzedaż Detaliczna. RAPORT 2026”, prepared by the editors of Stacja Benzynowa & Convenience Store and Petrolnet.pl. Read the original on Petrolnet.pl.

Running a fuel station today is an exceptionally difficult balancing act. While fuel sales bring traffic to the forecourt, it is the shop and foodservice area that generate real margins and determine the profitability of the entire business. At the same time, operators and chains face enormous pressure: rising labour costs, recruitment difficulties, high staff turnover and ever-increasing customer expectations.

The leadership teams of convenience-store chains often live in a world of polished digital strategy. Headquarters invest millions in advanced ERP systems, predictive algorithms and business analytics. From behind a desk, everything looks perfect: planned promotions, optimised planograms and dynamic pricing. Yet in retail, the truth is not found in head-office system data. It reveals itself on the physical shelf on a Friday evening, when a station is experiencing peak traffic.

This is where the so-called Analog Gap emerges—a dramatic disconnect between what IT systems say is in the store and what customers can actually see and buy. In this harsh business reality, artificial intelligence (AI) and Computer Vision are no longer expensive, futuristic experiments. They are practical, everyday operational tools designed to protect profits, ease the burden of staff shortages and keep shelves in order—today.

The Phantom Inventory Epidemic and the Execution Gap

From Blind Spots to Full Control: How Computer Vision Analytics (Edge AI) Turns the Physical Shelf into a Real Business Asset illustration 1

At the heart of the problem created by the Analog Gap is so-called Phantom Inventory. It is a silent killer of operating margins, costing the global retail sector an estimated more than $1.77 trillion annually, according to IHL Group. Research and industry data are unforgiving: across most convenience-store chains, inventory accuracy at the individual-store level does not exceed 65–80%.

How does this work at a fuel station? The point-of-sale system indicates that five cans of a flagship energy drink are still on the shelf. In reality, they may have been stolen, miscounted during delivery or simply forgotten in the stockroom. Because the digital system “sees” the product as available, the auto-replenishment algorithm does not trigger a delivery. The product falls into a state of “coma” (zero sales, no demand signal), and a frustrated customer leaves empty-handed—often heading straight to a competitor. On top of this comes the Execution Gap: deviations from planograms and outdated price labels mean that even the best head-office strategy can evaporate at the local-store level.

The Edge AI Breakthrough: An Intelligent Assistant Instead of Expensive Cameras

From Blind Spots to Full Control: How Computer Vision Analytics (Edge AI) Turns the Physical Shelf into a Real Business Asset illustration 2

For operators, innovation is often associated with enormous upfront capital expenditure (CAPEX), IT infrastructure overhauls and station-wide disruption. The first generation of autonomous retail (for example, “Just Walk Out” formats) struggled in the market precisely because of the prohibitive cost of installing ceiling-based systems ($100,000–$150,000 per location) and the challenges of cloud-based data processing. Streaming video from hundreds of cameras creates costs and can overwhelm connectivity at fuel stations, which naturally contain “dead zones”, such as walk-in chillers where Wi-Fi does not reach.

The solution transforming the market is Edge AI—artificial intelligence at the edge. For Computer Vision to deliver a real return on investment, we must move from passive monitoring to active operational execution. Modern algorithms run directly on devices the station and its staff already have: several-year-old Android smartphones or standard handheld data terminals.

Image processing takes place entirely offline and in fractions of a second. Instead of carrying out tedious manual audits with a clipboard, an employee performs a five-second video scan of the shelf. The system immediately synchronises shelf status and overlays visual cues using augmented reality (AR): “missing facing for product X”, “price mismatch”. The technology (the StoreOS concept) becomes a “Second Brain” for the cashier, turning passive oversight into prioritised task guidance.

Foodvenience, Team Management and Perfect Promotions

Fuel stations increasingly serve as bistros and cafés, creating challenges around the short shelf lives of fresh sandwiches and ready meals. Unsold products become waste, directly affecting a location’s financial results. Intelligent cameras can work with station systems to analyse expiry dates. Combined with electronic shelf labels (ESL), the system can automatically reduce prices in real time as part of “happy hours” promotions, reducing fresh-food waste by as much as 49%.

Computer Vision also solves the challenge of Retail Media verification. FMCG brands pay stations for dedicated displays and promotional islands. In seconds, an algorithm analyses a photo, verifies “Share of Shelf”, checks for point-of-sale marketing materials (POSM) and detects incorrectly placed products. The station owner gains irrefutable digital proof that promotional contracts have been fully executed, helping secure this additional revenue stream.

From Blind Spots to Full Control: How Computer Vision Analytics (Edge AI) Turns the Physical Shelf into a Real Business Asset illustration 3

With staff turnover at 75–150%, Edge AI is also a powerful organisational tool. We are entering the era of “Agentic AI”, in which an autonomous system can generate suggested orders for the distribution centre after identifying chronic out-of-stock situations. Employees can devote the time recovered to active selling. What is more, deeper image analysis—creating heatmaps—helps optimise store layout, bottlenecks and product assortment in ways that naturally drive cross-selling.

Checkout VAR: Self-Checkout Without Losses

Self-checkout (SCO) reduces the workload on staff, but the statistics are stark: it can generate up to 16 times more losses than staffed checkout lanes. This is caused by mistakes made by customers in a hurry as well as deliberate theft (for example, so-called “sweethearting”). Edge AI can create a discreet, intelligent support system called “Checkout VAR” (Video Assistant Referee). A camera above the checkout continuously analyses hand movements and scanned products—without intruding on privacy or tracking faces. When the algorithm detects a missed scan, the checkout smoothly pauses the transaction and displays a short video message to the customer: “Did you remember to scan this item?” As many as 75% of customers immediately correct the mistake themselves, helping cut losses in the SCO area by 15–30% without stressful confrontations with staff or security.

Proof at Scale: Technology That Works Today

Technological pragmatism is measured solely by scalability. Through OmniShelf StoreOS, we have demonstrated that Computer Vision analytics is ready to work here and now. Together with 7-Eleven Philippines, we deployed Edge AI across an existing fleet of mobile devices in more than 4,500 locations. The operational results decisively closed the Analog Gap:

Achieved a 35% increase in planogram compliance;
Saved more than 70 hours of staff time per location each month;
Recorded a 1.7x like-for-like sales increase in optimised categories;
Employee satisfaction (NPS) rose to 9.2/10, confirming that a significant cognitive burden had been removed from staff.

From Blind Spots to Full Control: How Computer Vision Analytics (Edge AI) Turns the Physical Shelf into a Real Business Asset illustration 4

The world’s best convenience-store chains are not perfect. The difference between market leaders and businesses losing margin lies in how quickly they detect and resolve shelf-execution problems. Let’s stop treating vision technology as an expensive experiment. By putting power in the hands of station staff and giving the physical shelf a digital voice, we can turn retail’s biggest blind spot into our most profitable business asset.

INSIGHTS & UPDATES

Explore More from the OmniShelf Blog

Stay ahead of the curve with deeper insights, product updates, and industry trends shaping the future of retail technology. Discover more stories that matter to your business.