The numbers are hard to ignore.

According to Coresight Research’s State of In-Store Retailing 2026, surveying 200 VP-level and above retail decision-makers across the U.S., in-store inefficiencies now cost retailers 6.4% of gross sales annually. That’s $196.4 billion across key retail sectors. And it’s getting worse: the same figure was 5.5% in 2025, and 4.5% in 2024. Three consecutive years of rising losses, despite three consecutive years of rising investment.

97% of retailers have deployed or plan to deploy store intelligence technology within the next year. 60% are already actively scaling those deployments, up 18% year over year.

So why are losses accelerating?

Why does more technology produce more inefficiency?

Coresight’s answer is precise: technology sequencing. Retailers are deploying pricing software, supplier platforms, and replenishment systems before they’ve established the one input all of those systems depend on; verified, physical shelf data.

Only 33% of retailers have invested in shelf digitization. The other 67% are running sophisticated systems on top of a gap.

Pricing software without accurate shelf data misprices. Replenishment systems without real inventory visibility over-order or under-stock. Planogram compliance tools without shelf-level confirmation can’t flag what they can’t see.

What shelf-level intelligence actually looks like

This is the problem WeightAI™ was built to close and it does so from inside the machine, not above it.

SHEKEL’s Smart Cooler embeds weight-based AI at the shelf level. Every product removal is verified in real time. Not estimated from transaction logs. Not inferred from a camera feed. Confirmed by weight, at 99.8% accuracy, with no cameras, no RFID, and no biometric data. That means it’s GDPR-compliant by design, a non-negotiable in European deployments where camera-based recognition creates compliance overhead that most operators can’t absorb.

The result is accurate product recognition; live, verified data layer that feeds every system downstream: replenishment, planogram verification, route decisions, sell-through by SKU. The same shelf intelligence the Coresight study identifies as missing from 67% of operator deployments today.

For standard shelving environments, outside the cooler, Sentinel applies the same weight-based logic without replacing existing infrastructure. Any shelf becomes a real-time inventory tracking system. Operators who’ve invested in fixtures don’t need to rip them out. They need to make them intelligent.

Operators buy revenue. Not technology.

The Coresight data makes the business case plainly. Retailers with shelf digitization in place build what the study calls “a compounding competitive advantage that is difficult to replicate.” Those without it keep patching around a foundation that isn’t there.

For autonomous retail operators, the stakes are more direct. 2x+ revenue versus traditional vending in the same footprint. ROI in under 18 months. Transaction completed in under 30 seconds. 100% payment capture because every pick is verified at the point of removal, not reconciled after the fact.

The operators scaling profitably already know this. The Coresight numbers tell you what it costs to find out late.