By Amir Khoshniyati, VP, Wiliot

‍

Grocery supply chains have long run on the assumption that tomorrow will look a lot like yesterday. Retailers could forecast demand using historical purchasing patterns, promotional cycles, and seasonal rhythms, and high-velocity center-store categories like snacks, bakery, frozen meals, and soft drinks moved reliably enough to reward that assumption. Replenishment systems were reliable and predictability itself became a competitive advantage.

‍

Then the weight-loss drug boom rewrote the American shopping list.

‍

GLP-1 medications like Ozempic and Wegovy have gone mainstream, and the categories grocers built their businesses around are starting to feel it. A recent Cornell University study published in the Journal of Marketing Research found that GLP-1 households reduced overall grocery spending, with the steepest declines being in historically stable categories such as savory snacks and baked goods. One in eight American adults is now on one of these drugs. A market that size doesn’t bend to forecasting models, it breaks them.

‍

But the consumption decline only tells part of the story. The deeper-rooted operational issue is what changing demand does to physical visibility across the store environment.

‍

The Forecasting Problem Is Nonlinear

‍

Grocery forecasting systems are, in essence, pattern-recognition engines. Fed on years of historical movement data and promotional elasticity, they get very good at optimizing around consistency. GLP-1 adoption is the kind of change those engines were never calibrated for. It arrives gradually, unevenly, and along demographic lines that don't announce themselves in historic transaction data.

‍

The distortion isn't uniform. One store may see softening across traditional center-store categories while a nearby location picks up stronger movement in fresher, higher-protein, or smaller-portion products. The same category can be telling two different stories, two miles apart.

‍

That creates operational risk in both directions. Over-forecasting legacy high-velocity categories drives up markdown exposure and spoilage. Under-forecasting the substitution categories means missed sales and empty shelves. And because these changes tend to emerge slowly and unevenly, they often don't surface clearly until the inefficiencies have already been compounding for weeks.

‍

The more granular and real-time the shelf and inventory visibility, the faster retailers can adapt to those new baselines before the margin damage is done.

‍

Slower Movement Does Not Mean Simpler Operations

‍

There is a counterintuitive dynamic at work here that can be easy to miss. When a category cools, the instinct is to assume operations simplify. In practice, it's typically the opposite that rings true.

‍

Slower movement breeds lower replenishment urgency, and phantom inventory, shelf gaps, and execution failures piggyback off of that blind spot. A product might be sitting on a backroom pallet or misplaced in the wrong aisle for days before anyone notices, because the signal that typically triggers a check has weakened. The system is designed to respond to velocity, and when velocity falls, so does its resolution.

‍

A store that cannot read itself becomes a puzzle with missing pieces, and those misreads compound in three specific ways: genuine demand decline looks nearly identical in the data to poor shelf execution, slower-moving goods accumulate expiration risk before anyone flags them, and localized preference shifts stay invisible until they show up in shrink. 

‍

With assortment fragmentation increasing, retailers need a clearer operational picture of what is actually happening inside the store; one that allows merchandising, store operations, and supply chain teams to distinguish shifting demand from execution breakdowns before inefficiencies take hold.

‍

Fresh Categories Face a More Complex Balancing Act

‍

Early signals suggest GLP-1 users are gravitating toward more nutrient-dense products, which means grocers may be about to face a more difficult balancing act in perishables precisely as those categories heat up. Fresh departments already operate with tighter spoilage tolerances and more variable demand curves than packaged goods. It’s an unforgiving environment under stable conditions.

‍

Under unpredictable ones, even modest changes in consumer mix can reverberate through replenishment cadence, safety stock assumptions, labor planning, and shrink rates all at once. Static planograms and weekly planning cycles were not designed to absorb that kind of pressure. What fresh categories demand instead is near real-time inventory intelligence and adaptive replenishment; the ability to sense and respond to what is happening inside the store as it happens.

‍

Moving From Passive Inventory to Intelligent Inventory

‍

The underlying problem GLP-1s are exposing is that grocery systems were never designed to continuously understand the physical world. They were designed to infer it – and for a long time, inference was good enough. But not anymore.

‍

So how does grocery keep pace today? Physical AI.

‍

With Wiliot’s Physical AI platform and battery-free IoT Pixels, products gain a persistent digital presence that follows them throughout the supply chain and into the store itself. Goods that once went silent between scans now continuously communicate their movement, location, dwell time, and environmental conditions. The store stops being an environment retailers piece together from fragmented transactions and becomes one that can reflect what’s actually happening inside it, in real time.

‍

For grocers navigating the uncharted waters of ever-evolving GLP-1-driven purchasing behavior, that opens up new operational possibilities. Friction surfaces earlier, before it has time to harden into margin pressure. Retailers gain a clearer picture of how inventory is actually behaving as the consumer habits beneath it continue to change.

‍

Appetite is no longer a stable variable. For an industry that built its entire operating infrastructure around the assumption that it was, real-time visibility has become the starting point for everything that comes next.

‍

The Operational Winners Will Be the Fastest Learners

‍

GLP-1 drugs are holding a mirror up to grocery supply chains, and what they are reflecting back is uncomfortable. The entire operating model was built on the assumption that consumer behavior would remain legible through historical data alone.

‍

The real issue goes deeper than declining snack sales. It’s a long-term erosion of consumption predictability, and for an industry that spent decades engineering precision around consistency, that is a far more unsettling problem. The retailers that come out ahead will be the ones detecting and responding to micro-shifts across the physical store environment before those shifts hollow out a category's margin structure. Physical AI is what makes that speed and granularity of detection possible.

‍

A prescription changed what America puts in its cart. The supply chains sharp enough to see that change as it unfolds – not after the fact, not at the end of the quarter, but in real time – are the ones that will define what modern grocery looks like on the other side of it.

‍