October 20, 2026
Ben Kuethe Oaks
Walk into any grocery store’s back room during the morning receiving shift and you will find some version of the same scene: a store associate pulling near-date product off the shelf, sorting it into categories, and either donating it, calling a hauler, or writing it off. The workflow has changed very little in twenty years. Neither has the hit to the bottom line.
Perishable shrink, unsold product leaving through that back door daily, accounts for 65% of total store shrink while representing only 38% of sales. The average store loses roughly $82,000 annually to produce shrink alone, plus $93,000 in meat and $74,000 in deli: a quarter million dollars a year, before a single item is stolen. Across a 300-store chain, that’s $75 million no security camera or self-checkout audit will touch.
The industry has known this for a decade. What’s missing isn’t awareness. It’s the operating model to act on it. And the environment around that gap has changed.
A squeeze from both ends
Grocery margins are being compressed from two directions, and the pressure isn’t likely to ease.
Consumer behavior has shifted structurally: McKinsey’s 2026 State of Grocery North America found more than half of shoppers have cut impulse purchases, nearly half traded into private label, and three in four changed buying behavior because of price. FMI puts 70% as extremely or very concerned about rising prices. These are durable shifts, not anomalies, limiting operators’ ability to pass costs through.
On the supply side, tariffs on Mexican, Canadian, and Chinese imports have raised procurement costs across perishables. Produce prices are up 16% year over year, cattle inventories sit at historic lows, and meat prices are elevated. Tariff increases typically take 12 to 18 months to reach shelves, so 2025’s full impact hits consumers in mid-to-late 2026. Operators are absorbing margin now while the recovery window narrows.
Perishable shrink is one of the few remaining margin levers operators control.
The instrument has been incomplete, not the theory
Every version of the shrink-reduction argument shares one flaw: correct at the category level, nearly impossible to act on operationally. A 0.5% reduction in shrink as a percentage of sales is a significant P&L event. Identifying which SKUs to cut, in which stores, and when to mark down preemptively rather than manage loss after the fact requires data most operators lack.
The reason is structural. Most retailers manage unsaleables through a patchwork of vendors: a hauler, a donation partner, a composter, a landfill contract as backstop. Each solves a narrow problem; together they consume the time needed to build something more intelligent, leaving nothing for the question that matters: what’s going unsold, and why?
Fragmented diversion programs don’t just fail to generate intelligence. They consume the capacity intelligence requires.
What the operating layer makes possible
A vertically integrated diversion program — managing store-level pickup through processing under one model — does two things. It replaces a fragmented, unpredictable cost structure with a managed one, freeing the bandwidth vendor coordination has consumed. It also generates data consistently, at scale, without adding to the store team’s workload.
A reasonable challenge: shouldn’t stores already be capturing this data through associate workflows? In theory, yes. In practice, manual processes face compliance gaps, training variation, and a grocery store’s competing demands. The associate on shift is also receiving deliveries and helping customers. Data entry gets skipped, leaving fragmented records that can’t be aggregated across locations or tied to purchasing decisions.
Vertically integrated diversion captures the same information differently: as a byproduct of the diversion operation, recorded during processing, where consistency can be managed. Product composition, category, volume, and store of origin are logged at a mandated event, not a discretionary task among a dozen priorities. Data quality is structurally higher and scales across a network in ways associate-driven workflows haven’t.
That dataset makes an intelligence layer possible. Store-level data on what goes unsold, tied to ordering history, markdown cadence, and donation logs, reveals root causes aggregate figures never do: a chronic over-buy in one produce category, a recurring markdown-timing failure in deli that repeats every Tuesday, a category mix that doesn’t match local demand. These are fixable once the data surfaces them.
The sequence matters: a consolidated program’s first job is bringing a chaotic cost line under control. Once that operating layer functions, the data layer becomes possible and feeds back into ordering — reducing over-buy before product hits the floor, not just managing what didn’t sell. Shrink and diversion volume decline together.
Why now
Grocery is in an operational reset. AI-driven forecasting and inventory tools are attracting real investment and sophistication in fresh-department ordering. But those models are only as good as their data. What’s been missing is a consistent, store-level record of what went unsold. Vertically integrated diversion is the infrastructure that generates it. The two capabilities are more complementary than the industry has recognized. Diversion consolidation produces the signal that AI-driven ordering needs to be worth the investment.
The opportunity has been visible for a decade. Unexecuted opportunities at this scale usually have an infrastructure problem, not a logic problem, and that infrastructure now exists. The retailers that recognize it first will be best positioned to turn shrink from a cost center into a competitive advantage.
Ben Kuethe Oaks is SVP of Commercial at Divert, Inc., a vertically integrated circularity platform operating inside major U.S. grocery retailers and CPG companies.
