35 percent fewer stockouts and 28 percent less shelf time within 60 days — that is what DeCicco & Sons measured in its shelf-robot pilot, and as of November 12, 2026 the twelve-store New York and Connecticut grocer has taken the system into every store. In the same week, Dollar General put roughly 20,000 stores on the Instacart marketplace and Lowes Foods began marking down surplus fresh food through an app at five South Carolina stores. Three different moves rest on one assumption: that somebody knows, hour by hour, what is actually on the shelf.
Three cheques written against the same shelf
Start with the chain that measured. DeCicco & Sons reported a 35 percent reduction in out-of-stocks, up to a 28 percent reduction in the average time an item stayed missing, and enterprise-wide shelf data flowing through APIs into e-commerce availability, supplier conversations and store routines. The company's chief executive, John DeCicco Jr., frames it less as a robot story than as a data story: every system the chain runs now gets an accurate picture of what is happening in the store.
Now the scale case. Dollar General and its pOpshelf banner went live on the Instacart Marketplace across 48 states, with orders promised in as quickly as 30 minutes and an introductory offer of 10 dollars off a first qualifying 50 dollar order through December 31. The retailer operated 21,148 stores in the United States and Mexico as of July 31, 2026, and Instacart carries more than 2,200 national and local banners. A listing that says same-day delivery is a promise about stock, not about software.
The third move is about time. Five Lowes Foods stores in the Columbia, South Carolina market list surplus meat, dairy, produce and bakery items on the Flashfood app at up to 60 percent off, collected at the service desk the chain calls the Front Porch. Flashfood says shoppers have saved more than 55 million dollars in 2026 and that its marketplace works with more than 2,300 stores. Markdown is what happens when a count arrives too late to fix the order.
What a machine count is actually for
A shelf robot does not sell anything. It produces a repeated, timestamped measurement of the same physical space, which is the one input that store operations, e-commerce and markdown decisions all lack today. Most chains still take that measurement by walking the aisle with a clipboard, which means the number is late, partial and impossible to compare week to week.
Time out of stock beats the count itself
A store can report 200 missing items and still lose less money than a store reporting 150, because what a shopper experiences is duration, not incidence. A gap that opens at 9 a.m. and closes by noon costs a fraction of a gap that opens on Friday and closes on Monday. That is why the 28 percent reduction in average time out of stock is the more interesting number in the DeCicco disclosure, and why any store-level target built only on counts will drift.
Data has to land inside the systems people already use
The failure mode for shelf technology is a dashboard nobody opens. DeCicco's approach — pushing shelf data into the tools associates and department managers already work in, rather than asking them to adopt a new console — is the difference between a pilot that ends and a routine that sticks. Simbe, which reports more than 3,000 stores under contract across 75-plus banners, is selling exactly that integration argument, and the same logic applies to a marketplace feed: if the availability file sent to a delivery partner is refreshed daily while the shelf is counted hourly, the listing will keep overselling.
Three cases, three time horizons
The comparison below is the practical part. Each row answers a different question, and each pays back on a different clock: a robot count is a capability the chain owns, a marketplace listing is reach built on somebody else's platform, and a markdown app is a disposal channel for stock that is already in the building. Putting them side by side makes the trade-off visible, because a retailer can buy all three and still not know which one moved the number.
| Case | What it changes | Measured effect | Payback horizon |
|---|---|---|---|
| DeCicco & Sons with Simbe | How often the shelf is counted and who sees the result | Out-of-stocks down 35 percent in 60 days; shelf time down up to 28 percent | One to two quarters |
| Dollar General on Instacart | Where the assortment can be bought from | About 20,000 stores in 48 states, delivery in as little as 30 minutes | Immediate reach, permanent stock risk |
| Lowes Foods with Flashfood | How surplus fresh food leaves the store | Up to 60 percent off, five stores, pickup at the Front Porch | Weekly, per store |
Read the rows in order of evidence. The first row has a measured effect on a named set of stores and a stated window; the second has a footprint but no availability data; the third has a discount ceiling but no volume. A supplier or a category manager comparing them should weight them accordingly, and should ask the same question of all three: who owns the number that says an item is on the shelf right now.
There is a second-order effect worth planning for. Once a robot walks the aisle several times a day, the store stops being a place that gets audited and becomes a place that reports. Assortment decisions that used to be taken once a quarter on a rep's visit can be taken weekly on observed gaps, which changes what a field team is for, and changes how quickly a distribution gap shows up as a sales gap rather than as a complaint.
FAQ
Does a shelf robot replace the store team?
No. It replaces the clipboard walk. The chain that reported results describes staff spending more time on the floor resolving gaps, not less.
Is a marketplace listing the same as availability?
No. A listing is a promise about assortment; availability is a fact about a shelf at a moment. The gap between them shows up as cancellations.
Why discount surplus food at all?
Because a markdown recovers part of the cost of an order that was already too large. It treats the symptom; the count treats the cause.
What should a supplier ask for first?
Item-level availability by store, with timestamps. Without it, every conversation about distribution coverage is an anecdote.
Sources
- The Shelby Report: DeCicco & Sons expands Simbe shelf-scanning robots to all 12 stores, citing a 35 percent cut in out-of-stocks
- The Shelby Report: Dollar General and pOpshelf go live on Instacart across about 20,000 stores in 48 states
- The Shelby Report: Lowes Foods pilots Flashfood markdowns at five Columbia, South Carolina stores
