As of 9 November 2026, the grocery reset inside Amazon has moved from closing stores to explaining the ones that remain. The company shut 58 Amazon Fresh locations in the United States and 19 in Britain across late 2025 and early 2026, kept Fresh as an online-only banner, and now points to Whole Foods and to small Daily Shop formats as its physical answer. The interesting question for anyone running stores is not who won the argument, but which cost each format is trying to remove.
What Amazon said on the earnings call
The numbers behind the strategy are plain enough. Amazon delivered 1.5 billion items same day or overnight in the third quarter, and says fresh and everyday essentials grew roughly three times faster than the rest of the store while representing about half of units sold. The company also says shoppers who buy perishables with same-day delivery buy nearly seven times more everyday essentials than those who do not, which is the argument for keeping the delivery promise rather than the box.
That argument has a cost side that rarely appears in the same sentence. A delivery-led format without a clear price position does not earn its rent, and a 40,000 square foot store built around a wide assortment is expensive to run when the order arrives through an app. The Fresh exit was a statement about cost to serve as much as about customer preference.
The vacancy left behind is being repriced
Empty anchor boxes do not stay empty for long, and the tenants that take them tell you what the market now believes. In Silver Spring, Maryland, Lidl confirmed on 6 November that it will take space that Amazon Fresh vacated in February after less than a year, a swap from a delivery-first format to a narrow-assortment discounter that displays products in their shipping boxes to save labour.
Two formats, two different cost drivers
Lidl earns its margin on turns and labour, with nearly 80 percent of what it sells coming from private-label suppliers, while Amazon earns its grocery growth on delivery density and basket adjacency. Neither is a better store; they are two answers to different cost problems, and a supplier has to re-earn shelf space under a different rule in each one.
| Format | What it optimises | First signal to watch |
|---|---|---|
| Discount discounter, narrow assortment | Labour cost and turns per square foot | Share of sales from own brand |
| Small-format urban grocery | Basket adjacency and trip frequency | Items per basket after the first visit |
| Delivery-led online grocery | Cost to serve per order | Orders per route and delivery density |
| Warehouse club | Volume per square metre | Membership renewal and basket size |
Read together, the cases rule out the simplest story, which is that physical grocery is losing. What is losing is a format that carries the cost of a supermarket and the promise of an app without a price position strong enough to pay for either, and the vacancy is being re-let to whoever can carry that cost more cheaply.
Why coverage still decides the outcome
Every format in this reset depends on the same underlying variable. A discounter needs a narrow range to be in stock every day, a small-format store needs the top few hundred items to be present at all times, and an online grocery needs the shelf to match the app. On-shelf availability is the shared constraint, and it is the one that breaks first when a store changes hands.
Stockouts move before sales do
When a banner is converted, planograms are reset, suppliers change and delivery schedules are rebuilt, so gaps appear on specific pack sizes for a few weeks. Category growth then looks flat while the real problem is a missing facing, and consumer demand is misread as a preference shift rather than as a supply failure. A panel that measures availability weekly separates the two before a range review is ordered.
How to test a format change in one quarter
Format arguments are settled by evidence from converted stores, not by national averages. The test below compares a converted store with an unchanged one in the same trade area, so that a promotion week or a payday does not get mistaken for a format effect. It takes one quarter and one analyst, and it produces a number a landlord or a supplier can argue with.
FAQ
Does the Fresh exit mean physical grocery is shrinking?
No. The square footage is being re-let, mostly to discount and small-format operators, so the store count barely moves while the operating model behind each box changes completely.
Why does cost to serve matter more than sales growth?
Because delivery growth without density raises the cost of every order, and a format that cannot cover that cost per trip will eventually be closed regardless of how fast its top line grows.
What is the first thing to measure after a conversion?
Availability of the top items by pack size, measured weekly, because it is the variable that a planogram reset damages first and the one that recovers last.
How long before the comparison is meaningful?
Two full delivery and promotion cycles, which in most markets means six to eight weeks, shorter than that and the result is mostly noise from the opening week.
