As of November 9, 2026, the T&T Supermarket that opened on 5 November in San Francisco occupies a 50,000-square-foot unit on the third floor of the City Center mall at 2675 Geary Blvd., a space that sat empty for nearly a decade after Whole Foods dropped its plan. Queues formed before the 9 a.m. opening. The hard part is not filling the shelves on day one but holding them for a month.
A store this size breaks availability in week three
Opening week is a rehearsal with a full cast. The first deliveries are packed against a forecast rather than against local sales, so the shelves look complete for five to seven days. From the second week the replenishment orders follow scanned movement instead, and that is when the distance between a ten-thousand-item assortment and a 50,000-square-foot floor starts to show: fast sellers are reordered late while slow movers occupy the back room.
| Category | First-week replenishment | Where week three fails | How to measure |
|---|---|---|---|
| Leafy vegetables | Daily, pushed by forecast | Afternoon gap on peak days | End-of-day on-shelf rate |
| Mushrooms | Every other day | Tail items quietly drop out | SKU coverage against plan |
| Live seafood | Daily delivery | Waste and stockout at once | Weighed availability |
| Hot food bar | Set by store | Empty trays after the peak | Hourly availability |
The 200 vegetable varieties and 17 mushroom types reported at opening are a width claim, not an availability claim. Width is measured once, on day one. Availability is measured every day after that, and the two move in opposite directions when a store is short of back-room space and labour: buyers keep adding items while staff keep facing the same forty feet of chilled shelving. A stockout that starts in the loading dock ends on a shelf two days later, which is why the count has to be taken at the same hour every day.
- WidthCount items once, at opening, against the plan.
- AvailabilityCount the same items at the same hour, every day.
- DifferenceThe gap between the two is the shelf gap that matters.
The delivery launch moves picking onto the same aisle
Online shopping, express delivery and in-store pickup are scheduled for 4 December, a month after the doors opened. From that day the store carries two demand streams on one floor: walk-in shoppers between 8 a.m. and 10 p.m., and pickers working orders against a promise window. Picking competes with replenishment for the same narrow service corridors, and the loss shows up first in the chilled and frozen aisles, where a picker's cart blocks a pallet and a pallet blocks a picker.
Local coverage of the opening notes that the store sits on the third floor of a mall and that parking is limited, which pushes a share of demand toward pickup and short-radius delivery rather than a car boot. That makes the picking schedule a store-level constraint rather than a head-office setting: the same promise window costs more in a mall unit with two service lifts than in a standalone box with a loading yard.
Picking load is a store number, not a chain number
A twelve-item grocery order touches four or five aisles and takes far longer per item than a shopper filling the same basket, because the picker walks the route in the order the app dictates and then queues at a packing bench. Measure pick minutes per order in the first two weeks and use that number, not the chain average, when a thirty-minute promise is set for this store.
Store records disagree before the shelves do
Two English-language trade reports describe the same opening with different numbers. One puts the opening on 5 November and the floor at 50,000 square feet; another says the store opened on 6 November in a 52,000-square-foot unit. Neither is unreasonable. The earlier date matches the public 9 a.m. opening, while the later one may follow a dateline, and the area gap can come from gross versus selling space. A store file that copies whichever number arrived first will disagree with its own delivery records for years.
| Field | Report A | Report B | Value to store |
|---|---|---|---|
| Opening date | 5 November 2026 | 6 November 2026 | First day of public trading |
| Selling area | 50,000 sq ft | 52,000 sq ft | State gross or selling |
| Item count | About 10,000 | Not disclosed | Take from store master data |
| Online start | 4 December 2026 | Not disclosed | Backfill on the live date |
Store master data is the base layer for distribution and delivery dispatch. If one system holds 50,000 square feet and another holds 52,000, capacity planning drifts by four percent before a single pallet moves. If the opening date differs by a day, the first-week sales baseline is off by a day and every later comparison inherits the error.
Definitions beat values
A field without a definition is an opinion. Write down whether area means gross or selling, whether the item count includes seasonal lines, and whether the opening date means the first trading day or the ribbon cutting. Then keep one owner for the field. It is dull work that prevents a category manager and a logistics planner from arguing about the same store for a quarter.
Ninety days, three numbers
Track three numbers per store for the first ninety days: end-of-day on-shelf availability by category, fresh waste as a share of fresh sales, and the share of orders that arrive through delivery or pickup once it launches. Footfall, basket size and press coverage are context, not evidence. These three tell you whether the store can hold its assortment, whether it is over-ordering fresh to protect availability, and whether the delivery layer earns its picking cost.
FAQ
Why would on-shelf availability fall in week three rather than week one?
Because the first week runs on forecast-driven deliveries packed at head office. Once replenishment switches to scanned sales, ordering lag, back-room congestion and labour limits show up together, and the shelf gap becomes visible in the third week.
Should a new store be recorded by gross or selling area?
Either works if the definition is written down and applied to every store. Mixing the two makes capacity and sales-density comparisons meaningless, which is why the field needs a definition before it needs a value.
What changes when delivery starts a month after opening?
Picking begins to compete with replenishment for the same corridors and lifts. The store gains a second demand stream but not a second floor, so pick waves have to be scheduled around the walk-in peak rather than added on top of it.
How should a store file treat figures that differ between reports?
Record the value from the most authoritative source, note the alternative and the definition behind each, and backfill once the operator publishes its own numbers. Two unexplained figures in one file will keep resurfacing in planning meetings.
