204 percent, 104 stores, one quarter. Assaí's last-mile line looks like a breakout until the same filing reports same-store sales up 0.1 percent and monthly traffic holding above 40 million visits. A cash-and-carry chain built for bulk buyers is being re-read as a delivery network one store at a time, and the question for a retail operator is no longer whether to fulfil from stores but which stores pass the test. Every figure below comes from disclosures available through November 12, 2026.
The 204 Percent Number Counts a Small Base
The 204 percent is a growth rate on a base that was close to nothing a year earlier, which is why it sits so comfortably beside a 0.1 percent same-store number. A chain that adds a delivery channel to 104 existing stores can triple that channel without moving traffic at all. Read as a share of revenue, the channel is still small. Read as a share of stores touched by an operating change, it is already a third of the estate, and that is the more useful denominator for anyone planning the next four quarters.
What changes inside a wholesale format is the unit of picking. A cash-and-carry buyer takes a case; a delivery customer takes a bottle. The same shelf now serves two demand patterns with different pack sizes and different tolerance for substitution, and the operator has to decide which one wins when stock runs thin on a Saturday evening. Assaí's customer traffic, which rose 2.1 percent to more than 40 million visits a month, measures the door. It says nothing about the aisle where the picking happens.
Boxes and bottles on one shelf
A store that sells by the case can look fully stocked while being unpickable for single units, because the last case is committed to a wholesale order. That is the failure mode nobody sees in a monthly report, and it is why stockout records kept in cases read so differently from the same records kept in units: availability measured in units stays healthy, availability measured in cases collapses, and the delivery order that arrives at eight in the evening is cancelled for an item that is technically in the building. Separating the two demand types at store level is the only way to catch it.
One Hundred and Four Stores Set the Real Ceiling
The disclosed network is 313 stores across 24 Brazilian states and the Federal District, with roughly 90,000 employees and gross revenue of 21.3 billion reais in the quarter, up 2.7 percent. Only a third of that estate currently ships. The constraint is not willingness but geometry: a wholesale floor is laid out for pallets and trolleys, and a picking operation needs a staging area, a separate exit and a rider waiting zone that do not compete with the checkout queue.
Assortment adds a second constraint that is easy to miss. The chain launched more than 30 private-label items in the quarter, taking the range to 140 SKUs across four brand tiers, and has said it targets 250 by the end of the year. It also extended its pharmacy concept to 11 units in São Paulo with a target of 25, and pushed dietary supplements into 300 stores. Every one of those additions widens the pick face, and a wider pick face is slower to walk. A private-label programme that improves gross margin at the shelf can still raise the cost per delivered order.
Private label as a fulfilment decision
Own brands are usually discussed as a pricing lever, and at 17.2 percent gross margin with operating expenses growing 4.2 percent against 4.4 percent inflation, they clearly are one. But in a store that ships, the same range carries a second job: it is the part of the assortment the retailer can standardise, substitute and re-pack without a supplier conversation. A delivery catalogue built mostly from own-brand staples is easier to keep in stock than one built from national brands with rigid case packs.
The Screen That Replaces Footfall
Store selection for delivery does not start with population density. It starts with three physical questions: how much unpicking space sits behind the last gondola, how wide the fast-moving assortment is inside a two-hour walking radius of the picker, and how often the store is replenished. A store with 40 million visits behind it and no staging area will lose to a quieter store with a loading bay and a narrow, deep assortment.
| Input | What it predicts | Where it fails |
|---|---|---|
| Back-room and staging area | Orders per hour the store can assemble without blocking shoppers | Stores with mezzanine storage that cannot take trolleys |
| Assortment width in fast categories | Substitution rate when the exact item is out | Categories where the shopper refuses substitution, such as infant formula |
| Replenishment frequency | Evening availability for orders placed after 4 p.m. | Trucks that arrive after the delivery window closes |
| Case versus unit mix | Whether fill rate is measured on the right unit | Stores serving restaurants and households from one shelf |
None of these inputs appear in a quarterly release, which is why the 204 percent number cannot be benchmarked against a peer by reading filings alone. A related test on radius and saturation appears in this analysis of fulfilment polygons. It can, however, be tested store by store, and the test is cheap: pick fifty orders, time the walk, count the substitutions, and repeat it on a Friday evening. A store that fails that test will fail again next quarter, whatever the growth rate says.
Four Steps Before a Store Starts Shipping
The practical sequence is shorter than a capital committee usually assumes, and it is worth running before a lease is signed rather than after a rider is hired. Each step produces one number that can be compared across stores, which is what makes the screen portable from one market to the next. A team that skips the fourth step is measuring a launch rather than a store.
- Measure pickable assortment, not total assortment: count SKUs that can be taken as single units without breaking a case.
- Time a mock order in the worst hour, and record the walk from the pick face to the staging point with a trolley.
- Split availability reporting into case-level and unit-level fill rate for the top 500 sellers.
- Re-check the store after four weeks, because the first week of a new channel is carried by novelty and the fourth is carried by habit.
FAQ
Does last-mile growth of 204 percent mean the channel is now material?
Not by revenue share. It means the operating model has been tested in a third of the estate, which is the stage where processes either standardise or fragment.
Why does case-level fill rate matter for a consumer delivery order?
Because the last case on the shelf is usually committed to a wholesale buyer, so a unit-level dashboard shows stock that cannot be sold to the delivery customer.
Can a cash-and-carry format serve delivery without new space?
Sometimes, if the store has a loading bay and a narrow fast-moving range. Without staging space, orders compete with shoppers for the same aisle.
What should a brand watch in this model?
Whether its SKUs survive substitution, and whether the retailer's own-brand expansion is quietly replacing them in the delivery catalogue.
