Three numbers from Walmart's latest quarter explain why store selection is being reopened as a topic. US e-commerce sales rose 24%, store-fulfilled deliveries grew about 43%, and deliveries completed within three hours accounted for roughly 37% of store-fulfilled orders. Add Express Pickup — one-hour collection, live at 500 stores and slated for national rollout before Thanksgiving — and the store stops being a place customers walk into and becomes the node that keeps a promise. As of November 12, 2026, the trade area is no longer drawn only around people who show up.
Pickup shortens the promise, delivery widens the area
The mechanics are simple and the consequences are not. A pickup promise is kept by a picker walking a shelf within the hour, so on-shelf availability has to be true at the moment of the order, not at the start of the day. A delivery promise is kept the same way, only with a rider attached and a catchment to protect. Both turn quiet shelf gaps into visible failures, because the order has already been accepted and paid before anyone reaches the aisle.
That is why the store density argument has flipped. When the store is also the warehouse, a closer store means a faster delivery, fresher product and a lower delivery cost per drop. A market that a drive-time model would call fully penetrated can hold one more store, because that store serves delivery customers the existing sites cannot reach inside the promise (Zacks).
Two sales lines, two margins
Most site forecasts still carry one sales line. Delivery orders carry platform fees, picking labour and promotional cost that walk-in sales do not, so a single line understates sales and overstates profit at the same time. The fix is structural rather than analytical: split in-store and online, each with its own margin, before the site goes to committee. A store that looks accretive on one line can be dilutive on two.
What the last mile now costs
| Model | Who carries the last mile | Where it breaks |
|---|---|---|
| One-hour pickup | Customer | Order assembly needs a substitute |
| Three-hour delivery from store | Rider, from the store's shelf | Stock file diverges from the shelf |
| Delivery from a dark store | Rider, from a dedicated node | Fixed cost per node before demand arrives |
The third row explains why retailers with dense store networks keep choosing the first two. A dark store of a few hundred to a thousand square metres, a range of a few thousand SKUs and a radius of two to three kilometres is a perfectly efficient format — until the fixed cost has to be carried by orders that have not arrived yet (Retail News Asia). A store that already exists carries that cost on someone else's rent.
Search data points the same way. Demand for delivery capacity shows up in the labour market before it shows up in a P&L: driver-related queries are the clearest signal that last-mile capacity, not consumer appetite, is the constraint (Rising Trends). For a retailer, that means the pickup lane, the staging area and the cold holding space are site criteria, not fit-out details — six drivers waiting at 5:30 on a Friday is a design problem, not a staffing surprise.
Cannibalisation arrives faster than the model says
A new store going live on an app does not wait for shoppers to discover it. Orders inside its zone can route to it automatically from the first day, which means sales transfer from existing stores can be larger and faster than a drive-time model predicts — and the new store ramps faster too. Both belong in the committee package; neither appears in a forecast built on walk-in capture rates.
What to measure before signing a lease
| Metric | Definition | Warning sign |
|---|---|---|
| Line-level availability by day part | Share of advertised lines found on first pass | Below 95% in the 4pm to 8pm window |
| Cancellations by reason | Cancelled lines with time of pick | Clustered minutes before the slot |
| Orders per store per day | Fulfilled orders divided by active stores | Flat while store count grows |
| Pick minutes per order | Labour time from accept to handoff | Rising after range extension |
The second row is the one that changes decisions. A cancellation minutes before pickup is a customer standing at a counter being told the item is gone; it costs more than a lost sale because it is witnessed. Measuring it by reason and by time of pick turns an anecdote into a site-level number.
None of this makes speed the objective. Express pickup and three-hour delivery are ways of competing for the top-up trip, which is frequent, low-value and intolerant of a missing item. The retailer that wins it is the one whose shelf file is honest at the hour the promise is made — and that is a property of the site, not of the app.
FAQ
Why does one-hour pickup change store selection?
Because the store stops being only a place to shop and becomes the node that keeps the promise. A site that looked fully covered by drive-time maths can become viable again if it serves delivery customers existing stores cannot reach quickly.
What does store-fulfilled delivery growth of 43% imply?
That the fulfilment node is the store, not a separate warehouse. Growth of that size inside existing square footage puts pick labour and stock accuracy on the store's own P&L.
Which metric should a new site be judged on?
Line-level availability against the advertised assortment, by day part, plus cancellations by reason. Catalogue size says nothing about whether the promise can be kept at 5pm on a Friday.
Is pickup cheaper than delivery?
Per order, usually yes, but not per basket. Pickup removes the delivery cost and hands the last mile to the customer, which works only if the store can assemble the order without substitutions.
