The Maze: Online grocery does not have one last mile. It has three different businesses hiding behind one checkout. Brick Meets Click's June 2026 data puts Delivery and Pickup near $95 per order, while Ship-to-Home averages $58. The growth runs the other way: Ship-to-Home rose 28.6% year over year, versus 2.2% for Delivery and a 1.3% decline for Pickup. The important split is not speed alone. It is the shopping mission underneath: roughly 53% of Ship-to-Home orders land below $60, while Delivery and Pickup build much more weight in the $100-plus bands.
The method label is now an economic variable. Ship-to-Home peaks in the $30-$59.99 band at about 35% of its orders. Delivery and Pickup peak at $150-plus, at roughly 38% and 39%. That changes pick density, handling costs, carrier economics, minimum-order logic, and the revenue available to absorb fulfillment. A blended `online grocery` average can therefore hide the health of each model.
Sub-same-day is redrawing the old taxonomy. Delivery comes from a store or warehouse; Ship-to-Home uses common or contract carriers. The boundary blurs when Amazon offers 30-minute Amazon Now delivery and same-day fresh grocery through different networks. Walmart adds drones: its network passed one million deliveries, averaging 23 minutes. The customer sees speed. The operator must still see the network.
Growth can improve reach while compressing the basket. US e-grocery sustained more than 20% year-over-year growth for six quarters through Q1 2026. Delivery and Ship-to-Home grew about three times faster than Pickup; nearly 80% of Delivery and more than 30% of Ship-to-Home orders arrived the same day. But faster fulfillment does not erase method economics. A $58 Ship-to-Home AOV creates less merchandise revenue per order to fund convenience.
Measurement has to follow the mission, not the button. David Bishop argues for comparing spending, product composition, order-cycle time, and customer demographics by method—much as a convenience store should not be managed like a supermarket. Retailers should add cost-to-serve, contribution margin, substitution, fees, repeat frequency, and incremental demand. The survey covers about 1,500 US adults, weighted for age and income. It is market evidence, not a retailer P&L; that is a reason to segment the scorecard, not blend it.
Why it matters: Ultrafast grocery is making different networks look interchangeable at the storefront. They are not interchangeable in the ledger. Retailers that classify by customer-facing promise alone will misprice service levels, misread cohort quality, and fund growth with the wrong basket assumptions. The practical move is simple: preserve the fulfillment method in every performance view, then judge growth against method-specific order value and cost-to-serve. One doorstep can hide three margin structures.
