As of 6 November 2026, Luckin Coffee's third-quarter results show a business that can lose a double-digit share of same-store volume and still post record profit. Total net revenues rose 17 per cent to RMB17.9 billion, gross merchandise value rose 20 per cent to RMB20.7 billion, and GAAP operating profit climbed 55 per cent to roughly RMB2.8 billion. Self-operated same-store sales fell 11.1 per cent. For teams that plan store networks and monitor on-shelf availability, the interesting part is not the headline profit but the mechanism behind it: delivery expenses dropped 38 per cent to about RMB1.8 billion as orders moved back to pickup, and store-level margin recovered to 23.3 per cent. That recovery now puts the constraint somewhere else, at the store counter rather than at the finance line.
Delivery costs fell because the order mix changed
The saving is a mix effect, not an efficiency gain. A year earlier, food-delivery platforms were at the peak of their subsidy war, and Luckin's quarterly delivery bill reached RMB2.89 billion. As those subsidies faded, customers returned to pickup, and the bill fell to RMB1.79 billion. Management was explicit on the call that the roughly RMB1.1 billion gap is a year-on-year expense difference rather than new profit, but it still lifted store-level economics back to where they sat before the subsidy period.
Ten per cent of revenue is the number to watch
The ratio matters more than the absolute amount. At 10 per cent of net revenues, down from 19 per cent, roughly nine yuan out of every hundred now stays inside the store rather than going to a rider. That is why store-level operating profit for self-operated outlets rose 55.5 per cent to RMB3.02 billion and store-level margin improved 5.8 percentage points to 23.3 per cent, close to the level seen before the delivery war.
Availability is the next bottleneck
With pickup dominant, demand arrives in narrow peaks: the morning commute, the post-lunch dip, the afternoon office round. A store that can absorb a delivery order spread over twenty minutes may not absorb the same volume of customers standing at the counter. Availability stops being a warehouse question and becomes a counter question: whether the pastry case is stocked, whether the second machine is staffed, whether the mobile order queue is cleared before the peak begins.
Store-level capacity replaces warehouse coverage
For most consumer goods, availability is about distribution coverage across outlets. For a chain with 38,909 stores, most of them small, availability is about throughput per outlet. The company added 41 new beverages and more than 20 snack items in the quarter, which widens the menu but also lengthens the preparation time per order. Menu breadth and counter speed pull in opposite directions, and the balance shows up in complaints and abandoned orders rather than in the revenue line.
Store density and the trade area problem
| Metric | Q3 2026 | Year earlier | Change |
|---|---|---|---|
| Total net revenues | RMB17.9bn | RMB15.3bn | Up 17 per cent |
| Same-store sales, self-operated | Down 11.1 per cent | Up 14.3 per cent | Turned negative |
| Delivery expenses | RMB1.8bn | RMB2.9bn | Down 38 per cent |
| Store-level margin | 23.3 per cent | 17.5 per cent | Up 5.8 points |
| Store count | 38,909 | About 29,200 | Up about 33 per cent |
The same-store decline is not a single-quarter event. The metric moved from minus 0.1 per cent in the first quarter to minus 5.3 per cent in the second and minus 11.1 per cent in the third. Management points to a high base, which is fair, but the sequence also says something about density: store count grew about 33 per cent while revenue grew 17 per cent, so average revenue per store is being diluted. When new outlets open inside an existing trade area, the older store loses volume first, and the loss is recorded as a same-store decline rather than as a new-store success.
Why store selection needs a trade area test
A trade area test is not a radius drawn on a map. It is an estimate of how much demand a new outlet takes from the nearest existing outlet versus how much it adds from customers who previously did not buy. In dense Chinese cities, metro-station outlets and street-corner outlets draw from different flows, so the test has to be run per format. Treating a station kiosk and a neighbourhood store as the same unit overstates coverage and understates cannibalisation.
What to track next quarter
Three numbers will settle the argument. The first is the delivery ratio: if it holds near 10 per cent, the margin recovery is structural. The second is the monthly trajectory of same-store sales, which management says narrowed from July to September and improved further in October. The third is store-level availability during peaks, which no filing reports but every operator can measure internally through abandoned orders and queue times.
FAQ
Does the delivery saving equal extra profit?
No. Management described the roughly RMB1.1 billion as a year-on-year expense difference, not new income. Its real effect is to restore store-level cost structure, which is why the margin, not the profit line, is the better measure of what changed.
Is the same-store decline caused by opening too many stores?
Partly. Store count grew about 33 per cent while revenue grew 17 per cent, so average volume per store fell. A high base from subsidies explains the timing; density explains the depth.
How should availability be measured for a coffee chain?
At the counter, not in the warehouse. Useful proxies are abandoned mobile orders, queue length at peak, and the share of menu items actually available during the two daily peaks.
What does a trade area test look like in practice?
Estimate the volume a new outlet takes from the nearest existing one, then compare it with the volume it adds from new customers. If the transfer share is high, the outlet adds cost without adding demand.
