"A node cannot sell what the lease does not let it hold." That sentence is the least glamorous line in quick commerce, and it explains most of the gap between network growth and availability. Swiggy's Instamart ended the September quarter with 1,240 nodes across 133 cities after adding 69 net stores, with net order value up 34.8% year on year to Rs 6,594 crore as of November 11, 2026, while its adjusted core loss widened to Rs 895 crore (Urban Acres). More space, more orders, thinner contribution — the three moved together.
The lease decides assortment depth
A small-format node is a fixed box: rent buys square metres, square metres buy shelf positions, and shelf positions buy SKUs. Every SKU added beyond the shelf capacity forces a trade against another one, which is why networks converge on a fast-moving core and quietly drop the long tail. Availability then reads well for the core and badly for everything a shopper actually searched for.
That trade-off is visible in the numbers before it is visible in customer complaints. Contribution margin in quick commerce deteriorated to negative 1% of gross order value from negative 0.2% in the prior quarter, even as net order value rose. Higher sales did not resolve cost pressure, because the cost of holding and moving that assortment sits in rent, picking and delivery rather than in the marketing line.
Density is not the same as capacity
What another node actually adds
Adding a node inside an existing catchment shortens the delivery leg and splits the order book. If picking is already the bottleneck, the split does not raise throughput: each node inherits the same pick path, the same shelf depth and the same refill rhythm, so the constraint is copied rather than relieved. Density pays when the binding constraint is ride time; it costs when the constraint is inside the box.
Rent per order, not rent per square foot
The useful lease metric is rent divided by orders served, not rent per square foot. A cheap unit in a thin catchment can carry a higher rent per order than an expensive unit in a dense one. Operators expanding into 133 cities are effectively testing that ratio in every new market, and the ones that report store counts without it are reporting activity rather than economics (Urban Acres).
Three numbers per node
| Metric | Definition | Reference band |
|---|---|---|
| Orders per node per day | Delivered orders / active nodes / day | >=400 |
| Availability | Searched SKUs sellable at the moment of order | >=92% |
| Contribution after rent | Order value minus goods, pick, delivery and rent | >0 |
| Shelf utilisation | SKUs carried / shelf positions available | 80%-90% |
A practical test sits underneath all three. Divide a node's fixed daily cost by the contribution of an average order, and the result is the number of orders the box has to serve before it stops losing money on rent; a negative contribution cannot be repaired by adding volume to it (Retail News Asia). That is the same lease question asked from the other end.
Shelf utilisation is the quiet one. Running at 100% leaves no room for a new SKU without removing an existing one, which slows every assortment test; running below 80% means paying rent for empty space. The band in between is where a node can still respond to what shoppers search for, which is the whole point of holding inventory close to demand.
None of this argues against expansion. It argues for reading expansion correctly: a network that adds nodes while contribution stays negative is buying market position, and a network that adds nodes while availability falls is buying the wrong kind of position. The lease is where that difference is decided, long before a rider ever reaches a door (The Entrepreneur Story).
FAQ
Q: Why is a lease an availability question?
A: Because rent fixes how much space a node can hold, and space fixes assortment depth. A lease that is cheap per square foot but too small caps availability permanently.
Q: Do more nodes always mean faster delivery?
A: No. Density shortens the ride but not the pick. When pick capacity is the constraint, another node copies the bottleneck instead of removing it.
Q: What should a node be judged on?
A: Orders per node per day, availability, and contribution after rent, picking and delivery. Store count alone tells you nothing about whether the network works.
