Fill rates above 90 percent and 60 million monthly buyers now describe the same channel, and the distance between those two numbers is where quick commerce is actually decided. As of 10 November 2026, the public record has stopped talking about growth in the abstract. Hindustan Unilever told its capital markets day that an AI-led availability engine, nano distribution centres placed closer to demand and faster replenishment lifted fill rates by 1,400 basis points, with front-end availability up 1,000 basis points. Redseer, cited by the Economic Times the same week, put India's quick commerce market at roughly USD 9 billion in the first half of 2026, up from about USD 1 billion in the first half of 2023, while monthly transacting users climbed from 8 million to more than 60 million(Economic Times).
Why fill rate replaced growth as the metric
The promise in this channel is measured in minutes, which changes what a missing item costs. In traditional e-commerce an out-of-stock SKU delays a sale; the shopper waits, or buys the same item a week later. In quick commerce the basket disappears, because the reason for the order was that the shopper needed it within the hour. That is why fill rate has moved out of supply chain reviews and into investor presentations, and why door counts alone no longer describe capacity.
What the 1,400 basis points actually came from
The disclosed mechanics are unglamorous: integrated forecasting, smaller distribution nodes placed closer to demand, and faster replenishment. None of the three is new. What changed is that they were run as one system rather than three separate projects, which is why the gain appeared in fill rate instead of in a single function's scorecard. Moving inventory closer to demand shortens the replenishment loop, but it only helps if the forecast is granular enough to decide what belongs in that node.
System stock is not shelf stock
Most retail organisations still report availability from system stock, which counts cases in the backroom and SKUs misplaced on a neighbouring shelf as available. The shopper can buy neither. Field audits that count the pick face separately usually find a gap of ten to twenty percentage points against the system figure, and that gap is exactly the part a ten-minute promise cannot absorb. A store that reports 95 percent availability from the system and 80 percent from the shelf is not a reporting problem, it is a cancelled order waiting to happen.
Where availability breaks when the range widens
The channel is no longer a grocery story. Blinkit's disclosures describe an assortment above one lakh SKUs backed by roughly 8.5 million square feet of warehousing, and non-grocery categories are growing faster than grocery inside the same dark stores. That widens the failure surface: a grocery backroom picks from a few hundred fast movers, while a distributed warehouse has to hold variant depth for phones, size runs for apparel and batch records for medicines inside a footprint that was never designed for it.
| Channel | Availability metric | Typical failure |
|---|---|---|
| Dark store | Fill rate on live orders | High system stock, empty pick face |
| Modern trade outlet | On-shelf availability by SKU | Backroom stock counted as available |
| Marketplace listing | SKU coverage per store | Listing live, variant missing |
| Store as node | Cancellations caused by stock | Promise made before availability check |
Variant depth before stationery
A store may list a phone model while carrying a single colour, so the listing looks complete and the actual choice is missing. The same happens with apparel size runs and with chilled beauty products, where storage compliance rather than stock is the binding constraint. Networks that treat participation as a listing exercise rather than an availability exercise tend to see cancellation rates climb within weeks, usually after the first promise has already been made to a shopper.
A weekly node audit in five steps
Expansion plans are still judged on door counts and sales per square foot, which is the wrong order. The earlier signal is availability per node: whether a new location can hold its listed range through a full week, not just on opening day. A node that opens at 95 percent and settles at 82 percent has not added capacity, it has added a promise it cannot keep.
- Category teams separate listing coverage from fulfilment coverage for every store and SKU pair in scope.
- Field auditors count the pick face by SKU once a week and log the timestamp with the count.
- Operations track fill rate on live orders separately from system stock, and report the gap as its own number.
- Logistics tags every cancellation with a cause: stock, rider, or system, and publishes the split weekly.
- Commercial reviews use the three numbers together; a node below threshold gets a range cut before it gets new doors.
FAQ
Why does fill rate matter more than market size here?
Market size describes demand. Fill rate decides whether that demand can be served at the moment it appears, and in a channel measured in minutes the moment does not come back.
Can a brand fix store-level stockouts with more safety stock?
Only partially. Extra stock helps when the forecast is roughly right and hurts when the mix is wrong; the shelf carries the cost either way. Fixing the mix usually moves the number more than adding cases.
Is a marketplace listing a reliable availability signal?
No. A listing confirms that a store is onboarded, not that the item can be picked in the next hour. Listing coverage and fulfilment coverage should be reported as two numbers.
What should a new node prove before it opens?
That it can hold its listed range for a full week at the fill rate it promised on day one. Opening-day numbers describe the launch, not the operation.
