The 30-minute promise now depends on a shop assistant in a convenience store remembering to top up a shelf. As of 9 November 2026 Dollar General and pOpshelf are live on the Instacart marketplace across roughly 20,000 stores in 48 states, with delivery in as fast as 30 minutes and the banner added to a marketplace that already carries more than 2,200 national and local retailers (Instacart). Shipt put 500-plus Sprouts Farmers Market stores across 25 states onto its platform ahead of the heaviest grocery weeks, and Gophr connected three easy-brand platforms so independent UK shops can offer 30 to 60 minute delivery (Foodservice News, Asian Trader). Coverage expanded; the shelf did not get faster.
Coverage grew by twenty thousand doors
The arithmetic of scale is straightforward and slightly misleading. Twenty thousand stores is a footprint claim, and footprint is what a marketplace sells to a brand that wants national reach. What the shopper experiences is a fill rate, and fill rate in a convenience or grocery format depends on whether the item was replenished that morning by someone whose job description is running a till, not picking orders. A marketplace that adds 20,000 doors adds 20,000 replenishment routines it does not control.
The promotional layer makes the gap visible. Instacart is offering 10 dollars off a first Dollar General order of 50 dollars or more through 31 December 2026, and Shipt ran new-member access at 0.99 dollars a month through 7 November with half off a first Sprouts order (Instacart, Foodservice News). A subsidised first basket lands on a shelf that has never served a picker before, which is exactly when substitution rates spike and the first review is written.
What a banner count hides
Banner count hides three things that decide whether a fast order can be filled: the depth of the shelf behind the item, the hour at which that shelf is topped up, and whether anyone on shift knows where the item lives. A store carrying 4,000 lines and selling 3,000 of them a week has no spare depth on a Saturday evening. The platform sees inventory as a number; the shop sees it as a gap on a shelf it will fill tomorrow.
| Layer | What it measures | Who owns it | Typical failure |
|---|---|---|---|
| Footprint | Stores connected to the marketplace | Platform and retailer | Doors open, shelves empty |
| Availability | Sellable units at the hour of order | Store manager | Replenishment done at close, not before the peak |
| Pick readiness | Minutes to collect a standard basket | Store staff | Item located in the back room, not on the shelf |
| Promise | Delivery window shown to the shopper | Platform | Window shorter than the pick plus the drive |
Stockout is traced to a packer rota
In a dark store the picker is a dedicated role. In a convenience or grocery store connected to a marketplace, the picker is whoever is free, which makes the rota the real availability schedule. When the shift changes at four in the afternoon and the person who knows where the seasonal stock lives goes home, the pick time doubles and the substitution rate follows. That is not a demand problem, and no forecasting model fixes it.
The same logic explains why small-format retail is harder to serve than it looks. A 30-minute window leaves roughly eight minutes for the pick once travel is accounted for, which is enough for a basket collected from a known shelf position and not enough for a basket that requires opening a pallet in the back. Retailers that treat marketplace orders as a separate workflow, with a fixed pick route and a named owner per shift, hold their fill rate; those that fold the order into the normal shop routine lose it during the busiest hour, which is also the hour the order arrives.
Where the promise bends
The promise bends at the categories with the least depth. Chilled and frozen lines have the shortest shelf life and the tightest replenishment, so they are the first to show a gap; seasonal and general merchandise have the most depth and the lowest pick familiarity, so they show a different failure, which is a pick that takes three times as long. Reading a substitution log by category separates the two, and the separation matters because one is fixed with a delivery schedule and the other with a shelf map.
Four checks before a brand counts the doors
First, ask for fill rate by banner rather than by marketplace, because the average across 2,200 banners describes nobody. Second, ask for the replenishment hour, since a shelf topped up after the evening peak is unavailable exactly when the fast orders arrive. Third, time a standard basket in five stores, which shows whether the constraint is layout or staffing. Fourth, read the substitution log by category, because the reasons cluster and each cluster has a different owner.
FAQ
Does a large store footprint make a 30-minute promise reliable?
No. Footprint decides how close inventory sits to the shopper; reliability depends on whether the shelf was filled before the order arrived. A store that replenishes at closing time is unavailable during the evening peak regardless of how many doors the marketplace counts.
Why do convenience formats struggle more than dedicated picking sites?
Because the pick is done by staff whose primary job is serving walk-in customers. The rota becomes the availability schedule, and the busiest store hour is also the hour when a fast order is most likely to arrive.
Which number should a brand ask for first?
Fill rate by banner, followed by the replenishment hour for the fastest-moving lines. The marketplace average across thousands of banners describes no store a shopper will actually use.
