What does a thirty-minute delivery window actually buy a bar that runs out of limes at nine on a Saturday night? As of November 13, 2026, the answer sits inside an embedded Instacart checkout on Provi's ordering platform, where operators can now buy garnishes, mixers, juice, cleaning supplies and paper from grocery retailers and take delivery the same day. The partnership was announced on November 12. Its real subject is not speed. It is a second supply route running beside the scheduled distributor drop, drawing on a different shelf and a different item file.
Two supply routes behind one bar
Provi’s scale is the number that matters for planning. The platform counts more than 1,400 distributor partners, nearly 50,000 active retailers and over 1.4 million products inside the three-tier system that governs beverage alcohol in the United States, as the joint announcement sets out. That route is built for scheduled drops. It is efficient for cases of spirits and wine, and it is close to useless for the small-pack, fast-turn items that run out in the middle of service.
The second route works from grocery shelves instead. Instacart brings more than 2,200 retail banners and nearly 100,000 stores, and the Provi integration is limited to non-alcohol lines: fresh garnish, mixers, juice, cleaning supplies, paper. Nothing in the arrangement replaces the distributor, and nothing in it touches the licensing rules that keep alcohol on a separate order path.
| Route | What it covers | Where it fails | Who owns the record |
|---|---|---|---|
| Distributor drop | Alcohol and full cases, on a schedule | The gap between drops; small packs and perishable garnish | Distributor |
| Marketplace top-up | Non-alcohol, retail pack, same day | The item when it is not on a grocery shelf | Retailer and Instacart |
| Staff run | Anything, immediately | Labour taken off the floor during service | Store manager |
Why the split runs along alcohol
The three-tier system is the reason the new route stops where it does. A restaurant cannot top up vodka from a grocery aisle, so the value of a same-day grocery order sits entirely in the low-value, high-frequency tail: the limes, the mixers, the napkins. That tail is where on-shelf availability is worst, because nobody schedules a drop for a case of citrus.
The promise measures a shelf, not a supplier
Instacart’s own description of its data network explains how a same-day window can be promised at all. The company counts about 600,000 shoppers, more than fifteen visits per store per day, over ten million unique data points daily and 1.6 billion lifetime orders, a base it now resells to retailers as shelf-level visibility (Instacart network figures). The promise is not a claim about the distributor’s truck. It is a reading of what a grocery store has on hand when the order lands.
Substitution is where the promise breaks
A shopper who cannot find fresh limes will offer bottled juice, and the app will record a completed order. For a bar, that substitution changes the drink, the ticket and sometimes the menu. The partnership terms is explicit that substitution rules sit with the platform and the shopper, not with the operator, which means the cheapest part of the promise to measure is also the part most likely to fail quietly.
The item file is the constraint
Both platforms carry an item file, and the two files were never built to match. A garnish that appears in a distributor catalogue under a case code may sit on a grocery shelf under a consumer name, a different pack size and a barcode the operator has never seen. SKU coverage, not delivery speed, decides whether a top-up order can be filled at all.
That mismatch is why availability has to be measured by item rather than by category. A store can look healthy on a category report while the twenty items that drive the basket are out, and a top-up route that is judged on order volume will look successful while the shelf it depends on keeps failing.
What to check in the first ninety days
Three numbers separate a working second route from a press release. Cancellations caused by a missing item, by store. Substitution rate on the hundred items an operator orders most. Repeat rate at sixty days, split between accounts that already used the distributor and accounts that did not. Only the third answers whether the route added supply or moved it.
The useful discipline is to keep the two routes on separate ledgers for a quarter. A blended margin hides which side pays for the other, and a top-up route that quietly absorbs the cost of a distributor gap will look cheaper than it is right up until the volume grows.
FAQ
Does the Provi integration cover alcohol?
No. The embedded checkout is limited to non-alcohol items such as garnish, mixers, juice, cleaning supplies and paper. Alcohol stays on the distributor order path because of the three-tier rules that govern it.
How should an operator judge whether a top-up order helps?
Track fill rate by item rather than by category, then split cancellations caused by a missing item from cancellations caused by the customer. A top-up route that fills ninety percent of lines but misses the twenty items that stop service has not solved the problem.
What breaks first when a second supply route is added?
The item file. Pack size, naming and barcodes differ between a distributor catalogue and a grocery shelf, so on-shelf availability at store level has to be checked item by item before the channel can promise a window; every unmatched item becomes a substitution or a cancellation.
