What areas are buying
Grocery demand, added up by area, item, and how often people buy it — weekly, monthly, or quarterly — and shown as volumes only. This is the page we take to a store: not who is buying, but how much of what, and where.
Collecting its first lists now The pilot is in its test phase. Households are sending in their weekly, monthly, and quarterly lists — the first aggregate read publishes with the pilot’s Wednesday flyer sweep, once submissions land. The Grocery Pilot is a price matching and planning service. It is not a food program. Project PAW does not supply, deliver, donate, or pay for groceries.
The anonymization promise
Areas and volumes only — never a name, an email, an address, or any single household’s list. A person’s list stays with that person; what reaches this page is the total, and only the total.
Demand by area
Each line is one item in one area, on the rhythm households actually buy it — the weekly shop, the monthly stock-up, or the quarterly bulk buy — with the combined volume for that rhythm.
| Area | Item | Cadence | Volume |
|---|---|---|---|
No aggregate lines yetThe columns are standing and the table is empty because the first lists are still coming in. The first aggregate read publishes with the pilot’s Wednesday flyer sweep once submissions land — and only where at least 5 households in an area contribute to a line. |
|||
How this page gets used at a store
We take this page in and have a short conversation, in plain numbers:
Here is the volume
We need this much of this item, in this area, next week — not a guess, the added-up demand from real household lists.
Here are the two prices
Your shelf price is one number. Your competitor’s price on the same item is another. Both are already public — we just put them side by side.
Fill it at the number that works
Fill the order at a price that works for the households and the store, and the difference is a keepable spread — value created by the volume, shared instead of lost.
This aggregate is the only thing sellers ever see — it is what the blind bid runs on. When sellers bid on a volume line, bids are sealed: no seller sees another seller’s price, and no bid is ever shopped to the next store. Blind is what makes a seller price their true floor.
Behind the page: the triple back end
This is the architecture the app build follows — three back ends, one per side of the pilot. In the test phase the pilot runs on a borrowed set: the intake form, the pilot spreadsheet, and the price book. The three back ends below are the structure being built; the anonymization rule above is the rule the test already runs under.
Household back end
The listsAccounts, the three cadence lists per household, the stores they shop at, approximate area, and household size. A person’s list never leaves this side except as that person’s own plan, emailed back to them.
Seller back end
The bidsSeller accounts, the aggregate volume lines put out to bid, and the sealed bid intake. Blind lives here: a bid is visible to no other seller, ever — sellers see volumes only, never a list, a name, or an email.
Admin back end
The house’s side — engine room and deskThe house’s own side, and the only side that touches both of the others. It runs the aggregation (lists into volumes, with the 5-household gate applied), the price book and Wednesday flyer pricing, and the matching that turns winning bids and flyer prices into household plans — plus the desk work: reviewing submissions, holding back any line under the threshold, publishing each sweep’s aggregate read, tracking who has been sent a plan, and keeping an audit trail on every number. Nothing here is public — this page is the only view sellers and stores get.
And work, when orders move
Aggregated volume is also how local work gets created. When a filled order needs picking, packing, or delivering, those jobs are offered to people in the area — Instacart-style shopping and pickup work, offered through Project PAW’s existing contractor bench and matcher — the same bench that dispatches errands and pickup work — not through the pilot’s own back ends. The test phase takes no orders and fulfils nothing; the jobs layer is designed in now so it is ready when the pilot moves past testing.
Your lists are what fill this table
Send in your weekly, monthly, and quarterly lists with your approximate area and the stores you shop at. Your list stays private — only the combined volumes ever appear here.
This pilot is the grocery front end of the Reverse Marketplace — buyers state what they need, sellers bid to fill it. The grocery pilot adds the three-cadence lists, the area volumes, and the blind bid.
One engine, several faces
This page is one face of the Project PAW matching engine. The same matching that adds up grocery demand here also runs for goods — wants and offers, matched and bid on blind — and for work — jobs matched to the people with the tools to do them. Every face below runs on that one engine.
Demand by area — the public view
Live from the pilot book: the aggregate lines that pass the gate — an area, an item, a cadence, and a household count of five or more. Volumes only; no name, email, address, or single household's list ever appears here. The tabs at the top of the grid switch cadence.
