COUPON
STACK

Seven stores, one engine. Every retailer’s rules are encoded as data, and on top of that data sit the machines: an auto-stack solver that brute-forces every legal combination, a trip planner that prices a whole basket and tells you which store should get the trip, and a money-maker scanner that hunts negative-cost stacks so you don’t have to. Every offer carries confidence and expiry — and the library remembers which layer failed at the register.

v3 status — read before clipping: the full concept plan is built out below — solver, scanner, rankings, list import with a split-trip sorter, in-aisle UPC lookup, alerts, and a trust system with decay, verification, and per-layer failure learning. The offers are still seed data for exercising the engine, not live deals — automatic ingestion stays gated on purpose (lane status at the bottom). Add your own clipped offers with the form in the library; they persist in your browser and feed every engine on this page.

Money-maker scanner

The engine continuously solves every catalogued item at every store that stocks offers for it, floor-gated by the confidence slider in the library. Negative effective cost means the stack pays you. Near-free counts too. This is the alert Matt asked for on day one: surface the crazy stack the moment it exists.

Alerts

Evaluated fresh every time you open this page, against your last visit: money-makers live above the confidence floor, watched offers approaching a known expiry (inferred expiries never alert — per the plan), and stacks that got cheaper since you were last here. Star an offer in the library to watch it.

Best-stack rankings

The league table: every catalogue scenario solved at its best store, floor-gated first, then ranked the plan’s way — absolute savings, then discount %, ties broken by average confidence. Click a row to load it in the stack lab.

Store policies, as data

The same table every engine on this page enforces. Generic shape is sale + store + manufacturer + cashback — the store’s row decides what’s actually allowed. Exclusions override everything; an expired coupon never enters a stack. Sourcing is flagged per store: official policy page beats folklore, and folklore is labeled as folklore.

The library

Search it, sort it, floor-gate it. Mark an offer “failed at register” and its confidence drops on the spot, with the failure recorded against that layer — the trust loop from the plan, running in your browser. Your own clipped offers join the same library and the same engines.

CONFIDENCE FLOOR 60% — offers below the floor are gated out of the solver, scanner, and trip planner. Alerts only fire above it. Show archived (long-dead & thrice-failed records)

Nothing in the library matches that. The register remains undefeated.

Clip your own offer — manual lane, live now

Offer name

Store

Layer

Item keyword

Value

Type

Expires

Confidence %

Source

Stack lab

One item, full control. Build by hand and the checker enforces policy tap by tap — or hit Auto-stack and let the solver brute-force every legal combination, ranked the way the plan demands: floor-gate first, then absolute savings, then discount %, ties broken by confidence.

Store

Item

Shelf price ($)

Item is on sale (percent-off store coupons sit this out at CVS)

Paying with Circle Card (5% payment discount — not a coupon)

Offers for this store — tap to add

Trip planner

The transaction-level pass. Build a basket once — or import a whole list and watch it parse, match, and sort itself. The planner solves every item at a chosen store, applies that store’s transaction layer (Dollar General’s $5-off-$25 Saturday coupon, Target’s Circle 5%), prices the same basket everywhere, and then sorts the list into per-store piles so you can see exactly what splitting the run buys you. Dollar General overage is real and modeled: a coupon worth more than its item spills onto the rest of the basket.

Import a list — one item per line

Load .txt / .csv

Add an item

Shelf price ($)

Trip store

Circle Card if Target (5% after coupons)

Same basket, every store

In-aisle lookup

Standing in the aisle, barcode in hand: type or paste a UPC and the lookup chain runs — local cache first (every barcode is looked up once, then kept), then Open Food Facts live — and whatever it resolves to gets solved at every store, floor-gated, right now. On the phone this is where a camera scan plugs in; the chain behind it is the one from the plan.

UPC / barcode

Sources & ingestion lanes

Where offers come from, graded — and what each lane is allowed to do today. Nothing phones a retailer without a sanctioned pipe.

COUPONSTACK