A promotion can look right and still start from the wrong price. This is a working prototype that reads the deal sheet and the system side by side, runs sixteen checks the moment a deal lands, says what is wrong in plain words, and names the team that owns the fix.
Three minutes and forty-six seconds. Then a six-minute walkthrough of every check, and a live copy of the engine you can break yourself.
The submitting team fills one row: product, offer, stores, dates, prices, funding and readiness flags. Eleven fields are mandatory, including the SKU, both prices and the limit per customer. What they submit stays on the sheet; QA writes the verdict back in the last two columns, so the claim and the answer sit side by side.
Every check returns one of four answers. One failure blocks the deal. Warnings with no failure mean a person has to decide. Sixteen passes clears it to build. "Not applicable" is kept separate from "pass", because a check that did not run is not a check that agreed.
PassWarn — a person decidesFail — blockedN/A — did not run
Friday's report
Written from the same rows
How many cleared first time, which checks keep firing, and which team keeps hitting them. In the modeled week: 20 submitted, 7 cleared first pass (35%), 11 blocked before a menu, 2 held for a decision. The dominant failure was vendor-agreement coverage, then vendor caps, then pricing. One root cause: deal sheets written without the agreement terms or the live system price in front of the writer.
Exposure figures on the report are modeled examples, not money anyone banked.
Identity
C1–C3
Row complete
SKU exists
Product, brand, category match the master
Price
C4–C5
Reg price matches the system
Promo price = reg minus the stated discount
Policy
C6–C9
Inside the category cap
Margin clears the floor
Signed vendor agreement covers brand and dates
Vendor discount ceiling
Timing & place
C10–C12
Dates valid, lead time
No other promo live on the SKU
Store codes are real
Readiness
C13–C16
Stock covers the offer
Built on every channel
Copy and image approved
Wording passes the sample policy
Try it live copy of the engine
D-1009, the blocked one. Fix it yourself.
Thirty-five percent off a disposable cart. The deal sheet says the regular price is $39.99; the SKU master says $34.99. Change the numbers and watch the checks re-run — the price checks, the category cap and the margin floor are computed here exactly as the workbook does it; the other twelve are shown as they stand for this deal.
D-1009 Orchard Haze Disposable · 35% Off · NRG-VP-3003 · ALL stores · 08/24–09/21
Blocked
C1
PASS
C2
PASS
C3
PASS
C4
PASS
C5
PASS
C6
PASS
C7
PASS
C8
PASS
C9
PASS
C10
PASS
C11
PASS
C12
PASS
C13
PASS
C14
PASS
C15
PASS
C16
PASS
Correct the prices and the two failures clear — but the row still doesn't go green: 35% sits exactly on the category's 35% cap, so the cap check warns and the deal moves from blocked to needs-review. Correct arithmetic does not remove the decision; it moves it to the right desk.
Stills from the series — the same rules, drawnMore stills from the series — the same rules across four stacks, search scoring, flags and the rule-reason-cost pattern
Disclosure. This is an independent portfolio prototype built to show process. The company in the workbook, Northgate Retail Group, is modeled, and every product, brand, vendor, price and inventory record is fictional. The channel columns name Dutchie, Weedmaps and Leafly only as the places a promotion gets built; no live menu was inspected and nothing here is affiliated with those companies. The workbook does not store an approval history; a production version would.