Stop 1 · Phase 0 · Opening balance01 / 10
Nothing works until this is signed
i2 was architecturally sound and produced garbage, because nobody ever loaded the opening stock. Every sale deducted from a starting balance of zero, so sellable drifted to minus 140,753 units and 410 "stockouts" meant nothing at all. The tool to fix it existed the whole time. The physical count never happened. i3 makes the count the gate: no location produces a trustworthy number until a named manager has signed its opening balance.
What to look at
- Three sources, ranked. A fresh physical count at a store, an accepted warehouse count paired with the RFID mass-tagging run, or Shopify levels as a fallback that is explicitly flagged unverified and never presented as truth.
- One signature per location, by name. Not a checkbox. The signer, the timestamp and the source are stored on the row, because in six months somebody will ask who agreed to this number.
- The mass-tagging campaign is the count. 139,561 units at HQ_319, three staff, two weeks, roughly 500 tags an hour. One operation produces both the RFID tags and the signed opening balance. Doing them separately would mean counting twice.
- GATE-0 is the exit. Balances signed per location, the hash chain verifies, and the total sits within 1% of the Shopify total. It passed on 22 July at 0.31%.
- Unverified is visible forever. A location seeded from Shopify carries that flag on every page until a real count replaces it. Nobody gets to forget which numbers were guessed.
Why this stop is first
- Ordering is the lesson. Every other stop in this tour describes machinery that only produces correct answers if this one already happened. A perfect ledger on a wrong opening balance is a perfect record of a wrong number.
- It is the cheapest possible fix. Two weeks of counting, once, versus years of reconciling numbers nobody believes.