Run-out countdown
Every in-stock selling variant, sorted by projected run-out. Runway = stock ÷ pace (live 7-day pace when SURGE); arrivals that land in time extend it. Order-by = the last safe day to order (run-out minus the 28d lead). Order by that date and the item never runs out; past it, the outage can only be shortened, not prevented.
Pre-orders per week — goal 0
Each bar = units sold while at ≤0 (Amai pre-order lines). This is the proof metric: it dies as availability is fixed.
data table
Restock queue
The scoreboard — cost of past ordering
Lagging metrics. You can't act on these today; they grade the last six weeks of ordering. The countdown above is where the future is decided.
Stock-out trend
Share of tracked variants at ≤0 per snapshot. Two scopes plotted separately — full warehouse vs Locked-Range-only checks. History accrues every refresh.
Full warehouseLocked Range scope
data table
Fill rate by tier
Demand-weighted: in-stock velocity ÷ total velocity of selling variants. Tiers are variant-level 90d units (A≥25 · B 14–24 · C 9–13 · D 1–8).
WFF order scorecard
Read-only from the Active Order Sheet. Cancelled lines re-enter the queue as RETRY until bought or killed (cancel_rule).
Violations — no-4th-state
Anything at ≤0 for 14+ days must be on order, pre-order enabled, or archived.
Retry backlog — factory-cancelled, still dead, on no order
Check-in gaps — arrived on paper, still negative
Units bought on arrived orders while the OOS run predates arrival → never checked in. Physical audit list.
Outside the radar
Non-Locked-Range variants, selling with real pre-sellout velocity, now at ≤0 — team review per full_price_rule, never auto-ordered.
Sold while out of stock — preorders, last 45d
Bleeding styles
Style-level rollup, ranked by estimated forgone revenue/day.