November dispute ops — the daily rhythm through peak

The daily 15

Same time every morning:

1. Decline rate vs baseline — the card-testing and payment-issue tripwire. A doubling overnight is a runbook activation, not a curiosity.
2. Flag-queue depth — queue growing faster than review capacity burns it down means calibration drift or an attack; catch it at 20 orders deep, not 200.
3. Arrivals — new alerts actioned within their 24–72h windows, new disputes acknowledged same-day.
4. Refund SLA — cancellations and oversells refunded same-minute, returns processed within 48h of receipt. Refund lag is the quietest dispute generator of the season.

The weekly 30

1. Forecast check: actual orders and flags vs the forecast — big deviations reprice the January dispute wave and its capacity plan.
2. False-positive sample: pull 10 flagged-then-cleared orders; if 9 were obviously fine, the calibration is taxing review capacity you'll want later in the month.
3. Early tail reading: disputes arriving now are mostly October orders — their reason mix previews what BFCM's own tail will look like at 3x. A cluster of unrecognized-charge filings now means a descriptor or receipt problem compounding daily.

Everything else — rule edits, threshold moves, process changes — waits for December unless a runbook says otherwise. Aurai carries the dispute-response load throughout, which is what keeps the daily 15 at 15.