NERVE — Network Earnings & Risk Visibility Engine

KID: Kiosk ID — unique identifier for each locker. Links to locker.amazon.com.
Net PnL: Monthly net = transport savings + ad revenue - concessions - rent - ad wrap - maintenance - depreciation - other.
DPMO: Defects per million opportunities (conceded/shipped × 1M). <8K good, >15K critical.
Throughput: Packages per slot per active day. >0.15 healthy, <0.08 critical.
Dispatches: Vendor maintenance visits (from invoices). Different from tickets (complaints filed).
Ad Wrap Cost: $209/mo charged ONLY when an actual ad campaign is running, not when the physical wrap is just installed.

Tier vs EWS — They Measure Different Things:
Tier (CRITICAL/HIGH/EMERGING/STABLE): Based on PnL SLOPE — "Is this locker actively getting worse month-over-month?" A STABLE locker has a flat or improving PnL trend over the last 6 months. It may still be losing money, but it’s not deteriorating further.
Classification Path: HOW the model assigned the tier — via slope decline, maintenance spikes, run-ads opportunity, or emerging pattern. STABLE lockers show "N/A" because they were never classified into a problem tier.
EWS (Early Warning System): Based on RELATIVE PERFORMANCE — "How does this locker compare to network average AND its own history?" A locker can be STABLE (not getting worse) but EWS HIGH (consistently underperforming vs peers).
When Tier = STABLE but EWS = HIGH: The locker is “consistently bad but not deteriorating.” EWS is flagging it as a proactive watchlist item. The tier model won’t alarm until it starts actively declining, but EWS says “this one is already underperforming and deserves attention.”
EWS Risk: HIGH RISK = worse than network AND worse than its own history (both axes). MODERATE = one axis only.