Complete reference of every metric used in Project NERVE. Includes formula, calculation method, and business meaning.
| Category | Metric | Formula / Threshold | What It Means |
|---|---|---|---|
| PnL — Revenue & Costs | transport_savings_mo | $1.15 x shipped_units | Dollar value of delivery savings. Primary revenue source. |
| PnL — Revenue & Costs | ad_revenue_mo | Actual per-KID from Kiosk Attribution / PRUNE / Financials | Ad campaign revenue. $0 if no campaign running. |
| PnL — Revenue & Costs | lease_revenue_mo | renewal_price / term_months OR original_price / 60 | Fee partner pays Amazon to host locker. Residential only. $17.6M/yr. |
| PnL — Revenue & Costs | concession_cost_mo | $32 x conceded_units | Refund cost per customer concession. |
| PnL — Revenue & Costs | rent_mo | Actual rent from rent file | Rent Amazon pays to store partner. Commercial only. |
| PnL — Revenue & Costs | ad_wrap_cost_mo | $209 if has_actual_ads else $0 | Wrap cost. ONLY charged when campaign is running. |
| PnL — Revenue & Costs | depreciation_mo | $105 Core/Odin, $99 Apartment, $18 Dobby | Monthly hardware depreciation. |
| PnL — Revenue & Costs | maint_fixed_mo | 3-month rolling avg of vendor invoice costs | Actual maintenance spend smoothed over 3 months. |
| PnL — Revenue & Costs | other_fixed_mo | $30 Core/Odin, $15 Apartment, $5 Dobby | Miscellaneous operational costs. |
| PnL — Revenue & Costs | net_savings_mo | Sum of revenue - sum of costs (3-month avg) | Net monthly PnL. Positive = profitable. |
| PnL — Revenue & Costs | estimated_2026_pnl | YTD actuals + net_savings_mo x remaining months | Full-year projection. |
| Performance | shipped_mo | Sum of daily shipped / months (3-month avg) | Monthly package volume processed. |
| Performance | conceded_mo | Sum of daily conceded / months (3-month avg) | Monthly delivery failures. |
| Performance | dpmo | (conceded / shipped) x 1,000,000 (3-month avg) | Defects per million. <8K good, >15K critical. |
| Performance | throughput | packages / (active_days x slot_count) (3-month avg) | Utilization. >0.15 healthy, <0.08 critical. |
| Performance | tput_velocity | OLS slope of throughput over 6 months | Rate of volume change. Negative = eroding demand. |
| Tickets & Maintenance | tickets_ytd | Count of unique IssueIds in current year | Complaints filed. >5 = high risk. |
| Tickets & Maintenance | tickets_6m | Count of unique tickets in last 6 months | Recent complaint activity. |
| Tickets & Maintenance | dispatches_6m | Count of months with vendor dispatches in 6 months | Technician visits. 3+ = alert. |
| Offline | offline_days_ytd | Count of days where offline_days > 0 in 2026 | Total days locker was down YTD. |
| Offline | offline_ytd_pct | offline_days_ytd / total_days_in_period x 100 | % of year locker has been offline. |
| Offline | consecutive_offline_days | Count backward from latest date while offline | Current continuous outage duration. |
| Model — Slope | net_slope | OLS regression slope over 6 months ($/mo per mo) | Rate of financial deterioration. |
| Model — Slope | net_slope_dt | OLS slope of deseasonalized PnL over 6 months | Confirms trend is real, not seasonal. |
| Model — Slope | r_squared | R-squared of 6-month OLS regression (0-1) | Confidence in slope. Higher = more linear decline. |
| Model — Slope | seasonal_idx | avg_month_pnl / overall_avg_pnl (clipped 0.3-3.0) | Per-locker seasonal adjustment factor. |
| Model — Tiers | EMERGING RISK (P0) | Positive today + 4+ months consecutive decline | Still profitable but degrading fast. |
| Model — Tiers | CRITICAL (P1) | Persistent negative + steep slope OR maintenance spike | 95.6% precision. Near-certainty of continued loss. |
| Model — Tiers | HIGH (P2) | Includes 'Run Ads' quality gate candidates | 93.8% precision. Operational fix may resolve. |
| Model — Tiers | ELEVATED (P3) | Moderate decline detected | 91.1% precision. |
| Model — Tiers | LOW (P4) | Early throughput erosion | 87.2% precision. |
| Model — Tiers | STABLE | No decline signal. Flat or improving PnL slope. | Not flagged by the model. |
| Model — Tiers | classification_path | slope | maintenance | run_ads | emerging | N/A | Which mechanism flagged this locker. |
| EWS | ews_vs_net | Composite: how much worse than network avg (0-100) | Relative underperformance vs peers. |
| EWS | ews_vs_self | Composite: how much worse than own baseline (0-100) | Deterioration vs own history. |
| EWS | risk_level | HIGH if both >= 40. MODERATE if one >= 40. | AND logic. Both axes must be bad for HIGH. |
| EWS vs Tier | STABLE + EWS HIGH | Locker not declining (flat) but underperforming | Proactive watchlist. Tier won't catch until slope starts. |
| Cluster (DBSCAN) | cluster_health | Bad: 75%+ neg AND 2x+ worse. Good: DPMO<8K, tput>0.15. | Zone-level classification (500m radius). |
| Cluster (DBSCAN) | multiplier | max(DPMO/avg_DPMO, avg_tput/cluster_tput) | How many times worse than network average. |
| Proximity | distance_m | Haversine straight-line distance (meters) | Physical proximity between two lockers. |
| Proximity | both_neg | Both lockers in pair have net_status = negative | Consolidation candidate if also <500m. |
| DFR | total_dfr | Sum of all failure reason counts (last 90 days) | Pre-delivery failures (different from DPMO/concessions). |
| Alert Thresholds | Maintenance Spike | dispatches_6m >= 3 OR maint_fixed_mo > $50 | Hardware degradation signal. |
| Alert Thresholds | DPMO Spiral | dpmo > 15,000 | Quality critically impaired. |
| Alert Thresholds | Volume Drop | tput_velocity < -0.01 AND throughput < 0.15 | Demand eroding. |
| Alert Thresholds | High Tickets | tickets_ytd > 5 | High complaint volume. Partner dissatisfaction risk. |
| Alert Thresholds | Currently Offline | offline_days > 0 on latest date | Locker not processing packages now. |
$209/mo charged ONLY when has_actual_ads = True (campaign running with revenue > $0). NOT when wrap is merely installed. Violating this inflates losses by ~$15M.
Source file contains ~51% duplicate rows. Pipeline deduplicates by IssueId before counting. Always use unique ticket counts.
Tier = Is PnL slope declining? (trend-based)
EWS = Is locker underperforming vs peers AND vs self? (relative comparison)
A STABLE locker with EWS HIGH = consistently bad but not getting worse. Proactive watchlist signal.
PnL, volume, DPMO, throughput, maintenance all use 3-month rolling averages. Model targets PERSISTENT problems, not one-month dips. Tickets, offline days, dispatches use exact counts (no averaging).
Partner pays Amazon. This is REVENUE, not cost. "Billing annually" means price is annual (divide by 12). Expired + not renewed = $0 rent = free transport savings. Do NOT disturb.
Actual vendor invoices (B&H + Velociti), NOT annual/12 flat estimates. A locker can have $0 for 9 months then $500 spike. 3-month rolling average preserves trends while smoothing one-offs.