Evidence boundary: Operational records and outcomes are synthetic. Public context establishes property fit and guest-facing value, not internal policy, costs, or performance.
This report stress-tests a synthetic policy simulation. It does not validate business outcomes or estimate causal recovery effects.
Stability Summary
- Cases evaluated:
430 - Median recommendation stability:
100.0% - High-confidence cases:
285 - Moderate-confidence cases:
133 - Low-confidence cases:
12 - Cases where a tested context removal changed the gesture:
83 - Cases below 60% stability:
2 - Most common gesture:
rooftop_f_and_b_credit(42.6%)
Each recommendation is rescored under ±20% perturbations to fit, cost, occupancy, context, the overall recovery-need scale, and every individual recovery-need weight. Stability is the share of perturbations that preserve the selected gesture.
Cost Uncertainty
| Measure | Synthetic run |
|---|---|
| Low estimated internal-cost bound | $21,864 |
| Midpoint policy estimate | $39,807 |
| High estimated internal-cost bound | $61,817 |
| Cases where midpoint cost exceeds modeled recovery value | 35 |
The range is intentional. Public prices can anchor guest-facing value, but property contribution margin remains unavailable.
Stability By Guest Context
| Guest context | Cases | Average stability | Low-confidence cases |
|---|---|---|---|
| event or suite guest | 10 | 93.7% | 1 |
| loyalty guest | 59 | 96.4% | 3 |
| new guest | 229 | 97.3% | 6 |
| returning guest | 100 | 97.1% | 1 |
| vip guest | 32 | 98.4% | 1 |
Lowest-Stability Cases
| Case | Issue | Recommendation | Stability | Alternatives |
|---|---|---|---|---|
case_00209 | room readiness delay | room upgrade | 55.6% | late checkout, Calabra or Palma dining credit |
case_00220 | room readiness delay | future-stay credit | 59.3% | Calabra or Palma dining credit, late checkout |
case_00286 | room readiness delay | late checkout | 63.0% | room upgrade, Calabra or Palma dining credit |
case_00083 | noise disruption | late checkout | 66.7% | Calabra or Palma dining credit, Palma lounge credit |
case_00353 | noise disruption | Surya Spa or Recovery Suite credit | 66.7% | Calabra or Palma dining credit, Palma lounge credit |
case_00377 | room readiness delay | room upgrade | 66.7% | late checkout, future-stay credit |
case_00397 | noise disruption | Calabra or Palma dining credit | 66.7% | Palma lounge credit, in-room amenity gesture |
case_00170 | room readiness delay | room upgrade | 77.8% | late checkout, Calabra or Palma dining credit |
case_00193 | housekeeping miss | future-stay credit | 77.8% | Calabra or Palma dining credit, Surya Spa or Recovery Suite credit |
case_00314 | room readiness delay | room upgrade | 77.8% | late checkout, Calabra or Palma dining credit |
Interpretation
- High stability means the selected gesture survives the tested policy perturbations; it does not mean the gesture is empirically optimal.
- Low stability should trigger manager review and parameter discussion rather than a stronger automated claim.
- Real comp decisions, overrides, satisfaction recovery, reviews, and repeat-stay outcomes are required for outcome validation.