Problem

• You have an unusual day (concert, outage, trade-fair spike, etc.).
• If you do nothing, PMI Prediction will treat that demand as “normal” next year and the SMART/rooms models will try to repeat it.

Quick fix (what you do today)

Step 1

• Where: Rooms ▸ Live Forecast
• What to do: Find the date. Over-write OTB / Rev / RN with your true expectation
• Why it works: Stamps the day as manual (orange triangle). PMI stops auto-updating it overnight.

Step 2

• Where: Tools ▸ Seasons (same module)
• What to do: Click + Add label. Choose Outliers. Tick the date(s) and Save
• Why it works: Adds the day to the outlier database so the ML model ignores it when learning next year’s pattern.

Step 3

• Where: Sense-check bar (top of Live Forecast)
• What to do: Confirm the red “outlier” icon appears for that day
• Why it works: Visual confirmation the flag is in place.

Step 4

• Where: Save / Send to Forecast
• What to do: Locks the change and pushes it to Cockpits & labour tools
• Why it works: Ensures staffing and cost drivers now follow the corrected forecast.

Tip: If the day is extreme (hotel closed, whole-property buy-out) flag all affected departments, not just Rooms.

What not to do

• You don’t need the Planning (P&L) submission flow just to handle a one-off event.
• Avoid leaving the spike in OTB and “fixing it later” – by then the model has already learned the wrong pattern.

Result:  The outlier tag + manual override stop the live model from copying the anomaly into future seasons, and your labour/cockpit numbers stay realistic.