Forecast Miss in a Compression Month
It is the 20th of the month. Your 180-room hotel forecast 84% occupancy and $168 ADR for the month, but actuals are tracking to 76% and $159 ADR. The gap is concentrated in the last 10 days, where you had assumed a citywide convention would fill the house. The convention's attendance came in 30% below projection, and two corporate accounts cut their midweek travel. Your GM has already submitted the month's forecast to ownership and is asking you to 'explain the miss' in Friday's report. Group pace for next quarter is also soft, down 9% in room nights on the books.
1. How do you diagnose and communicate the variance without either hiding it or overcorrecting?
Read the worked answer
Separate the variance into controllable and uncontrollable components. The convention shortfall is external and non-recurring - quantify it (e.g., the 8-point occupancy gap × 180 rooms × ~10 days ≈ 1,440 room nights, plus ADR dilution). The two account cuts are semi-controllable (relationship/negotiation) and should be flagged separately. Do not blend them into one 'we missed' story; ownership needs to know which part recurs.
Build a bridge: forecast → convention shortfall → account attrition → rate/mix effect → actual. Present it as a waterfall with dollars, not just points. State the ADR gap explicitly: $9 × occupied rooms.
On communication: report the miss plainly with the cause, then immediately present the recovery and prevention actions - e.g., displaced-demand triggers, a convention-contingency rule (never forecast a citywide at 100% of projected attendance; apply a haircut and a floor), and a mid-month re-forecast checkpoint. The judgement point is that credibility comes from accurate attribution and a credible fix, not from spin or from sandbagging next month's forecast to look good later.
2. Group pace is down 9% for next quarter. What forecasting discipline should you apply now?
Read the worked answer
Do not simply lower the number to match pace; decompose pace into segment, day-of-week, and lead-time. A 9% room-night decline might be concentrated in one segment (e.g., association midweek) with transient leisure still strong, in which case the blended forecast change is smaller and the sales response is targeted.
Apply a lead-time-adjusted pickup model: for each future period, use historical pickup curves to project how much of the remaining gap will close before arrival. If you are 60 days out and historically 70% of final business is on the books, the 30% still to come should be projected from the pickup curve, not assumed away. Then layer in known risks (the soft convention calendar) and a scenario range (downside/base/upside) rather than a single point.
Finally, tie the forecast to action: if the gap will not close on its own, the sales team needs a defined volume target and pricing latitude now, while there is still lead time. A forecast that only informs the report and not the selling behavior is a failure of the forecasting process.