Too much prep wastes money, too little disappoints guests
Every restaurant faces the same dilemma: prepare too much and throw food away, or prepare too little and tell customers "we are out." Demand forecasting is the science of finding the sweet spot. A restaurant that forecasts well wastes 2-3% of purchases. One that does not wastes 7-10%. The annual difference runs into thousands of euros.
Six factors that drive demand
- Day of the week — Saturday evening 80 covers, Tuesday 35. Historical data by day is the most reliable predictor.
- Weather — Rain means 10-20% fewer covers (especially with outdoor seating). Summer sun means 15-20% more.
- Local events — Football matches, concerts and fairs shift cover counts depending on your location.
- Seasonality — December full, January empty. August at the coast packed, August in the city quiet.
- Reservations — The most reliable data point. 40 reserved covers usually means 45-55 total (with walk-ins).
- Daily specials — If the daily special is fish, the standard fish main sells less.
A practical forecasting method
- Calculate average covers for each day of the week over the last 4 weeks.
- Adjust for known factors: confirmed reservations, weather forecast, local events.
- From predicted covers, estimate menu mix (40% first course, 30% main, etc.) based on history.
- Calculate ingredient quantities needed and compare with current stock.
- Order the difference.
BiteBase automates this with AI that analyzes sales history, reservations and seasonality to suggest quantities to prepare and order.
Common mistakes
- Preparing for the best case — do not prep for 80 covers when Tuesday averages 40.
- Ignoring seasonality — January is not December.
- Ignoring reservations — they are the most reliable predictor.
- Not tracking stockouts — every "we are out" is data telling you that dish needed more quantity.
FAQ
How accurate is AI forecasting? With 3+ months of history, 85-90% accuracy on cover counts and 75-80% on menu mix.
How do I handle unpredictable events? Keep a 10-15% safety stock on non-perishables. For perishables, order just in time.