Back
Case Studies

Ancillary Spend Prediction — Food, Beverage & Merch Forecasting

Mission:

Turn event crowds into smarter revenue streams by predicting how much they’ll spend on food, drinks, and merch — before the doors open.

Tagline:

🎯 Predicting what fans will eat, drink, and buy — before they show up.
🍔🎤 From popcorn to band tees — optimized by machine learning.

The Challenge:

At live events, ticket sales aren’t the only money-maker — concessions and merch can be a huge slice of revenue. But spend varies wildly by:
  • Artist & event type
  • Audience demographics
  • Time, day, & seasonality
  • Venue & moment (pre-show rush vs mid-show orders)
The goal? and recommend which products will perform best.

How the Khares Analytics team helped:

  • ML models predicting F&B spend per attendee + merch spend per attendee
  • Audience-aware forecasting using:
  • Event metadata (artist, genre, sub-genre)
  • Demographics (age, gender, income profiles)
  • Timing trends (month, weekday, hour, show period)
  • Split training/scoring pipeline:

• Trained on historical events
• Scored future events with zero leakage
• Feature engineering across product categories & time behavior

Key Insights:

  • Spend patterns differ dramatically between genres
  • Higher-income crowds ≠ automatic higher F&B spend — product mix matters
  • Timing windows drive spend spikes (pre-show > intermission > end)
  • Not all events benefit from the same inventory mix — customization wins

The Outcome:

A data-driven recommendation system that helps venues:

✅ Predict concession & merch revenue before events
✅ Plan inventory & staffing with accuracy
✅ Tailor product mix to event-type & audience behavior
✅ Reduce waste + increase per-head yield

Translation:

More profitable bars, smarter merch tables, happier fans, and zero “oops we ran out of nachos again” moments.
Next Project
On-Sale Price Optimization — Live Event Demand AI