The business problem
You're the regional operations manager. Four product categories — coffee, donuts, breakfast sandwiches, and lunch — each behave differently. Coffee is level with a slight weekly pattern. Donuts have weekend peaks. Sandwiches have a clear morning-vs-noon split. Lunch is highly weather-sensitive. Build a forecast for each, validate it with a tracking signal, and translate it into a staffing recommendation for the next quarter.
What makes this case interesting
The right method depends on the data. Forcing one method on all four categories is the wrong answer. Show the work for why you picked what you picked.
Case kit
Everything you need
- Tim Hortons Weekly Sales Dataset52 weeks × 4 categories, with weather and event columns.
- Forecasting Examples (Excel)Worked examples of every method you might pick.
- FRED Macro DemoUseful for the trend-vs-seasonality framing in your write-up.
- Forecasting & S&OP Industry GuideHow real S&OP cycles use a forecast like this one.
- AI Lab InstructionsThis dataset is the AI Lab's first scenario — prompts included.
Deliverable
A one-slide recommendation to the regional ops director:
- The forecast. Next 4 weeks per category, plus the method you used and why.
- The error. MAD or MAPE on your hold-out, plus a tracking-signal check.
- The action. What staffing or inventory changes you'd make next week as a result.
Topics you'll be applying
- Decision 4 — Forecasting & S&OP · the core technique
- Decision 5 — Inventory & MRP · the forecast feeds the inventory plan