LET’S ‘TACO-BOUT’ NUMBERS
Turning Sales Data into Actionable Insights
Project Overview
This project focused on optimizing inventory and sales strategy for Tacovore, a local restaurant in Corvallis, Oregon. Using Power BI and Excel, we analyzed historical sales and ingredient usage data to identify seasonal patterns and areas of inefficiency. Our recommendations included improved forecasting models, new menu configurations to increase average order sizes, and targeted initiatives such as Lotería nights to boost traffic during slow periods.
TOOLS & METHODOLOGIES
Power BI: Developed interactive dashboards to visualize sales trends and ingredient usage.
Holt-Winters Forecasting: Applied to predict seasonal avocado demand with a 3-month outlook.
Transaction Analysis: Evaluated order patterns to identify popular menu combinations and pricing opportunities.
Microsoft Excel: Used for data cleaning, aggregation, and exploratory analysis.
Demographic Contextualization: Incorporated local age and income data to support customer-targeted recommendations.
Insight-to-Action Framework: Translated findings into operational strategies tailored to a small business setting.
DATA-DRIVEN RECCOMENDATIONS
Introduce taco plate bundles to encourage higher average order quantities without relying on discounts.
Customers typically ordered 2–3 tacos per transaction (calculated using DAX measures in Power BI to analyze item counts per transaction). Introducing a 4-taco plate could help anchor expectations, simplify ordering decisions, and subtly encourage higher order volumes without relying on discounts.
Host Lotería nights during historically slow periods.
Average ticket sizes were smaller on Tuesdays and during afternoon hours (between lunch and dinner rush, based on Power BI analysis by day and hour). Hosting Lotería nights during these slower periods could increase foot traffic both on event nights and through repeat visits from winners returning to redeem their prizes.
Expand and highlight the kids’ menu to attract more families.
“Niños”, or kids’ menu items, showed a gradual increase in orders over time. This trend aligns with demographic data we pulled for Corvallis and the area immediately surrounding Tacovore. Compared to the city overall, the neighborhood near the restaurant appeared more family-oriented, with slightly older median ages and higher household incomes — suggesting a greater presence of parents with young children.
Improve avocado forecasting using seasonal models.
Avocado usage showed strong seasonality, peaking in summer and dipping in winter. The Holt-Winters model was selected for forecasting due to its ability to capture both trend and seasonal components, making it well-suited for ingredient demand with recurring patterns. Forecasts were further refined to account for social and cultural factors, such as increased usage during holidays and local events like game day. These small changes improved the real-world accuracy of the forecasts.