Understanding what drives revenue in a coffee business
This project analyses a year of transaction-level coffee shop sales data to uncover patterns in customer purchasing behaviour, product performance, and payment preferences. The central question: does it matter when a customer walks in, and how they pay?
The dataset — sourced from Kaggle — contains individual transaction records spanning March 2024 to February 2025, capturing purchase time, coffee type, payment method (cash or card), and sales value. An interactive Power BI dashboard was built to allow stakeholders to slice the data across three time-of-day periods and two payment channels.
Morning brings the crowds. Evening brings the spend.
One of the clearest findings in this dataset is that transaction volume and transaction value move in opposite directions across the day. The morning rush is real — but it isn't where the money is made.
Morning
Highest volume, lowest average transaction value. Customers are habitual and convenience-driven — fast, routine purchases dominate.
Afternoon
Moderate volume with a noticeable uptick in basket size. A mix of leisure and work-break consumption — customers linger and upgrade.
Night
Fewest transactions but the highest average spend per visit. Social and discretionary purchases inflate the per-customer revenue.
Key takeaway
A business optimised only for morning traffic is leaving afternoon and evening revenue on the table.
Card payments quietly dominate revenue
Cash is more common in the morning, where transactions are small and fast. But as the day progresses, card payments take over — and they are consistently linked to higher transaction values.
"Higher-value transactions are more frequently completed using cards, especially in the Afternoon and Night periods — reflecting both convenience and consumer trust in digital payments."
This matters operationally. Businesses that make cash handling the path of least resistance may inadvertently cap their per-transaction ceiling. Incentivising card use — through loyalty points or small discounts — could increase both revenue and checkout speed during the morning rush.
Two products carry the menu. One is left behind.
Latte and Americano with Milk are the clear leaders, each contributing over $20K in total sales. At the other end, Espresso trails significantly — despite typically commanding a premium price elsewhere.
The gap between premium beverages ($35–$36 average) and lower-tier options ($20–$26) suggests clear pricing tiers — and an opportunity to either reposition the bottom performers or bundle them to increase their contribution.
Five actions to improve performance
- Optimise morning operations for speed. The volume is there — remove friction. Bundled offers (e.g. coffee + pastry) can lift average transaction value without slowing down the line.
- Incentivise card payments. A small loyalty reward for card use reduces cash handling costs and correlates with higher spending across afternoon and night segments.
- Run upsell campaigns in the afternoon and at night. These periods have willing spenders. Promote premium beverages, seasonal specials, and add-ons through menu placement and staff training.
- Reposition Espresso. The lowest-selling product is an opportunity — either through targeted promotions, pairing suggestions, or a price-point reassessment to drive trial.
- Align staffing and inventory to demand patterns. Morning needs more staff; afternoon and night need the right products in stock. Data-driven scheduling directly reduces both waste and lost sales.
Data doesn't just describe performance — it reveals opportunity
This analysis demonstrates that time of day and payment method are not passive variables — they are active levers. The morning window is operationally demanding but margin-thin. The afternoon and night windows are quieter but commercially rich, particularly for card-paying customers ordering premium products.
By combining intuitive Power BI visualisations with dynamic filtering, this report enables any stakeholder to identify trends, compare segments, and act — without needing to be a data professional themselves.