Summary & Reflection
Summary of Results
- Ordering Trends
Order volumes decline in the evenings after 4 p.m., reaching their lowest levels before 7 a.m. Spending remains relatively steady in the evenings, with a slight increase toward midnight on Thursdays and throughout Friday evenings. Early morning spending is sporadic but occasionally peaks between 3 a.m. and 6 a.m.
To optimise product placement, products were grouped into low, mid, high, and premium price ranges based on availability and interquartile range (IQR) analysis. The most frequently purchased products fall into the food and beverage categories, with fresh produce leading, followed by dairy, eggs, snacks, and beverages.
- Customer Profiling & Ordering Habits
Analysis of order behaviour and demographics revealed five distinct customer profiles, shaped by loyalty, region, age, income, and family structure.
High-loyalty customers order more relative to their population size and spend the most per item, showing consistent purchasing patterns and a preference for healthier products. Moderate-loyalty customers place the highest total number of orders, following broader market trends. Low-loyalty customers are more price-sensitive, ordering less frequently relative to their population size, and favouring non-perishables and beverages.
Regional, age, and family based spending habits generally align with overall trends, though customers in the Northeast spend slightly less than those in other regions.
Income influences both order volume and spending, with high-income customers placing more orders relative to their population size, and middle and high income customers consistently spend more per item than low income customers.
Recommendations
To optimise engagement and maximise revenue, advertisements should be scheduled during periods of reduced order activity, specifically after 4 p.m. in the evenings and before 7 a.m. in the early morning.
Mid-priced products should be promoted in the evenings, when spending remains moderate. Higher-priced products are better suited for late Thursday and Friday evenings, as well as early mornings, when spending is elevated. Premium products should be targeted between 3 a.m. and 6 a.m., when spending reaches its highest levels.
Food and beverage products should be the primary focus of promotions, with fresh produce, dairy, and eggs leading demand.
Moderate loyalty customers, who place the most orders, should be targeted with a mix of staple and premium products. High loyalty customers, with their stable spending habits and preference for healthier products, are ideal targets for premium and high-quality fresh food promotions. Low loyalty customers, who are more price sensitive, should be targeted with budget friendly non-perishables and beverages.
Income based targeting should be considered, with high income customers engaging more with premium and high priced products, while low-income customers respond better to lower cost essential goods.
Reflection
Successes
The project provided actionable insights into ordering behaviour across customer profiles, enabling a more targeted marketing strategy through better product placement.
Challenges
While data wrangling and cleaning were straightforward, a key challenge in this project was generating visualisations in Python. While Python provides powerful visualisation libraries like Matplotlib and Seaborn, creating clear and intuitive charts required significant customisation. Compared to Tableau, which offers an interactive and visually guided approach, fine-tuning Python visualisations to ensure clarity and interpretability took more effort, from adjusting axis labels to optimising layouts for readability.
Moving Forward
The next step in this project is to create a Tableau storyboard to visually present key findings in an interactive format. While Python was effective for data wrangling and statistical analysis, Tableau’s dynamic visualisation capabilities will allow for clearer storytelling, enabling stakeholders to explore ordering trends, customer profiles, and spending behaviours with filters, dashboards, and drill-down insights. This will enhance the interpretability and impact of the analysis, making it easier to identify actionable insights.
David Griesel 